Request to postpone some Python 3.9 incompatible changes to Python 3.10
Hi, Python 3.9 introduces many small incompatible changes which broke tons of Python projects, including popular projects, some of them being unmaintained but still widely used (like nose, last release in 2015). Miro and me consider that Python 3.9 is pushing too much pressure on projects maintainers to either abandon Python 2.7 right now (need to update the CI, the documentation, warn users, etc.), or to introduce a *new* compatibility layer to support Python 3.9: layer which would be dropped as soon as Python 2.7 support will be dropped (soon-ish). Python 3.9 is too early to accumulate so many incompatible changes on purpose, some incompatible changes like the removal of collections aliases to ABC should be reverted, and reapplied on Python 3.10. Python 3.9 would be the last release which still contain compatibility layers for Python 2.7. Said differently, we request to maintain the small compatibility layer in Python for one more cycle, instead of requesting every single project maintainer to maintain it on their side. We consider that the general maintenance burden is lower if it's kept in Python for now. == Fedora COPR notify packages broken by Python 3.9 == In Python 3.9, Victor introduced tons of incompatible changes at the beginning of the devcycle. His plan was to push as many as possible, and later decide what to do... This time has come :-) He wrote PEP 606 "Python Compatibility Version" and we wrote PEP 608 "Coordinated Python release", but both have been rejected. At least, it seems like most people agreed that having a CI to get notified of broken projects should help. We are updating the future Fedora 33 to replace Python 3.8 with Python 3.9. We are using a tool called "COPR" which is like a sandbox and can be seen as the CI discussed previously. It rebuilds Fedora using Python 3.9 as /usr/bin/python3 (and /usr/bin/python !). We now have a good overview of broken packages and which incompatible changes broke most packages. https://fedoraproject.org/wiki/Changes/Python3.9 - Describes the Fedora change. https://copr.fedorainfracloud.org/coprs/g/python/python3.9/monitor/ - Has package failures. Some packages fail because of broken dependencies. https://bugzilla.redhat.com/showdependencytree.cgi?id=PYTHON39 - Has open Python 3.9 bug reports for Fedora packages. Some problems have been fixed upstream already before reaching Fedora, most are only fixed when the Fedora maintainers report the problems back to upstream maintainers. Right now, there are 150+ packages broken by Python 3.9 incompatible changes. == Maintenance burden == Many Python projects have not yet removed Python 2 support and Python 2 CI. It's not that they would be in the "we will not drop Python 2 support ever" camp, it's just that they have not done it yet. Removing Python 2 compatibility code from the codebase and removing it from the documentation and metadata and CI is a boring task, doesn't bring anything to users, and it might take a new major release of the library. At this point, we are very early in 2020 to expect most projects to have already done this. At the same time, we would like them to support Python 3.9 as soon in the release cycle as possible. By removing Python 2 compatibility layers from the 3.9 standard library, we are forcing the projects maintainers to re-invent their own compatibility layers and copy-paste stuff like this around. Example: try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence While if we remove collections.Sequence in 3.10, they will face this decision early in 2021 and they will simply fix their code by adding ".abc" at the proper places, not needing any more compatibility layers. Of course, there will be projects that will still have declared Python 2 support in 2021, but it will arguably not be that many. While it's certainly tempting to have "more pure" code in the standard library, maintaining the compatibility shims for one more release isn't really that big of a maintenance burden, especially when comparing with dozens (hundreds?) of third party libraries essentially maintaining their own. An good example of a broken package is the nose project which is no longer maintained (according to their website): the last release was in 2015. It remains a very popular test runner. According to libraries.io, it has with 3 million downloads per month, 41.7K dependent repositories and 325 dependent packages. We patched nose in Fedora to fix Python 3.5, 3.6, 3.8 and now 3.9 compatibility issues. People installing nose from PyPI with "pip install" get the unmaintained flavor which is currently completely broken on Python 3.9. Someone should take over the nose project and maintain it again, or every single project using nose should pick another tool (unittest, nose2, pytest, whatever else). Both options will take a lot of time. == Request to revert some incompatible changes == See https://docs.python.org/dev/whatsnew/3.9.html#removed Incompatible changes which require "if <python3>: (...) else: (...)" or "try: <python3 code> except (...): <python2 code>": * Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API == Kept incompatible changes == Ok-ish incompatible changes (mostly only affects compatiblity libraries like six): * _dummy_thread and dummy_threading modules removed: broke six, nine and future projects. six and nine are already fixed for 3.9. OK incompatible changes (can be replaced by the same code on 2.7 and 3.9): * isAlive() method of threading.Thread has been removed: Thread.is_alive() method is available in 2.7 and 3.9 * xml.etree.ElementTree.getchildren() and xml.etree.ElementTree.getiterator() methods are removed from 3.9, but list()/iter() works in 2.7 and 3.9 == Call to check for DeprecationWarning in your own projects == You must pay attention to DeprecationWarning in Python 3.9: it will be the last "compatibility layer" release, incompatible changes will be reapplied to Python 3.10. For example, you can use the development mode to see DeprecationWarning and ResourceWarning: use the "-X dev" command line option or set the PYTHONDEVMODE=1 environment variable. Or you can use the PYTHONWARNINGS=default environment variable to see DeprecationWarning. You might even want to treat all warnings as errors to ensure that you don't miss any when you run your test suite in your CI. You can use PYTHONWARNINGS=error, and combine it with PYTHONDEVMODE=1. Warnings filters can be used to ignore warnings in third party code, see the documentation: https://docs.python.org/dev/library/warnings.html#the-warnings-filter -- Victor Stinner and Miro Hrončok for Fedora
Two pieces of feedback on this. One, nose is a bad example because that project has been telling people for years to switch to nose2 so the fact that people have still not switched something that should mostly be a direct swap after years of being asked to does not motivate in wanting to postpone one more year for that project's benefit. ;) (And where did you get the download stats on libraries.io? https://libraries.io/pypi/nose doesn't show any download counts.) Two, you buried the list of things you would like to revert and postpone to Python 3.10. :) Was this copied-and-pasted from somewhere, hence why you explain to us how to use `-X dev`? Basically it isn't easy to discuss the things you want to revert here, especially since you grouped them all together. Are you planning to start separate conversations on the (I think) 5 things you want to change?
Le jeu. 23 janv. 2020 à 20:45, Brett Cannon <brett@python.org> a écrit :
Two pieces of feedback on this. One, nose is a bad example because that project has been telling people for years to switch to nose2 so the fact that people have still not switched something that should mostly be a direct swap after years of being asked to does not motivate in wanting to postpone one more year for that project's benefit. ;) (And where did you get the download stats on libraries.io? https://libraries.io/pypi/nose doesn't show any download counts.)
Oh right, download stats can be found at: https://pypistats.org/packages/nose Downloads last day: 150,272 Downloads last week: 824,557 Downloads last month: 3,077,860 150k downloads yesterday.
Two, you buried the list of things you would like to revert and postpone to Python 3.10. :)
I'm not sure of the meaning of "buried" here. What do you mean? We propose to revert 5 changes: * Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
Was this copied-and-pasted from somewhere, hence why you explain to us how to use `-X dev`?
Well, use whatever you want to see DeprecationWarning, as soon as you fix all of them in your project :-) PYTHONWARNINGS=default should be fine as well. IMO it's simpler to remind "-X dev" than using PYTHONWARNINGS=default. By the way, -X dev and PYTHONWARNINGS=default shows more warnings than just DeprecationWarning: they also show ImportWarning, PendingDeprecationWarning and ResourceWarning. -X dev helps to make your project ready for the next Python version.
Basically it isn't easy to discuss the things you want to revert here, especially since you grouped them all together. Are you planning to start separate conversations on the (I think) 5 things you want to change?
I'm not sure of what you mean. Would you prefer one mail thread per revert? It's not only about specific changes, but more a discussion about a general policy to decide if a deprecated feature should stay until 3.10, or if it's ok to remove it in 3.9. For example, another feature has been removed since we drafted this email: "The encoding parameter of json.loads() has been removed." If we follow the proposed policy, it should also be reverted. It seems like the incompatible change causing most issues is the collections change. We can try to produce statistics per change if you prefer. I understand that you disagree to revert changes, but you would prefer to decide on a case by case basis. Victor -- Night gathers, and now my watch begins. It shall not end until my death.
On Jan 23, 2020, at 14:03, Victor Stinner <vstinner@python.org> wrote:
I'm not sure of the meaning of "buried" here. What do you mean? We propose to revert 5 changes:
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
Basically it isn't easy to discuss the things you want to revert here, especially since you grouped them all together. Are you planning to start separate conversations on the (I think) 5 things you want to change?
It's not only about specific changes, but more a discussion about a general policy to decide if a deprecated feature should stay until 3.10, or if it's ok to remove it in 3.9.
Given that we’ve changed the release cadence to one major release per year, it doesn’t seem that much of a burden to revert and carry these changes forward into Python 3.10. And if it helps with the migration off of Python 2.7, then +1 from me. -Barry
On 01/23/2020 03:36 PM, Barry Warsaw wrote:
On Jan 23, 2020, at 14:03, Victor Stinner wrote:
It's not only about specific changes, but more a discussion about a general policy to decide if a deprecated feature should stay until 3.10, or if it's ok to remove it in 3.9.
Given that we’ve changed the release cadence to one major release per year, it doesn’t seem that much of a burden to revert and carry these changes forward into Python 3.10. And if it helps with the migration off of Python 2.7, then +1 from me.
I suspect anyone who's waited this long to switch to Python 3 will be okay with staying on 3.8 for a few years. -- ~Ethan~
On Fri, 24 Jan 2020 at 14:46, Ethan Furman <ethan@stoneleaf.us> wrote:
On 01/23/2020 03:36 PM, Barry Warsaw wrote:
On Jan 23, 2020, at 14:03, Victor Stinner wrote:
It's not only about specific changes, but more a discussion about a general policy to decide if a deprecated feature should stay until 3.10, or if it's ok to remove it in 3.9.
Given that we’ve changed the release cadence to one major release per year, it doesn’t seem that much of a burden to revert and carry these changes forward into Python 3.10. And if it helps with the migration off of Python 2.7, then +1 from me.
I suspect anyone who's waited this long to switch to Python 3 will be okay with staying on 3.8 for a few years.
The issue isn't end users, it is projects who will then be incompatible with 3.10 for "a few years." -Rob
On 01/23/2020 07:02 PM, Robert Collins wrote:
On Fri, 24 Jan 2020 at 14:46, Ethan Furman <ethan@stoneleaf.us <mailto:ethan@stoneleaf.us>> wrote:
On 01/23/2020 03:36 PM, Barry Warsaw wrote: > On Jan 23, 2020, at 14:03, Victor Stinner wrote:
>> It's not only about specific changes, but more a discussion about a >> general policy to decide if a deprecated feature should stay until >> 3.10, or if it's ok to remove it in 3.9. > > Given that we’ve changed the release cadence to one major release per year, it doesn’t seem that much of a burden to revert and carry these changes forward into Python 3.10. And if it helps with the migration off of Python 2.7, then +1 from me.
I suspect anyone who's waited this long to switch to Python 3 will be okay with staying on 3.8 for a few years.
The issue isn't end users, it is projects who will then be incompatible with 3.10 for "a few years."
I'm sure you meant 3.9, since 3.10 would still have the removals. My understanding is that this postponement of removals is aimed at those who have just migrated to Python 3, not those who have already done it nor those who have 2/3 straddling code bases. For those who have been on 3 for a while, updating to use the newer APIs for 3.9 vs 3.10 shouldn't make a difference. Like-wise for those with 2/3 straddling code bases (we'll just need to add a few more things to our shims). Anyone who hasn't supported/used Python 3 until now shouldn't have a problem with sticking with 3.8 until they are ready to make more adjustments, and those who have had to keep Python 2 around to run those applications/frameworks/whatevers will, I should think, be thrilled to use any Python 3 instead. :-) -- ~Ethan~
On 24. 01. 20 5:59, Ethan Furman wrote:
My understanding is that this postponement of removals is aimed at those who have just migrated to Python 3, not those who have already done it nor those who have 2/3 straddling code bases.
No, the motivation to pospone the changes to 3.10 are projects that alrady support both 2 and 3 at the same time, with or without compatibility libraries like six. Before they had anough time to make the necessary actions to abandon Python 2.7, we ask them to support 3.9 from the same code base. Hence, they need to replace simple code: from collections import Sequence With either: try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence Or: from compatlib.moves.collections_abc import Sequence In both cases, we move the burden of having a compatibility layer of some sort from one central location (Python 3.9 stdlib) to dozens (hundreds?) locations.
For those who have been on 3 for a while, updating to use the newer APIs for 3.9 vs 3.10 shouldn't make a difference.
For Python 3 only projects? No, no difference.
Like-wise for those with 2/3 straddling code bases (we'll just need to add a few more things to our shims).
And our point is that because it's too early in 2020 to drop Python 2 code, it is better to encourage projects to update the code as Python 3 only in a year than to let them add more shims now just to hopefully remove them in couple months.
Anyone who hasn't supported/used Python 3 until now shouldn't have a problem with sticking with 3.8 until they are ready to make more adjustments, and those who have had to keep Python 2 around to run those applications/frameworks/whatevers will, I should think, be thrilled to use any Python 3 instead. :-)
This does not affect projects who don't support Python 3 yet. -- Miro Hrončok -- Phone: +420777974800 IRC: mhroncok
Le ven. 24 janv. 2020 à 08:37, Miro Hrončok <mhroncok@redhat.com> a écrit :
No, the motivation to pospone the changes to 3.10 are projects that alrady support both 2 and 3 at the same time, with or without compatibility libraries like six. Before they had anough time to make the necessary actions to abandon Python 2.7, we ask them to support 3.9 from the same code base.
Let me give you an example. The python-ipmi project works on Python 2.7-3.8, but it is not longer compatible with Python 3.9 because tostring/fromstring methods have been removed from array.array. IMO this change is a good example to illustrate the issue: https://github.com/kontron/python-ipmi/pull/73/files I chose to add py3_array_tobytes() and py3_array_frombytes() functions rather than adding "if _PY3: ... else: ...". Example: - self.minor = int(data[1:2].tostring().decode('bcd+')) + self.minor = int(py3_array_tobytes(data[1:2]).decode('bcd+')) This change reminds me when I added "Python 3" support to dozens of Python projects. You can see in this PR that Python 3.9 currently requires to modify 9 files in a small project. It is likely to be worse in a larger project. With Python 3.9, we cannot simply group all Python 3.x versions under "Python 3" anymore. Currently, a code compatible with "Python 3" means more likely compatible with "Python 3.5-3.8" ("Python 3.3-3.8" for the most conservative projects?). We kept a compatibility layer with Python 2 on purpose, PEP 4 says: "In order to facilitate writing code that works in both Python 2 & 3 simultaneously, any module that exists in both Python 3.5 and Python 2.7 will not be removed from the standard library until Python 2.7 is no longer supported as specified by PEP 373." The rule was used since Python 3.0 until Python 3.8, but it changed in Python 3.9 which includes many incompatible changes for the first time in the Python 3 major version. Victor
from collections import Sequence
With either:
try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence
Or:
from compatlib.moves.collections_abc import Sequence
In both cases, we move the burden of having a compatibility layer of some sort from one central location (Python 3.9 stdlib) to dozens (hundreds?) locations.
For those who have been on 3 for a while, updating to use the newer APIs for 3.9 vs 3.10 shouldn't make a difference.
For Python 3 only projects? No, no difference.
Like-wise for those with 2/3 straddling code bases (we'll just need to add a few more things to our shims).
And our point is that because it's too early in 2020 to drop Python 2 code, it is better to encourage projects to update the code as Python 3 only in a year than to let them add more shims now just to hopefully remove them in couple months.
Anyone who hasn't supported/used Python 3 until now shouldn't have a problem with sticking with 3.8 until they are ready to make more adjustments, and those who have had to keep Python 2 around to run those applications/frameworks/whatevers will, I should think, be thrilled to use any Python 3 instead. :-)
This does not affect projects who don't support Python 3 yet.
-- Miro Hrončok -- Phone: +420777974800 IRC: mhroncok _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/TFCAMIAJ... Code of Conduct: http://python.org/psf/codeofconduct/
-- Night gathers, and now my watch begins. It shall not end until my death.
On Fri, Jan 24, 2020 at 7:05 PM Victor Stinner <vstinner@python.org> wrote:
We kept a compatibility layer with Python 2 on purpose, PEP 4 says:
"In order to facilitate writing code that works in both Python 2 & 3 simultaneously, any module that exists in both Python 3.5 and Python 2.7 will not be removed from the standard library until Python 2.7 is no longer supported as specified by PEP 373."
The rule was used since Python 3.0 until Python 3.8, but it changed in Python 3.9 which includes many incompatible changes for the first time in the Python 3 major version.
I'm sorry, I don't understand what 'changed'. Isn't that rule exactly WHY 3.9 is the removal point? Python 2.7 is no longer supported, and its final post-support release is scheduled earlier than 3.9's first beta and feature freeze. Doesn't that mean that PEP 4 is being followed precisely? What have I misunderstood? ChrisA
The change is that Python 2.7 is no longer supported (since 2020-01-01). Victor Le ven. 24 janv. 2020 à 09:19, Chris Angelico <rosuav@gmail.com> a écrit :
On Fri, Jan 24, 2020 at 7:05 PM Victor Stinner <vstinner@python.org> wrote:
We kept a compatibility layer with Python 2 on purpose, PEP 4 says:
"In order to facilitate writing code that works in both Python 2 & 3 simultaneously, any module that exists in both Python 3.5 and Python 2.7 will not be removed from the standard library until Python 2.7 is no longer supported as specified by PEP 373."
The rule was used since Python 3.0 until Python 3.8, but it changed in Python 3.9 which includes many incompatible changes for the first time in the Python 3 major version.
I'm sorry, I don't understand what 'changed'. Isn't that rule exactly WHY 3.9 is the removal point? Python 2.7 is no longer supported, and its final post-support release is scheduled earlier than 3.9's first beta and feature freeze. Doesn't that mean that PEP 4 is being followed precisely? What have I misunderstood?
ChrisA _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/GCCLU2O2... Code of Conduct: http://python.org/psf/codeofconduct/
-- Night gathers, and now my watch begins. It shall not end until my death.
On Fri, Jan 24, 2020 at 7:40 PM Victor Stinner <vstinner@python.org> wrote:
The change is that Python 2.7 is no longer supported (since 2020-01-01).
Which means that it's now okay to remove things, correct? Starting with 3.9, it's no longer necessary to maintain compatibility with 2.7? ChrisA
Adding Python 3.9 and dropping Python 2.7 support are two different things, but Python 3.9 somehow "enforces" to do both at the same time. The proposal is to give one year to project maintainers to drop Python 2.7 support, since Python 2.7 end of support just happened a few weeks ago (2020-01-01). Big projects have more resources to schedule dropping Python 2 support, coordinated with other projects: https://python3statement.org/ But for smaller projects with less resources, it's more complicated. Many very popular Python modules are maintained by a single person with limited resources. Even if we propose pull requests, it still takes time to review it, prepare a new release, etc. Victor Le ven. 24 janv. 2020 à 09:42, Chris Angelico <rosuav@gmail.com> a écrit :
On Fri, Jan 24, 2020 at 7:40 PM Victor Stinner <vstinner@python.org> wrote:
The change is that Python 2.7 is no longer supported (since 2020-01-01).
Which means that it's now okay to remove things, correct? Starting with 3.9, it's no longer necessary to maintain compatibility with 2.7?
ChrisA
-- Night gathers, and now my watch begins. It shall not end until my death.
On Fri, 24 Jan 2020 at 10:05, Victor Stinner <vstinner@python.org> wrote:
The proposal is to give one year to project maintainers to drop Python 2.7 support, since Python 2.7 end of support just happened a few weeks ago (2020-01-01).
IMO creating this kind of "gray areas" in support and deprecation issues is bad. What this will create is just more sources for arguing/debates. Once deprecation or EoL schedule is set, it is best to align to it. Discussions about the schedules should happen when setting them, not when the deadline is coming. Also I am not sure it is really worth it. For example, importing ABCs directly from collections was deprecated 8 years ago, what would 1 extra year change? -- Ivan
On 1/24/2020 5:50 AM, Ivan Levkivskyi wrote:
On Fri, 24 Jan 2020 at 10:05, Victor Stinner <vstinner@python.org <mailto:vstinner@python.org>> wrote:
The proposal is to give one year to project maintainers to drop Python 2.7 support, since Python 2.7 end of support just happened a few weeks ago (2020-01-01).
IMO creating this kind of "gray areas" in support and deprecation issues is bad. What this will create is just more sources for arguing/debates. Once deprecation or EoL schedule is set, it is best to align to it. Discussions about the schedules should happen when setting them, not when the deadline is coming.
Also I am not sure it is really worth it. For example, importing ABCs directly from collections was deprecated 8 years ago, what would 1 extra year change?
I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say. Eric
On 24. 01. 20 14:02, Eric V. Smith wrote:
On 1/24/2020 5:50 AM, Ivan Levkivskyi wrote:
On Fri, 24 Jan 2020 at 10:05, Victor Stinner <vstinner@python.org <mailto:vstinner@python.org>> wrote:
The proposal is to give one year to project maintainers to drop Python 2.7 support, since Python 2.7 end of support just happened a few weeks ago (2020-01-01).
IMO creating this kind of "gray areas" in support and deprecation issues is bad. What this will create is just more sources for arguing/debates. Once deprecation or EoL schedule is set, it is best to align to it. Discussions about the schedules should happen when setting them, not when the deadline is coming.
Also I am not sure it is really worth it. For example, importing ABCs directly from collections was deprecated 8 years ago, what would 1 extra year change?
I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say.
The concern is not that they don't want to drop 2.7 support, but that is is a nontrivail task to actaually do and we cannot expect them to do it within the first couple weeks of 2020. While at the same time, we want them to support 3.9 since the early development versisons in order to eb able to detect regressions early in the dev cycle. -- Miro Hrončok -- Phone: +420777974800 IRC: mhroncok
On 1/24/2020 9:14 AM, Miro Hrončok wrote:
On 24. 01. 20 14:02, Eric V. Smith wrote:
I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say.
The concern is not that they don't want to drop 2.7 support, but that is is a nontrivail task to actaually do and we cannot expect them to do it within the first couple weeks of 2020. While at the same time, we want them to support 3.9 since the early development versisons in order to eb able to detect regressions early in the dev cycle.
Ah. So in 3.8, they kept code that had deprecation warnings so that they could be compatible with 2.7. They'd like to now drop that code and be 3.9-only compatible, but they don't have enough time to do that because they couldn't start that work as long as they were supporting 2.7. Do I have that right? If so, I'd be okay with postponing the removal of the deprecated code until 3.10. But I don't think we should postpone it if the driver is so that libraries can remain 2.7 compatible. That could go on forever. This postponement would be a one-time thing. Eric
Eric V. Smith wrote:
On 24. 01. 20 14:02, Eric V. Smith wrote: I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say. The concern is not that they don't want to drop 2.7 support, but that is is a nontrivai task to actually do and we cannot expect them to do it within the first couple weeks of 2020. While at the same time, we want them to support 3.9 since the early development versisons in order to eb able to detect regressions early in the dev cycle. Ah. So in 3.8, they kept code that had deprecation warnings so that they could be compatible with 2.7. They'd like to now drop that code and be 3.9-only compatible, but they don't have enough time to do that because
On 1/24/2020 9:14 AM, Miro Hrončok wrote: they couldn't start that work as long as they were supporting 2.7. Do I have that right? If so, I'd be okay with postponing the removal of the deprecated code until 3.10. But I don't think we should postpone it if the driver is so that libraries can remain 2.7 compatible. That could go on forever. This postponement would be a one-time thing.
I'm also okay with a one-time delay in removals that are problematic for code trying to get off of Python 2.7 this year and might not quite cut it before 2021 hits. I'm sure some people will be caught off-guard once 3.9b1 comes out and they realize that they now have to start caring about deprecation warnings again. So I'm okay letting 3.9 be the release where we loudly declare, "deprecation warnings matter again! Keep your code warnings-free going forward as things will start being removed in 3.10".
Le ven. 24 janv. 2020 à 19:35, Brett Cannon <brett@python.org> a écrit :
I'm also okay with a one-time delay in removals that are problematic for code trying to get off of Python 2.7 this year and might not quite cut it before 2021 hits. I'm sure some people will be caught off-guard once 3.9b1 comes out and they realize that they now have to start caring about deprecation warnings again. So I'm okay letting 3.9 be the release where we loudly declare, "deprecation warnings matter again! Keep your code warnings-free going forward as things will start being removed in 3.10".
Maybe we can put a strong statement at the What's New In Python 3.9 document? Victor -- Night gathers, and now my watch begins. It shall not end until my death.
On 24. 01. 20 16:22, Eric V. Smith wrote:
On 1/24/2020 9:14 AM, Miro Hrončok wrote:
On 24. 01. 20 14:02, Eric V. Smith wrote:
I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say.
The concern is not that they don't want to drop 2.7 support, but that is is a nontrivail task to actaually do and we cannot expect them to do it within the first couple weeks of 2020. While at the same time, we want them to support 3.9 since the early development versisons in order to eb able to detect regressions early in the dev cycle.
Ah. So in 3.8, they kept code that had deprecation warnings so that they could be compatible with 2.7. They'd like to now drop that code and be 3.9-only compatible, but they don't have enough time to do that because they couldn't start that work as long as they were supporting 2.7. Do I have that right?
Yes.
If so, I'd be okay with postponing the removal of the deprecated code until 3.10. But I don't think we should postpone it if the driver is so that libraries can remain 2.7 compatible. That could go on forever. This postponement would be a one-time thing.
-- Miro Hrončok -- Phone: +420777974800 IRC: mhroncok
On Fri, 24 Jan 2020 at 14:14, Miro Hrončok <mhroncok@redhat.com> wrote:
On 24. 01. 20 14:02, Eric V. Smith wrote:
I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say.
The concern is not that they don't want to drop 2.7 support, but that is is a nontrivail task to actaually do and we cannot expect them to do it within the first couple weeks of 2020. While at the same time, we want them to support 3.9 since the early development versisons in order to eb able to detect regressions early in the dev cycle.
But couldn't they have done that by adding the compatibility shims Victor described. Sure, they are messy, and sure, it would be more convenient not to have to. But the various dates have been known for some time now. Unless you're saying that the change to a yearly schedule for 3.9 has suddenly compressed timescales to an extent that projects can't cope with. So maybe there is a reason why people might legitimately have an issue here. Having said that, I think that far more people will see this as yet another delay before 2.7 dies, and treat it as one more reason to do nothing. So I still have reservations. I have already said that the 5 deprecations Victor described seem minor enough that one release's delay is not a big deal. If that's enough for the people hit by the shorter 3.9 cycle, then great. But it should *not* set a precedent that we're willing to repeatedly have this sort of discussion. It should be a one-off, and definitely not set the scene for a series of "and can this one be added as well" requests. I assume it's obvious that 3.10 would be a hard deadline, as well. There's no justification for deferring deprecations past then. Paul PS As a pip developer, I fully expect to be supporting Python 2.7 for a while yet. So I'm not talking from a perspective of someone who has happily dropped Python 2 support and doesn't understand the issues. But I don't expect *Python* to maintain my compatibility for me, and I don't see why others should, either. And yes, that does include these 5.
On 01/24/2020 05:02 AM, Eric V. Smith wrote:
On 1/24/2020 5:50 AM, Ivan Levkivskyi wrote:
On Fri, 24 Jan 2020 at 10:05, Victor Stinner wrote:
The proposal is to give one year to project maintainers to drop Python 2.7 support, since Python 2.7 end of support just happened a few weeks ago (2020-01-01).
Also I am not sure it is really worth it. For example, importing ABCs directly from collections was deprecated 8 years ago, what would 1 extra year change?
I think the concern is that with removing so many deprecated features, we're effectively telling libraries that if they want to support 3.9, they'll have stop supporting 2.7. And many library authors aren't willing to do that yet. Will they be willing to in another year? I can't say.
I'm not dropping support for 2.7 for the foreseeable future for my libraries, so an extra year won't help me at all (other than give me another 12 months to procrastinate). -- ~Ethan~
On Fri, 24 Jan 2020 at 08:54, Victor Stinner <vstinner@python.org> wrote:
The change is that Python 2.7 is no longer supported (since 2020-01-01).
However the assertion here seems to be that some people are unprepared for this (which seems to me like it's their problem, not ours). Features getting deprecated in newer Python versions is routine. Anyone maintaining Python 2.7 compatibility should already be painfully aware that we're deprecating things in 3.x and they have to maintain compatibity code for that. We've been minimising the impact for such people as much as we can for years. When *should* we stop doing so, if not at the point where we finally declared 2.7 unsupported? Having said that, the 5 specific changes noted seem relatively minor and deferring them an extra release doesn't seem like a massive deal. My objection is only if we try to extrapolate a general principle from those specific examples. Paul P.S.
I'm not sure of the meaning of "buried" here. What do you mean? We propose to revert 5 changes: [...] It's not only about specific changes, but more a discussion about a general policy to decide if a deprecated feature should stay until 3.10, or if it's ok to remove it in 3.9.
I agree with Brett, it was hard to understand your message. And I think the problem was that you tried to discuss both the *specific* proposal to delay 5 deprecations, and the *general* principle of when we stop preserving 2.7 compatibility layers. These two are very different questions and (as I say above) don't necessarily have the same answer. I think you should have posted *two* messages. The first very focused on the 5 deprecations, and the second, following on, saying something along the lines of "the previous message raises a broader question of how long we should carry 2.7 compatibility code before removing it - what do people think?". That would separate the general from the specific, while not flooding the list with separate threads for each deprecation (which I assume wasn't a serious suggestion on your part).
On 24 Jan 2020, at 08:40, Victor Stinner <vstinner@python.org> wrote:
The change is that Python 2.7 is no longer supported (since 2020-01-01).
Isn't there going to be a final Python 2.7 release in April 2020? I know that Twisted has moved its python 2 removal to match the April 2020 date. Other projects may have made the similar decisions. Barry
Victor
Le ven. 24 janv. 2020 à 09:19, Chris Angelico <rosuav@gmail.com> a écrit :
On Fri, Jan 24, 2020 at 7:05 PM Victor Stinner <vstinner@python.org> wrote:
We kept a compatibility layer with Python 2 on purpose, PEP 4 says:
"In order to facilitate writing code that works in both Python 2 & 3 simultaneously, any module that exists in both Python 3.5 and Python 2.7 will not be removed from the standard library until Python 2.7 is no longer supported as specified by PEP 373."
The rule was used since Python 3.0 until Python 3.8, but it changed in Python 3.9 which includes many incompatible changes for the first time in the Python 3 major version.
I'm sorry, I don't understand what 'changed'. Isn't that rule exactly WHY 3.9 is the removal point? Python 2.7 is no longer supported, and its final post-support release is scheduled earlier than 3.9's first beta and feature freeze. Doesn't that mean that PEP 4 is being followed precisely? What have I misunderstood?
ChrisA _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/GCCLU2O2... Code of Conduct: http://python.org/psf/codeofconduct/
-- Night gathers, and now my watch begins. It shall not end until my death. _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/6YLZXTOJ... Code of Conduct: http://python.org/psf/codeofconduct/
See pyfound.blogspot.com/2019/12/python-2-sunset.html for the details. January or April doesn't matter much for this thread, since Python 3.9.0 final is scheduled for October. Victor Le lun. 27 janv. 2020 à 23:13, Barry Scott <barry@barrys-emacs.org> a écrit :
On 24 Jan 2020, at 08:40, Victor Stinner <vstinner@python.org> wrote:
The change is that Python 2.7 is no longer supported (since 2020-01-01).
Isn't there going to be a final Python 2.7 release in April 2020?
I know that Twisted has moved its python 2 removal to match the April 2020 date. Other projects may have made the similar decisions.
Barry
Victor
Le ven. 24 janv. 2020 à 09:19, Chris Angelico <rosuav@gmail.com> a écrit :
On Fri, Jan 24, 2020 at 7:05 PM Victor Stinner <vstinner@python.org> wrote:
We kept a compatibility layer with Python 2 on purpose, PEP 4 says:
"In order to facilitate writing code that works in both Python 2 & 3 simultaneously, any module that exists in both Python 3.5 and Python 2.7 will not be removed from the standard library until Python 2.7 is no longer supported as specified by PEP 373."
The rule was used since Python 3.0 until Python 3.8, but it changed in Python 3.9 which includes many incompatible changes for the first time in the Python 3 major version.
I'm sorry, I don't understand what 'changed'. Isn't that rule exactly WHY 3.9 is the removal point? Python 2.7 is no longer supported, and its final post-support release is scheduled earlier than 3.9's first beta and feature freeze. Doesn't that mean that PEP 4 is being followed precisely? What have I misunderstood?
ChrisA _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/GCCLU2O2... Code of Conduct: http://python.org/psf/codeofconduct/
-- Night gathers, and now my watch begins. It shall not end until my death. _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/6YLZXTOJ... Code of Conduct: http://python.org/psf/codeofconduct/
-- Night gathers, and now my watch begins. It shall not end until my death.
On 1/27/2020 4:56 PM, Barry Scott wrote:
On 24 Jan 2020, at 08:40, Victor Stinner <vstinner@python.org> wrote: The change is that Python 2.7 is no longer supported (since 2020-01-01). Isn't there going to be a final Python 2.7 release in April 2020?
Normal patching has stopped, much as if the final .rc1 release candidate had been issued. I believe Benjamin want to hit the final release button live at PyCon. -- Terry Jan Reedy
24.01.20 01:36, Barry Warsaw пише:
Given that we’ve changed the release cadence to one major release per year, it doesn’t seem that much of a burden to revert and carry these changes forward into Python 3.10. And if it helps with the migration off of Python 2.7, then +1 from me.
There are enough breaking changes planned in 3.10. Even changing the minor version number to two-digit is a large breaking change.
Victor Stinner wrote:
Le jeu. 23 janv. 2020 à 20:45, Brett Cannon brett@python.org a écrit :
Two pieces of feedback on this. One, nose is a bad example because that project has been telling people for years to switch to nose2 so the fact that people have still not switched something that should mostly be a direct swap after years of being asked to does not motivate in wanting to postpone one more year for that project's benefit. ;) (And where did you get the download stats on libraries.io? https://libraries.io/pypi/nose doesn't show any download counts.) Oh right, download stats can be found at: https://pypistats.org/packages/nose Downloads last day: 150,272 Downloads last week: 824,557 Downloads last month: 3,077,860 150k downloads yesterday.
Ah, thanks! To put this in perspective, pytest had 631,008 downloads yesterday.
Two, you buried the list of things you would like to revert and postpone to Python 3.10. :) I'm not sure of the meaning of "buried" here. What do you mean?
Sorry, North American phrase. It means you put the important part way down towards the bottom. Another North American phrase that might apply is "burying the lead". Basically you put the interesting, important stuff so far down it was easy to overlook. :) My suggestion would have been to structure the email as "Fedora is seeing some transition problems for projects ATM. Reverting these 5 changes and postponing them to 3.10 would help a lot, and here is a bunch of detail about why we are making this proposal." IOW putting the list towards the top makes it easier to set expectations of what you're asking for while you try to justify the request, otherwise I could be reading most of that email as if you're suggesting we bring back the print statement or something. ;)
We propose to revert 5 changes:
Removed tostring/fromstring methods in array.array and base64 modules Removed collections aliases to ABC classes Removed fractions.gcd() function (which is similar to math.gcd()) Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier Removed old plistlib API: 2.7 doesn't have the new API
Was this copied-and-pasted from somewhere, hence why you explain to us how to use -X dev? Well, use whatever you want to see DeprecationWarning, as soon as you fix all of them in your project :-) PYTHONWARNINGS=default should be fine as well. IMO it's simpler to remind "-X dev" than using PYTHONWARNINGS=default. By the way, -X dev and PYTHONWARNINGS=default shows more warnings than just DeprecationWarning: they also show ImportWarning, PendingDeprecationWarning and ResourceWarning. -X dev helps to make your project ready for the next Python version. Basically it isn't easy to discuss the things you want to revert here, especially since you grouped them all together. Are you planning to start separate conversations on the (I think) 5 things you want to change? I'm not sure of what you mean. Would you prefer one mail thread per revert? It's not only about specific changes, but more a discussion about a general policy to decide if a deprecated feature should stay until 3.10, or if it's ok to remove it in 3.9. For example, another feature has been removed since we drafted this email: "The encoding parameter of json.loads() has been removed." If we follow the proposed policy, it should also be reverted. It seems like the incompatible change causing most issues is the
collections change. We can try to produce statistics per change if you prefer. I understand that you disagree to revert changes, but you would prefer to decide on a case by case basis. Victor Night gathers, and now my watch begins. It shall not end until my death.
Le ven. 24 janv. 2020 à 19:28, Brett Cannon <brett@python.org> a écrit :
(...) otherwise I could be reading most of that email as if you're suggesting we bring back the print statement or something. ;)
By the way, about adding back print statement....... nah, I'm just kidding :-D Victor
23.01.20 17:20, Victor Stinner пише:
Incompatible changes which require "if <python3>: (...) else: (...)" or "try: <python3 code> except (...): <python2 code>":
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
You had years to fix deprecation warnings in your code. And many projects did this long ago. If revert these removals in 3.9 because they break existing code we will have the same problem with removing them in 3.10, 13.11, etc. I consider breaking unmaintained code is an additional benefit of removing deprecated features. For example pycrypto was unmaintained and insecure for 6 years, but has 4 million downloads per month. It will not work in 3.9 because of removing time.clock. Other example is nose, unmaintained for 5 years, and superseded by nose2.
On 24.01.2020 11:08, Serhiy Storchaka wrote:
23.01.20 17:20, Victor Stinner пише:
Incompatible changes which require "if <python3>: (...) else: (...)" or "try: <python3 code> except (...): <python2 code>":
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
You had years to fix deprecation warnings in your code. And many projects did this long ago. If revert these removals in 3.9 because they break existing code we will have the same problem with removing them in 3.10, 13.11, etc.
I consider breaking unmaintained code is an additional benefit of removing deprecated features. For example pycrypto was unmaintained and insecure for 6 years, but has 4 million downloads per month. It will not work in 3.9 because of removing time.clock. Other example is nose, unmaintained for 5 years, and superseded by nose2.
AFAICS the justification for this proposition is the fact that the necessary fixes to existing code are much simpler if that code needs to support just 3.x-3.10 vs 2.7+3.x-3.10 . In this light, with the breaking changes, we essentially tell the users: "abandon 2.7 or do extra work". Victor alleges that after a year, the option to abandon 2.7 and not do that extra work will be easier to choose. So the dilemma is essentially how hard we push users to abandon 2.7 -- how much tax we incur on them for keeping its support.
_______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/DDVY3DG5... Code of Conduct: http://python.org/psf/codeofconduct/
-- Regards, Ivan
On Fri, Jan 24, 2020 at 12:12 AM Serhiy Storchaka <storchaka@gmail.com> wrote:
I consider breaking unmaintained code is an additional benefit of removing deprecated features.
I'd like to warn against this attitude (even though in the past I've occasionally said such things). I now think core Python should not be so judgmental. We've broken enough code for a lifetime with the Python 2 transition. Let's be *much* more conservative when we remove things from Python 3. Deprecation is fine, and we should look for other ways to handle the problem of unmaintained code. But we should not rush language or stdlib changes for this purpose. -- --Guido van Rossum (python.org/~guido) *Pronouns: he/him **(why is my pronoun here?)* <http://feministing.com/2015/02/03/how-using-they-as-a-singular-pronoun-can-c...>
On 25Jan.2020 0348, Guido van Rossum wrote:
On Fri, Jan 24, 2020 at 12:12 AM Serhiy Storchaka <storchaka@gmail.com <mailto:storchaka@gmail.com>> wrote:
I consider breaking unmaintained code is an additional benefit of removing deprecated features.
I'd like to warn against this attitude (even though in the past I've occasionally said such things). I now think core Python should not be so judgmental. We've broken enough code for a lifetime with the Python 2 transition. Let's be *much* more conservative when we remove things from Python 3. Deprecation is fine, and we should look for other ways to handle the problem of unmaintained code. But we should not rush language or stdlib changes for this purpose.
I'd like to *strongly* agree with this sentiment. Marking things as deprecated when we don't like them is a perfectly good way to advise against their use (and give ourselves permission to let bit-rot set in). But unless they are an actual maintenance burden, we gain literally nothing by removing them, and actively hurt our already-hurting users. As much as we know that 3.x->3.y is a major version change, many of our users don't think of them like that (in part because they come out so often). The more we can keep things working between them, warts or not, the better. Cheers, Steve
23.01.20 17:20, Victor Stinner пише:
* Removed collections aliases to ABC classes
Adding loud warning was one of largest compatibility breaking changes in 3.8, because many active projects treat warnings in tests as errors. I had doubts about removing them. On one side, they were deprecated for a very long time. On other side, most time it was silent deprecation, and the removal will affect too much projects. I agree with deferring this removal.
* Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier
io.open() (don't confuse with codecs.open(), which was recommended in the past) has other advantages. Builtin open() has problems with threads in Python 2.
* Removed tostring/fromstring methods in array.array and base64 modules * Removed fractions.gcd() function (which is similar to math.gcd()) * Removed old plistlib API: 2.7 doesn't have the new API
As for other changes, they look minor, and I think they affect not much code. In particular, plistlib is specific for macOS.
On 24. 01. 20 22:54, Serhiy Storchaka wrote:
23.01.20 17:20, Victor Stinner пише:
* Removed collections aliases to ABC classes
Adding loud warning was one of largest compatibility breaking changes in 3.8, because many active projects treat warnings in tests as errors. I had doubts about removing them. On one side, they were deprecated for a very long time. On other side, most time it was silent deprecation, and the removal will affect too much projects. I agree with deferring this removal.
Some data points: This has so far affected ~65 from ~200 projects that we found fail to build in Fedora w/3.9.
* Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier
~10
* Removed tostring/fromstring methods in array.array and base64 modules
~10
* Removed fractions.gcd() function (which is similar to math.gcd())
0
* Removed old plistlib API: 2.7 doesn't have the new API
2 -- Miro Hrončok -- Phone: +420777974800 IRC: mhroncok
Is it possible to change some of those deprecation from Warn-by-default (but can add a warning filter to make them error) to error-by-default (but can add a filter to make them only warnings). That way it is easy to fix large projects (even w/o modifying the source) in only a couple of lines of codes; and they buy themselves a year or so and remove in 3.10. That way you have a more gradual transition; and a clear signal to project with a last chance to fix issues. And that way can't complained they did not know as they have to opt-in.
Matthias Bussonnier wrote:
Is it possible to change some of those deprecation from Warn-by-default (but can add a warning filter to make them error) to error-by-default (but can add a filter to make them only warnings).
Yes, although it would be a change to how Python functions. Basically we would need a default warnings filter with either a very specific regex or a new warnings subclass which made silencing warning more opt-out instead of opt-in. But considering the vast amount of discussions we have had about when to even turn on warnings I don't know if this idea will get traction. -Brett
That way it is easy to fix large projects (even w/o modifying the source) in only a couple of lines of codes; and they buy themselves a year or so and remove in 3.10. That way you have a more gradual transition; and a clear signal to project with a last chance to fix issues. And that way can't complained they did not know as they have to opt-in.
Is it possible to change some of those deprecation from Warn-by-default (but can add a warning filter to make them error) to error-by-default (but can add a filter to make them only warnings). That way it is easy to fix large projects (even w/o modifying the source) in only a couple of lines of codes; and they buy themselves a year or so and remove in 3.10. That way you have a more gradual transition; and a clear signal to project with a last chance to fix issues. And that way can't complained they did not know as they have to opt-in.
On Fri, 24 Jan 2020 at 00:27, Victor Stinner <vstinner@python.org> wrote:
Miro and me consider that Python 3.9 is pushing too much pressure on projects maintainers to either abandon Python 2.7 right now...
Let's not conflate py2 EOL with what-should-go-into-py39.
https://copr.fedorainfracloud.org/coprs/g/python/python3.9/monitor/ - Has package failures. Some packages fail because of broken dependencies.
Please provide a better summary. I've looked at first 3 failed packages, and: * `afflib` only has these error lines: ``` warning: extra tokens at the end of %endif directive in line 186: %endif # with_python2 warning: extra tokens at the end of %endif directive in line 195: %endif # with_python3 ``` I don't know if that's why the build is considered failed, but that issue seems to be on the build script side. * `python-django-post_office` doesn't have any stderr output or other indication why the build failed. * `ansible-lint` same, does this package even use Python?
https://bugzilla.redhat.com/showdependencytree.cgi?id=PYTHON39 - Has open Python 3.9 bug reports for Fedora packages. Some problems
I feel it's still early days for py39 and well-maintained packages are expected to catch up. I, for one, just started testing our crud against py39a3 yesterday, and yes some tools and libraries are broken, so I've opened issues and might post a PR, depending. I get the gist from your email that you are (or fedora folks are) doing the same. So, what's the rationale to change py39 change set?
try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence
I thought that collections compatibility was kept up to 3.8 specifically because py2 was alive. No that that requirement is gone, so should the shim, right? Ofc., my assumption could be just wrong...
While if we remove collections.Sequence in 3.10, they will face this decision early in 2021 and they will simply fix their code by adding ".abc" at the proper places, not needing any more compatibility layers. Of course, there will be projects that will still have declared Python 2 support in 2021, but it will arguably not be that many.
I see some semi-abandoned projects, as well as actively maintained projects not addressing DeprecationWarnings. I don't know if some projects just need a harder push (will break, not just deprecated). Either way, my gut tells me, if py39 was changed and became very compatible, we'd see the same discussion in 2021.
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
my 2c: all of these are truly minor. I have a feeling that the discussion is just wrong altogether. It should not be "let's remove" or "let's keep". It should not be "let's keep x but not y". It should be something along the lines of: * here's a great tool that fedora has that shows which libraries need changes (your links + better output) * here's a build server, where existing packages (maybe not fedora) can be tested and built on py39, new ABI, etc. * [optional] it would be awesome if anyone could submit their repo url and the server would automatically build wheels for all arch,abi and publish to pypi * let's notify package maintainers (presumably done?) and if that doesn't yield the desired result, * let's organise the open-source community (via fame or bounties) and fix (or fork) those that are important (common [transitive] deps to many fedora packages) Cheers, d.
On 2020-01-30 07:12, Dima Tisnek wrote:
On Fri, 24 Jan 2020 at 00:27, Victor Stinner <vstinner@python.org> wrote:
Miro and me consider that Python 3.9 is pushing too much pressure on projects maintainers to either abandon Python 2.7 right now...
Let's not conflate py2 EOL with what-should-go-into-py39.
https://copr.fedorainfracloud.org/coprs/g/python/python3.9/monitor/ - Has package failures. Some packages fail because of broken dependencies.
Please provide a better summary. I've looked at first 3 failed packages, and:
* `afflib` only has these error lines: ``` warning: extra tokens at the end of %endif directive in line 186: %endif # with_python2 warning: extra tokens at the end of %endif directive in line 195: %endif # with_python3 ``` I don't know if that's why the build is considered failed, but that issue seems to be on the build script side.
Those are just warnings; ignore them. afflib didn't build because it's missing a dependency (libtermcap-devel). It wasn't counted in the ones with Python-related problems.
* `python-django-post_office` doesn't have any stderr output or other indication why the build failed.
Same situation (this time missing python3-django-jsonfield).
* `ansible-lint` same, does this package even use Python?
And that one is missing python3-passlib, which fails several tests; one of them being ImportError: cannot import name 'MutableMapping' from 'collections'.
https://bugzilla.redhat.com/showdependencytree.cgi?id=PYTHON39 - Has open Python 3.9 bug reports for Fedora packages. Some problems
I feel it's still early days for py39 and well-maintained packages are expected to catch up.
I, for one, just started testing our crud against py39a3 yesterday, and yes some tools and libraries are broken, so I've opened issues and might post a PR, depending.
I get the gist from your email that you are (or fedora folks are) doing the same.
Definitely! That's exactly what we're doing: helping the well-maintained packages catch up, via bug reports and pull requests as capacity allows. And now we're reporting back that the most common change these packages have to make is the collections.abc removal, and that since many projects are planning to drop py2 compat later in 2020, the pull request we need to send is ugly -- and it could be much nicer one-liner in 2021 if the collections.abc removal was postponed.
So, what's the rationale to change py39 change set?
try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence
I thought that collections compatibility was kept up to 3.8 specifically because py2 was alive. No that that requirement is gone, so should the shim, right? Ofc., my assumption could be just wrong...
While if we remove collections.Sequence in 3.10, they will face this decision early in 2021 and they will simply fix their code by adding ".abc" at the proper places, not needing any more compatibility layers. Of course, there will be projects that will still have declared Python 2 support in 2021, but it will arguably not be that many.
I see some semi-abandoned projects, as well as actively maintained projects not addressing DeprecationWarnings. I don't know if some projects just need a harder push (will break, not just deprecated). Either way, my gut tells me, if py39 was changed and became very compatible, we'd see the same discussion in 2021.
Not really. For projects that want to stay be py2/py3-compatible for the long term, the long "try/except ImportError" dance makes sense.
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
my 2c: all of these are truly minor.
I have a feeling that the discussion is just wrong altogether. It should not be "let's remove" or "let's keep". It should not be "let's keep x but not y".
It should be something along the lines of: * here's a great tool that fedora has that shows which libraries need changes (your links + better output) * here's a build server, where existing packages (maybe not fedora) can be tested and built on py39, new ABI, etc. * [optional] it would be awesome if anyone could submit their repo url and the server would automatically build wheels for all arch,abi and publish to pypi * let's notify package maintainers (presumably done?) and if that doesn't yield the desired result, * let's organise the open-source community (via fame or bounties) and fix (or fork) those that are important (common [transitive] deps to many fedora packages)
Yes! Except a few details, that's just what I'd like to do in the long term. Do you want to help?
Dima Tisnek writes:
I thought that collections compatibility was kept up to 3.8 specifically because py2 was alive. No that that requirement is gone, so should the shim, right?
Python 2 is still very much alive (even in a Python 3 venv :-þ): (analysis.venv) 01 16:56$ /usr/bin/python Python 2.7.10 (default, Feb 22 2019, 21:55:15) [GCC 4.2.1 Compatible Apple LLVM 10.0.1 (clang-1001.0.37.14)] on darwin Type "help", "copyright", "credits" or "license" for more information. >>> ^D We have stopped supporting Python 2 in the sense that we are no longer going to release source updates to Python 2. That doesn't necessarily mean we don't consider the needs of users who continue to depend on Python 2 for one reason or another, excellent or totally (in our oh-so-ever-'umble opinion) mistaken though that reason may be.
Ofc., my assumption could be just wrong...
It might be, it might not be, it might be situation-dependent (ie, decisions should be made case by case on a cost to Python core developers versus benefits to Python 2 users and those who support them in their libraries etc basis). I think it would certainly be reasonable to make the decision that we're going to go full speed ahead with Python 3 and start stripping out Python 2-only features in the stdlib code. I don't support that, but I'm not a core developer, so that's a +1.0e-17 for keeping the code. I get the feeling that Guido is also not in favor of a Procrustean[1] approach to Python 2 support in the Python 3 stdlib, but I could be wrong. Since he's not BDFL, that's merely indicative, but IMO it's a pretty strong indicator. What would really distress me however, is if we forget that we're not supporting *code*, we're supporting *users* by creating and maintaining code. Of course we can choose that user base, and may have to make painful decisions. But this community is about people supporting people, just like any community. Footnotes: [1] Look up "Procrustes" in Wikipedia. It's not a story to tell in this family setting. ;-)
I think this particular mess was caused by the hiding of “DeprecationWarning”s by default: If you don’t see any warnings cropping up in your production code, you don’t know you have to fix something. Some languages handle it like this: 1. Silent deprecation warning (deprecated in docs and/or hidden like in Python, the most eagle-eyed maintainers will see it and fix it) 2. Visible deprecation warning (people start seeing it and bug you in issues) 3. Hide things behind a flag: You’ll have breakage but you get a grace perior by enabling some scary looking environment variable or so (nobody can deny that things are getting serious) 4. Removal Going from 1 to 4 is of course pretty harsh, and this is what we’re doing. I propose that at least in the future, we have a schedule for every single deprecation to go to a gradual process like above. Best, Phil Am Fr., 31. Jan. 2020 um 05:21 Uhr schrieb Stephen J. Turnbull < turnbull.stephen.fw@u.tsukuba.ac.jp>:
Dima Tisnek writes:
I thought that collections compatibility was kept up to 3.8 specifically because py2 was alive. No that that requirement is gone, so should the shim, right?
Python 2 is still very much alive (even in a Python 3 venv :-þ):
(analysis.venv) 01 16:56$ /usr/bin/python Python 2.7.10 (default, Feb 22 2019, 21:55:15) [GCC 4.2.1 Compatible Apple LLVM 10.0.1 (clang-1001.0.37.14)] on darwin Type "help", "copyright", "credits" or "license" for more information. >>> ^D
We have stopped supporting Python 2 in the sense that we are no longer going to release source updates to Python 2. That doesn't necessarily mean we don't consider the needs of users who continue to depend on Python 2 for one reason or another, excellent or totally (in our oh-so-ever-'umble opinion) mistaken though that reason may be.
Ofc., my assumption could be just wrong...
It might be, it might not be, it might be situation-dependent (ie, decisions should be made case by case on a cost to Python core developers versus benefits to Python 2 users and those who support them in their libraries etc basis).
I think it would certainly be reasonable to make the decision that we're going to go full speed ahead with Python 3 and start stripping out Python 2-only features in the stdlib code. I don't support that, but I'm not a core developer, so that's a +1.0e-17 for keeping the code. I get the feeling that Guido is also not in favor of a Procrustean[1] approach to Python 2 support in the Python 3 stdlib, but I could be wrong. Since he's not BDFL, that's merely indicative, but IMO it's a pretty strong indicator.
What would really distress me however, is if we forget that we're not supporting *code*, we're supporting *users* by creating and maintaining code. Of course we can choose that user base, and may have to make painful decisions. But this community is about people supporting people, just like any community.
Footnotes: [1] Look up "Procrustes" in Wikipedia. It's not a story to tell in this family setting. ;-) _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/WX4LA55V... Code of Conduct: http://python.org/psf/codeofconduct/
On 2020-01-23 07:20, Victor Stinner wrote:
Python 3.9 introduces many small incompatible changes which broke tons
There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number. Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead? Well, either of these strategies sound logical to me: - Python 4.0 with removal of all of the Python 3-era deprecations - Continuing Python 3.1X with no breaks In other words, we should keep compatibility, or not. In any case, from the looks of it these will be tiny breaks compared to the Unicode transition. Cheers, -Mike
On 01/31/2020 10:47 AM, Mike Miller wrote:
On 2020-01-23 07:20, Victor Stinner wrote:
Python 3.9 introduces many small incompatible changes which broke tons
There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number.
Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead?
I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations. -- ~Ethan~
On 2020-01-31 19:58, Ethan Furman wrote:
On 01/31/2020 10:47 AM, Mike Miller wrote:
On 2020-01-23 07:20, Victor Stinner wrote:
Python 3.9 introduces many small incompatible changes which broke tons
There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number.
Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead?
I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations.
It has a certain appeal for me too.
On 2020-01-31 19:47, Mike Miller wrote:
On 2020-01-23 07:20, Victor Stinner wrote:
Python 3.9 introduces many small incompatible changes which broke tons
There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number.
Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead? Well, either of these strategies sound logical to me:
- Python 4.0 with removal of all of the Python 3-era deprecations - Continuing Python 3.1X with no breaks
In other words, we should keep compatibility, or not. In any case, from the looks of it these will be tiny breaks compared to the Unicode transition.
The Unicode transition also looked very small back when 3.0 was planned. It only takes one such not-so-little thing to make a big breaking release like 3.0. And even if all the changes were little, I wouldn't want to inflict 10 years of papercuts at once. When the changes are rolled out gradually across minor releases, those that cause unforeseen trouble in real-world code can be identified in the alphas/betas, and rethought/reverted if necessary. Ethan Furman wrote:
I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations.
I don't. I hope the 10-year (and counting) transition from Python 2 to Python 3 will not become a tradition. I'd rather iterate on making removals less drastic (e.g. by making the DeprecationWarnings more visible). Iterate with a feedback loop, rather than do a one-time change and hope that everything goes well.
On Mon, Feb 3, 2020 at 10:53 AM Petr Viktorin <encukou@gmail.com> wrote:
On 2020-01-31 19:47, Mike Miller wrote:
On 2020-01-23 07:20, Victor Stinner wrote:
Python 3.9 introduces many small incompatible changes which broke tons
There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number.
Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead? Well, either of these strategies sound logical to me:
- Python 4.0 with removal of all of the Python 3-era deprecations - Continuing Python 3.1X with no breaks
In other words, we should keep compatibility, or not. In any case, from the looks of it these will be tiny breaks compared to the Unicode transition.
The Unicode transition also looked very small back when 3.0 was planned. It only takes one such not-so-little thing to make a big breaking release like 3.0. And even if all the changes were little, I wouldn't want to inflict 10 years of papercuts at once.
I agree with the sentiment that gradual deprecations are more easily managed, this statement about Python 3.0 is not true. The unicode transition was never thought to be small, and that's *why* 3.0 was such a big change. We knew it was going to break everything, so we took the opportunity to break more things, like the behaviour of indexing bytestrings. (Bytestrings were even going to be *mutable* for a while.) I think we can all agree that it was a mistake, and that's certainly something we don't want to repeat: even if we defer actual removal of features for a x.0.0 release, it must not become carte blanche for breaking things.
When the changes are rolled out gradually across minor releases, those that cause unforeseen trouble in real-world code can be identified in the alphas/betas, and rethought/reverted if necessary.
That's also the case if things are (loudly) deprecated *but not removed* until a x.0.0 release. The replacement would already be in use for years before the old way would go away.
Ethan Furman wrote:
I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations.
I don't. I hope the 10-year (and counting) transition from Python 2 to Python 3 will not become a tradition. I'd rather iterate on making removals less drastic (e.g. by making the DeprecationWarnings more visible). Iterate with a feedback loop, rather than do a one-time change and hope that everything goes well. _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/CHMOFONI... Code of Conduct: http://python.org/psf/codeofconduct/
-- Thomas Wouters <thomas@python.org> Hi! I'm an email virus! Think twice before sending your email to help me spread!
On 2020-02-03 12:55, Thomas Wouters wrote:
On Mon, Feb 3, 2020 at 10:53 AM Petr Viktorin <encukou@gmail.com <mailto:encukou@gmail.com>> wrote:
On 2020-01-31 19:47, Mike Miller wrote: > > On 2020-01-23 07:20, Victor Stinner wrote: > > Python 3.9 introduces many small incompatible changes which broke tons > > > There's a well-known and established way of signaling breaking changes > in software platforms—it is to increment the major version number. > > Rather than debating the merits of breaking code on 3.9 or 3.10, > wouldn't it make more sense to do it in a Python 4.0 instead? Well, > either of these strategies sound logical to me: > > - Python 4.0 with removal of all of the Python 3-era deprecations > - Continuing Python 3.1X with no breaks > > In other words, we should keep compatibility, or not. In any case, from > the looks of it these will be tiny breaks compared to the Unicode > transition.
The Unicode transition also looked very small back when 3.0 was planned. It only takes one such not-so-little thing to make a big breaking release like 3.0. And even if all the changes were little, I wouldn't want to inflict 10 years of papercuts at once.
I agree with the sentiment that gradual deprecations are more easily managed, this statement about Python 3.0 is not true. The unicode transition was never thought to be small, and that's *why* 3.0 was such a big change.
Alright, "very small" is an overstatement. But it did seem much smaller than it turned out to be. https://docs.python.org/3.0/whatsnew/3.0.html lists it as the last of the big breaking changes, while most of the porting efforts were spent on it.
We knew it was going to break everything, so we took the opportunity to break more things, like the behaviour of indexing bytestrings. (Bytestrings were even going to be *mutable* for a while.) I think we can all agree that it was a mistake, and that's certainly something we don't want to repeat: even if we defer actual removal of features for a x.0.0 release, it must not become carte blanche for breaking things. >
When the changes are rolled out gradually across minor releases, those that cause unforeseen trouble in real-world code can be identified in the alphas/betas, and rethought/reverted if necessary.
That's also the case if things are (loudly) deprecated *but not removed* until a x.0.0 release. The replacement would already be in use for years before the old way would go away.
I fear that this is only so if no unexpected issues come up. If we do a loud deprecation, how can we be sure we did it right?
Ethan Furman wrote: > I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations.
I don't. I hope the 10-year (and counting) transition from Python 2 to Python 3 will not become a tradition. I'd rather iterate on making removals less drastic (e.g. by making the DeprecationWarnings more visible). Iterate with a feedback loop, rather than do a one-time change and hope that everything goes well.
On Mon, 3 Feb 2020 13:18:46 +0100 Petr Viktorin <encukou@gmail.com> wrote:
I agree with the sentiment that gradual deprecations are more easily managed, this statement about Python 3.0 is not true. The unicode transition was never thought to be small, and that's *why* 3.0 was such a big change.
Alright, "very small" is an overstatement. But it did seem much smaller than it turned out to be. https://docs.python.org/3.0/whatsnew/3.0.html lists it as the last of the big breaking changes, while most of the porting efforts were spent on it.
I don't think the order in this page has any relationship to the difficulty of each change. I'm not sure there was any particular reason to order them in this way, but it seems to me that the easier to understand changes were put first. AFAIR, we were all acutely aware that the text model overhaul was going to be the hardest change for many users (especially those who have never encountered non-ASCII input). Most other changes in Python 3 have a predetermined context-free repair, while migrating to the new text model requires *thinking* about the context in which each Python 2 bytestring is used. It forces people to think about the problem they're solving in ways many of them didn't think of it before. Regards Antoine.
On 2020-02-03 16:44, Antoine Pitrou wrote:
On Mon, 3 Feb 2020 13:18:46 +0100 Petr Viktorin <encukou@gmail.com> wrote:
I agree with the sentiment that gradual deprecations are more easily managed, this statement about Python 3.0 is not true. The unicode transition was never thought to be small, and that's *why* 3.0 was such a big change.
Alright, "very small" is an overstatement. But it did seem much smaller than it turned out to be. https://docs.python.org/3.0/whatsnew/3.0.html lists it as the last of the big breaking changes, while most of the porting efforts were spent on it.
I don't think the order in this page has any relationship to the difficulty of each change. I'm not sure there was any particular reason to order them in this way, but it seems to me that the easier to understand changes were put first.
AFAIR, we were all acutely aware that the text model overhaul was going to be the hardest change for many users (especially those who have never encountered non-ASCII input). Most other changes in Python 3 have a predetermined context-free repair, while migrating to the new text model requires *thinking* about the context in which each Python 2 bytestring is used. It forces people to think about the problem they're solving in ways many of them didn't think of it before.
I stand corrected. Thank you!
On 2020-02-03 01:50, Petr Viktorin wrote:
When the changes are rolled out gradually across minor releases, those that cause unforeseen trouble in real-world code can be identified in the alphas/betas, and rethought/reverted if necessary.
To be clear, my suggestion was to maintain gradual deprecations and warnings, but have a single removal event at the start of a new version major number. So there will be many years of betas and releases to haggle over. Also, I believe it is possible to separate the "mechanically fixable" breaks from larger changes in fundamentals. Few folks are willing to entertain or stomach the latter at this point in the lifecycle of Python. If one happens to occur, scheduling it for X.4 instead of X+1.0 doesn't help much, and may serve to obscure it. In some sense, distributing the breaks avoids potential/temporary bad press at the cost of easy planning. However, it feels like a very regular, easy to understand process should trump that in the long run. As a refinement to the idea above, perhaps a sub rule could be added: - No deprecations should appear in a X.9 release to give folks time to prepare. -Mike
On 03/02/2020 22:06, Mike Miller wrote:
On 2020-02-03 01:50, Petr Viktorin wrote:
When the changes are rolled out gradually across minor releases, those that cause unforeseen trouble in real-world code can be identified in the alphas/betas, and rethought/reverted if necessary.
To be clear, my suggestion was to maintain gradual deprecations and warnings, but have a single removal event at the start of a new version major number. So there will be many years of betas and releases to haggle over.
I think that just enables laziness. "We don't need to worry about the deprecations, nothing is going to happen for years yet," is more or less what happened with the Python2 to Python3 shift. People naturally enjoy adding shiny new features to their projects over maintenance.
As a refinement to the idea above, perhaps a sub rule could be added:
- No deprecations should appear in a X.9 release to give folks time to prepare.
That feels like a very bad idea indeed. If you don't tell people things are going away, they definitely won't prepare for that. -- Rhodri James *-* Kynesim Ltd
On 2020-02-04 04:16, Rhodri James wrote:
I think that just enables laziness. "We don't need to worry about the deprecations, nothing is going to happen for years yet," is more or less what happened with the Python2 to Python3 shift. People naturally enjoy adding shiny new features to their projects over maintenance.
That's an uncharitable interpretation. Not everyone has the luxury of fixing broken code every year. (New release schedule) It's not how most of the industry works either.
As a refinement to the idea above, perhaps a sub rule could be added:
- No deprecations should appear in a X.9 release to give folks time to prepare.
That feels like a very bad idea indeed. If you don't tell people things are going away, they definitely won't prepare for that.
This point is not suggesting to break things and "not tell anyone." Rather to avoid any compatibility breaks on the ultimate release of a series to give folks a minimum of two years to react. -Mike
On 2020-01-23 07:20, Victor Stinner wrote:
Python 3.9 introduces many small incompatible changes which broke tons
On 2020-01-31 19:47, Mike Miller wrote:
There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number.
Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead? Well, either of these strategies sound logical to me:
- Python 4.0 with removal of all of the Python 3-era deprecations - Continuing Python 3.1X with no breaks
In other words, we should keep compatibility, or not. In any case, from the looks of it these will be tiny breaks compared to the Unicode transition.
Ethan Furman wrote:
I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations.
Petr Viktorin wrote:
I don't. I hope the 10-year (and counting) transition from Python 2 to Python 3 will not become a tradition. I'd rather iterate on making removals less drastic (e.g. by making the DeprecationWarnings more visible). Iterate with a feedback loop, rather than do a one-time change and hope that everything goes well.
As a user I would much rather know that my 3.2 code worked in every version of 3.x, and not have to make changes in 3.5 and 3.7 and 3.11. Talk about death by paper cuts! I'd either be stuck updating already working code to get the benefits of the latest Python 3, or having multiple versions of Python 3 on my system. Both options are galling. Having the latest Python 2, the latest Python 3, and the latest Python 4 is much more palatable. -- ~Ethan~
Ethan Furman wrote:
Python 3.9 introduces many small incompatible changes which broke tons On 2020-01-31 19:47, Mike Miller wrote: There's a well-known and established way of signaling breaking changes in software platforms—it is to increment the major version number. Rather than debating the merits of breaking code on 3.9 or 3.10, wouldn't it make more sense to do it in a Python 4.0 instead? Well, either of these strategies sound logical to me:
Python 4.0 with removal of all of the Python 3-era deprecations Continuing Python 3.1X with no breaks
In other words, we should keep compatibility, or not. In any case, from the looks of it these will be tiny breaks compared to the Unicode transition. Ethan Furman wrote: I've gotta say, I like that plan. Instead of going to x.10, go to x+1.0. Every ten years we bump the major version and get rid of all the deprecations. Petr Viktorin wrote: I don't. I hope the 10-year (and counting) transition from Python 2 to Python 3 will not become a tradition. I'd rather iterate on making removals less drastic (e.g. by making the DeprecationWarnings more visible). Iterate with a feedback loop, rather than do a one-time change and hope that everything goes well. As a user I would much rather know that my 3.2 code worked in every version of 3.x, and not have to make changes in 3.5 and 3.7 and 3.11. Talk about death by paper cuts! I'd either be stuck updating already working code to get the benefits of the latest Python 3, or having multiple versions of Python 3 on my system. Both options are galling. Having the latest Python 2, the latest Python 3, and the latest Python 4 is much more
On 2020-01-23 07:20, Victor Stinner wrote: palatable.
Until you're being asked to maintain all of that for a decade. We paid a major price keeping Python 2 alive for over a decade. Now I'm not saying it wasn't the right thing to do considering what we changed, but for the stuff we are talking about removing it doesn't require a massive rewrite on the behalf of users. And we know from experience anything that is left in will get used no matter how loudly we try to broadcast that fact (and we know people still do not have a habit of running their code with warnings turned on). I think people also forget that prior to worrying about maintaining backwards-compatibility with Python 2 we deprecated for a release and then we removed (so an 18 month deprecation period). Python survived, users survived, and we all had more time for other things than trying to keep deprecated stuff from completely rotting out (look at inspect and trying to wedge keyword-only arguments into argspec() and friends as to prices paid to keeping stuff we would have shifted users off of as having a cost). And I know some people think that if we flat-out say we won't touch deprecated code that it should be enough, but even triaging issues for deprecated stuff as "won't fix" is still not free.
On 2020-02-03 17:00, Brett Cannon wrote:
Until you're being asked to maintain all of that for a decade. We paid a major price keeping Python 2 alive for over a decade. Now I'm not saying it wasn't the right thing to do considering what we changed, but for the stuff we are talking about removing it doesn't require a massive rewrite on the behalf of users. And we know from experience anything that is left in will get used no matter how loudly we try to broadcast that fact (and we know people still do not have a habit of running their code with warnings turned on).
Potentially up to a decade, often not. We seem to agree that these are most likely minor things like duplicate import paths and such. The maintenance burden is low on these items. Also, due to skittishness, removals have drug out to five plus releases anyway. So, why the aversion to formalizing the process, instead of "winging it" every release? Why force devs/end-users to maintain state about what release is compatible or waste time on investigation? Really, this was solved decades ago. Everyone knows what to do when a major version number changes. Other language platforms are not afraid to change them, likely because the breaks were typically minor and not break-the-world events. Just like we're talking about here. More broadly, an agile cadence is great for innovative apps, not so great for mature platforms. Platforms are supposed to provide stability and predictability, and the popular ones do. -Mike
Mike Miller wrote:
Until you're being asked to maintain all of that for a decade. We paid a major price keeping Python 2 alive for over a decade. Now I'm not saying it wasn't the right thing to do considering what we changed, but for the stuff we are talking about removing it doesn't require a massive rewrite on the behalf of users. And we know from experience anything that is left in will get used no matter how loudly we try to broadcast that fact (and we know people still do not have a habit of running their code with warnings turned on). Potentially up to a decade, often not. We seem to agree that these are most likely minor things like duplicate import
On 2020-02-03 17:00, Brett Cannon wrote: paths and such. The maintenance burden is low on these items.
Please be careful making that claim. Over my 16 years of helping manage this project I can tell you that claim is not universally true no matter how small and simple you think something is.
Also, due to skittishness, removals have drug out to five plus releases anyway.
That has been due to Python 2 compatibility which is about to no longer be a concern. Prior to Python 2 this was not the case.
So, why the aversion to formalizing the process, instead of "winging it" every release?
There is work to formalize it in PEP 387. We are just dealing with an odd case of Python 2 compatibility which is a one-off situation.
Why force devs/end-users to maintain state about what release is compatible or waste time on investigation? Really, this was solved decades ago. Everyone knows what to do when a major version number changes. Other language platforms are not afraid to change them, likely because the breaks were typically minor and not break-the-world events.
Python predates semver. Assume every feature/minor release potentially has a breaking change and we have (hopefully) been raising warnings to the user for the past two years about the breaking change coming.
On 2020-02-04 12:10, Brett Cannon wrote:
Please be careful making that claim. Over my 16 years of helping manage this project I can tell you that claim is not universally true no matter how small and simple you think something is.
The answer to that concern is to not break compatibility in the first place, and/or revert it when the mistake is discovered. It happens. As mentioned, significant breaks are now discouraged. The code is already written, so I'd argue keeping it deprecated potentially a few years extra in a regular process is not an undue burden.
Python predates semver. Assume every feature/minor release potentially has a breaking change and we have (hopefully) been raising warnings to the user for the past two years about the breaking change coming.
Hmm, was thinking about MS-DOS when I wrote that. Still, I don't think this a good system having breaking changes in the minor version number. It's not traditional and it's not newfangled either, it's relatively unique. On that note, just noticed that Wikipedia has its own section for Python under the "Software Versioning" article: The PSF has published PEP 440 -- Version Identification and Dependency Specification,[16] outlining their own flexible (complicated) scheme, that defines an epoch segment, a release segment, pre-release and post-release segments and a development release segment. https://en.wikipedia.org/wiki/Software_versioning#Python The main argument here seems to be it this uncommon versioning scheme saves core developer time, (unsaid) at the expense of dev or end-user time. As there are probably a million end-users per Python contributor, I argue this is not a good tradeoff to make. -Mike
On Tue, 4 Feb 2020 at 22:10, Mike Miller <python-dev@mgmiller.net> wrote:
On 2020-02-04 12:10, Brett Cannon wrote:
Please be careful making that claim. Over my 16 years of helping manage this project I can tell you that claim is not universally true no matter how small and simple you think something is.
The answer to that concern is to not break compatibility in the first place, and/or revert it when the mistake is discovered. It happens.
That sounds to me like an argument for stagnation. We already take backwards compatibility very seriously. If you're suggesting that we don't even have the option of deprecating things, then I'm not sure how you expect the language to evolve.
As mentioned, significant breaks are now discouraged. The code is already written, so I'd argue keeping it deprecated potentially a few years extra in a regular process is not an undue burden.
The point has already been made that "keeping code around but deprecated" isn't the problem, it's bug triage, telling people who report problems with deprecated code that "this is not going to be fixed", educating people who copy/paste old and deprecated code and wonder why they are now getting warnings with it, etc. I think it's probably up to the core devs (who do the work) to judge what is an "undue burden", but if you do want to try to judge it, please take all of those extra tasks into account when reckoning up.
Python predates semver. Assume every feature/minor release potentially has a breaking change and we have (hopefully) been raising warnings to the user for the past two years about the breaking change coming.
Hmm, was thinking about MS-DOS when I wrote that. Still, I don't think this a good system having breaking changes in the minor version number. It's not traditional and it's not newfangled either, it's relatively unique.
I have seen many systems that have breaking changes in the minor version, as you describe it. (Java 1.1 to Java 1.4 are a good example - so good that they "rebranded" the minor version as the major version with 1.5!). Restricting compatibility breaks to the major version isn't that important in my experience. IMO, the version number isn't anywhere near as important as the *process* of handling backward compatibility and giving users a good transition path. Knowing what the rules being followed are is the most important thing - and people trying to claim that semver concepts apply to a system that doesn't use it, just because the version number syntax looks like semver, is one of the worst ways to confuse the message IMO.
On that note, just noticed that Wikipedia has its own section for Python under the "Software Versioning" article:
The PSF has published PEP 440 -- Version Identification and Dependency Specification,[16] outlining their own flexible (complicated) scheme, that defines an epoch segment, a release segment, pre-release and post-release segments and a development release segment.
PEP 440 is not about Python's own version scheme, it's a standard for versioning Python *packages* that are published on PyPI and managed by tools like pip. It's complicated because it has to allow for many different legacy versioning schemes which were in use by existing projects when the PEP was defined. Ideally we would have chosen a simpler scheme, but that would have required breaking many existing projects' version schemes, so we chose not to do that. So it's actually a great demonstration of how not being able to deprecate old usages *adds* complexity and maintenance burden. But it's an irrelevant side issue when talking about the Python language version itself, as PEP 440 doesn't apply to that.
The main argument here seems to be it this uncommon versioning scheme saves core developer time, (unsaid) at the expense of dev or end-user time. As there are probably a million end-users per Python contributor, I argue this is not a good tradeoff to make.
No, it's that conserving the *extremely* limited resource that is freely donated, volunteer core developer time is essential if we want to have a viable core development team at all. If in doing so, we need to have a controlled deprecation process that places some of the burden of keeping up to date on end users who benefit for no cost from the huge amount of donated software and support that makes up what people think of as "Python", then that should be an entirely acceptable trade-off. Paul
On 2020-02-04 14:40, Paul Moore wrote:
The answer to that concern is to not break compatibility in the first place, and/or revert it when the mistake is discovered. It happens.
That sounds to me like an argument for stagnation. We already take backwards compatibility very seriously. If you're suggesting that we don't even have the option of deprecating things, then I'm not sure how you expect the language to evolve.
No, I chose the wrong phrasing there. Have been arguing for deprecations to occur on schedule but removals to occur later. By "not break compatibility…" I meant not to break it in a large, destructive manner. Small ones, sure.
The point has already been made that "keeping code around but deprecated" isn't the problem, it's bug triage, telling people who report problems with deprecated code that "this is not going to be fixed", educating people who copy/paste old and deprecated code and wonder why they are now getting warnings with it, etc. I think it's probably up to the core devs (who do the work) to judge what is an "undue burden", but if you do want to try to judge it, please take all of those extra tasks into account when reckoning up.
This happens already and won't go away. I'm arguing that a *very* predictable release process helps and doesn't hurt in this department, resulting in fewer questions. Instead of every release is a unique snowflake to be considered. Still have questions? Go to FAQ #42.
Restricting compatibility breaks to the major version isn't that important in my experience. IMO, the version number isn't anywhere near as important as the *process* of handling backward
It matters, when say Ubuntu drops X.4 when X.6 comes out and X.6 isn't backwards compatible. You get forced to update or find one of the various workarounds. I argue this should happen less frequently at predictable times.
No, it's that conserving the *extremely* limited resource that is freely donated, volunteer core developer time is essential if we want to have a viable core development team at all.
Sorry, didn't mean that part as an attack, rather an illustration of the tradeoff being made. Now if PSF needs help to build a more predictable release process, well then, by all means ask. But that can't happen until the decision is made to do it. -Mike
On 2020-02-04, 01:00 GMT, Brett Cannon wrote:
I think people also forget that prior to worrying about maintaining backwards-compatibility with Python 2 we deprecated for a release and then we removed (so an 18 month deprecation period).
But then you mustn’t filter out deprecation warnings. And yes, we have to relearn that deprecation warning means the incoming doom, not something to ignore. Best, Matěj -- https://matej.ceplovi.cz/blog/, Jabber: mcepl@ceplovi.cz GPG Finger: 3C76 A027 CA45 AD70 98B5 BC1D 7920 5802 880B C9D8 Thou shalt not nede to be afrayed for any bugges by night. -- Coverdale's 1535 translation of Psalm 91 (or Christopher Higley's 1972 thesis explaining why programmers' are nocturnal beasts)
Hi, I created an issue: "Keep deprecated features in Python 3.9 to ease migration from Python 2.7, but remove in Python 3.10" https://bugs.python.org/issue39674 And proposed a first pull request to add again collections.Mapping: https://github.com/python/cpython/pull/18545 Victor Le jeu. 23 janv. 2020 à 16:20, Victor Stinner <vstinner@python.org> a écrit :
Hi,
Python 3.9 introduces many small incompatible changes which broke tons of Python projects, including popular projects, some of them being unmaintained but still widely used (like nose, last release in 2015).
Miro and me consider that Python 3.9 is pushing too much pressure on projects maintainers to either abandon Python 2.7 right now (need to update the CI, the documentation, warn users, etc.), or to introduce a *new* compatibility layer to support Python 3.9: layer which would be dropped as soon as Python 2.7 support will be dropped (soon-ish).
Python 3.9 is too early to accumulate so many incompatible changes on purpose, some incompatible changes like the removal of collections aliases to ABC should be reverted, and reapplied on Python 3.10. Python 3.9 would be the last release which still contain compatibility layers for Python 2.7.
Said differently, we request to maintain the small compatibility layer in Python for one more cycle, instead of requesting every single project maintainer to maintain it on their side. We consider that the general maintenance burden is lower if it's kept in Python for now.
== Fedora COPR notify packages broken by Python 3.9 ==
In Python 3.9, Victor introduced tons of incompatible changes at the beginning of the devcycle. His plan was to push as many as possible, and later decide what to do... This time has come :-) He wrote PEP 606 "Python Compatibility Version" and we wrote PEP 608 "Coordinated Python release", but both have been rejected. At least, it seems like most people agreed that having a CI to get notified of broken projects should help.
We are updating the future Fedora 33 to replace Python 3.8 with Python 3.9. We are using a tool called "COPR" which is like a sandbox and can be seen as the CI discussed previously. It rebuilds Fedora using Python 3.9 as /usr/bin/python3 (and /usr/bin/python !). We now have a good overview of broken packages and which incompatible changes broke most packages.
https://fedoraproject.org/wiki/Changes/Python3.9 - Describes the Fedora change.
https://copr.fedorainfracloud.org/coprs/g/python/python3.9/monitor/ - Has package failures. Some packages fail because of broken dependencies.
https://bugzilla.redhat.com/showdependencytree.cgi?id=PYTHON39 - Has open Python 3.9 bug reports for Fedora packages. Some problems have been fixed upstream already before reaching Fedora, most are only fixed when the Fedora maintainers report the problems back to upstream maintainers.
Right now, there are 150+ packages broken by Python 3.9 incompatible changes.
== Maintenance burden ==
Many Python projects have not yet removed Python 2 support and Python 2 CI. It's not that they would be in the "we will not drop Python 2 support ever" camp, it's just that they have not done it yet. Removing Python 2 compatibility code from the codebase and removing it from the documentation and metadata and CI is a boring task, doesn't bring anything to users, and it might take a new major release of the library. At this point, we are very early in 2020 to expect most projects to have already done this.
At the same time, we would like them to support Python 3.9 as soon in the release cycle as possible. By removing Python 2 compatibility layers from the 3.9 standard library, we are forcing the projects maintainers to re-invent their own compatibility layers and copy-paste stuff like this around. Example:
try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence
While if we remove collections.Sequence in 3.10, they will face this decision early in 2021 and they will simply fix their code by adding ".abc" at the proper places, not needing any more compatibility layers. Of course, there will be projects that will still have declared Python 2 support in 2021, but it will arguably not be that many.
While it's certainly tempting to have "more pure" code in the standard library, maintaining the compatibility shims for one more release isn't really that big of a maintenance burden, especially when comparing with dozens (hundreds?) of third party libraries essentially maintaining their own.
An good example of a broken package is the nose project which is no longer maintained (according to their website): the last release was in 2015. It remains a very popular test runner. According to libraries.io, it has with 3 million downloads per month, 41.7K dependent repositories and 325 dependent packages. We patched nose in Fedora to fix Python 3.5, 3.6, 3.8 and now 3.9 compatibility issues. People installing nose from PyPI with "pip install" get the unmaintained flavor which is currently completely broken on Python 3.9.
Someone should take over the nose project and maintain it again, or every single project using nose should pick another tool (unittest, nose2, pytest, whatever else). Both options will take a lot of time.
== Request to revert some incompatible changes ==
See https://docs.python.org/dev/whatsnew/3.9.html#removed
Incompatible changes which require "if <python3>: (...) else: (...)" or "try: <python3 code> except (...): <python2 code>":
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
== Kept incompatible changes ==
Ok-ish incompatible changes (mostly only affects compatiblity libraries like six):
* _dummy_thread and dummy_threading modules removed: broke six, nine and future projects. six and nine are already fixed for 3.9.
OK incompatible changes (can be replaced by the same code on 2.7 and 3.9):
* isAlive() method of threading.Thread has been removed: Thread.is_alive() method is available in 2.7 and 3.9 * xml.etree.ElementTree.getchildren() and xml.etree.ElementTree.getiterator() methods are removed from 3.9, but list()/iter() works in 2.7 and 3.9
== Call to check for DeprecationWarning in your own projects ==
You must pay attention to DeprecationWarning in Python 3.9: it will be the last "compatibility layer" release, incompatible changes will be reapplied to Python 3.10.
For example, you can use the development mode to see DeprecationWarning and ResourceWarning: use the "-X dev" command line option or set the PYTHONDEVMODE=1 environment variable. Or you can use the PYTHONWARNINGS=default environment variable to see DeprecationWarning.
You might even want to treat all warnings as errors to ensure that you don't miss any when you run your test suite in your CI. You can use PYTHONWARNINGS=error, and combine it with PYTHONDEVMODE=1.
Warnings filters can be used to ignore warnings in third party code, see the documentation: https://docs.python.org/dev/library/warnings.html#the-warnings-filter
-- Victor Stinner and Miro Hrončok for Fedora
-- Night gathers, and now my watch begins. It shall not end until my death.
And proposed a first pull request to add again collections.Mapping:
https://github.com/python/cpython/pull/18245 FYI, this patch is adding a fixer to handling abstract base classes. It will be helpful for the migration to new changes. I hope someone reviews this. 2020년 2월 18일 (화) 오후 8:44, Victor Stinner <vstinner@python.org>님이 작성:
Hi,
I created an issue:
"Keep deprecated features in Python 3.9 to ease migration from Python 2.7, but remove in Python 3.10" https://bugs.python.org/issue39674
And proposed a first pull request to add again collections.Mapping:
https://github.com/python/cpython/pull/18545
Victor
Le jeu. 23 janv. 2020 à 16:20, Victor Stinner <vstinner@python.org> a écrit :
Hi,
Python 3.9 introduces many small incompatible changes which broke tons of Python projects, including popular projects, some of them being unmaintained but still widely used (like nose, last release in 2015).
Miro and me consider that Python 3.9 is pushing too much pressure on projects maintainers to either abandon Python 2.7 right now (need to update the CI, the documentation, warn users, etc.), or to introduce a *new* compatibility layer to support Python 3.9: layer which would be dropped as soon as Python 2.7 support will be dropped (soon-ish).
Python 3.9 is too early to accumulate so many incompatible changes on purpose, some incompatible changes like the removal of collections aliases to ABC should be reverted, and reapplied on Python 3.10. Python 3.9 would be the last release which still contain compatibility layers for Python 2.7.
Said differently, we request to maintain the small compatibility layer in Python for one more cycle, instead of requesting every single project maintainer to maintain it on their side. We consider that the general maintenance burden is lower if it's kept in Python for now.
== Fedora COPR notify packages broken by Python 3.9 ==
In Python 3.9, Victor introduced tons of incompatible changes at the beginning of the devcycle. His plan was to push as many as possible, and later decide what to do... This time has come :-) He wrote PEP 606 "Python Compatibility Version" and we wrote PEP 608 "Coordinated Python release", but both have been rejected. At least, it seems like most people agreed that having a CI to get notified of broken projects should help.
We are updating the future Fedora 33 to replace Python 3.8 with Python 3.9. We are using a tool called "COPR" which is like a sandbox and can be seen as the CI discussed previously. It rebuilds Fedora using Python 3.9 as /usr/bin/python3 (and /usr/bin/python !). We now have a good overview of broken packages and which incompatible changes broke most packages.
https://fedoraproject.org/wiki/Changes/Python3.9 - Describes the Fedora change.
https://copr.fedorainfracloud.org/coprs/g/python/python3.9/monitor/ - Has package failures. Some packages fail because of broken dependencies.
https://bugzilla.redhat.com/showdependencytree.cgi?id=PYTHON39 - Has open Python 3.9 bug reports for Fedora packages. Some problems have been fixed upstream already before reaching Fedora, most are only fixed when the Fedora maintainers report the problems back to upstream maintainers.
Right now, there are 150+ packages broken by Python 3.9 incompatible changes.
== Maintenance burden ==
Many Python projects have not yet removed Python 2 support and Python 2 CI. It's not that they would be in the "we will not drop Python 2 support ever" camp, it's just that they have not done it yet. Removing Python 2 compatibility code from the codebase and removing it from the documentation and metadata and CI is a boring task, doesn't bring anything to users, and it might take a new major release of the library. At this point, we are very early in 2020 to expect most projects to have already done this.
At the same time, we would like them to support Python 3.9 as soon in the release cycle as possible. By removing Python 2 compatibility layers from the 3.9 standard library, we are forcing the projects maintainers to re-invent their own compatibility layers and copy-paste stuff like this around. Example:
try: from collections.abc import Sequence except ImprotError: # Python 2.7 doesn't have collections.abc from collections import Sequence
While if we remove collections.Sequence in 3.10, they will face this decision early in 2021 and they will simply fix their code by adding ".abc" at the proper places, not needing any more compatibility layers. Of course, there will be projects that will still have declared Python 2 support in 2021, but it will arguably not be that many.
While it's certainly tempting to have "more pure" code in the standard library, maintaining the compatibility shims for one more release isn't really that big of a maintenance burden, especially when comparing with dozens (hundreds?) of third party libraries essentially maintaining their own.
An good example of a broken package is the nose project which is no longer maintained (according to their website): the last release was in 2015. It remains a very popular test runner. According to libraries.io, it has with 3 million downloads per month, 41.7K dependent repositories and 325 dependent packages. We patched nose in Fedora to fix Python 3.5, 3.6, 3.8 and now 3.9 compatibility issues. People installing nose from PyPI with "pip install" get the unmaintained flavor which is currently completely broken on Python 3.9.
Someone should take over the nose project and maintain it again, or every single project using nose should pick another tool (unittest, nose2, pytest, whatever else). Both options will take a lot of time.
== Request to revert some incompatible changes ==
See https://docs.python.org/dev/whatsnew/3.9.html#removed
Incompatible changes which require "if <python3>: (...) else: (...)" or "try: <python3 code> except (...): <python2 code>":
* Removed tostring/fromstring methods in array.array and base64 modules * Removed collections aliases to ABC classes * Removed fractions.gcd() function (which is similar to math.gcd()) * Remove "U" mode of open(): having to use io.open() just for Python 2 makes the code uglier * Removed old plistlib API: 2.7 doesn't have the new API
== Kept incompatible changes ==
Ok-ish incompatible changes (mostly only affects compatiblity libraries like six):
* _dummy_thread and dummy_threading modules removed: broke six, nine and future projects. six and nine are already fixed for 3.9.
OK incompatible changes (can be replaced by the same code on 2.7 and 3.9):
* isAlive() method of threading.Thread has been removed: Thread.is_alive() method is available in 2.7 and 3.9 * xml.etree.ElementTree.getchildren() and xml.etree.ElementTree.getiterator() methods are removed from 3.9, but list()/iter() works in 2.7 and 3.9
== Call to check for DeprecationWarning in your own projects ==
You must pay attention to DeprecationWarning in Python 3.9: it will be the last "compatibility layer" release, incompatible changes will be reapplied to Python 3.10.
For example, you can use the development mode to see DeprecationWarning and ResourceWarning: use the "-X dev" command line option or set the PYTHONDEVMODE=1 environment variable. Or you can use the PYTHONWARNINGS=default environment variable to see DeprecationWarning.
You might even want to treat all warnings as errors to ensure that you don't miss any when you run your test suite in your CI. You can use PYTHONWARNINGS=error, and combine it with PYTHONDEVMODE=1.
Warnings filters can be used to ignore warnings in third party code, see the documentation: https://docs.python.org/dev/library/warnings.html#the-warnings-filter
-- Victor Stinner and Miro Hrončok for Fedora
-- Night gathers, and now my watch begins. It shall not end until my death. _______________________________________________ Python-Dev mailing list -- python-dev@python.org To unsubscribe send an email to python-dev-leave@python.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://mail.python.org/archives/list/python-dev@python.org/message/AOEVRTYS... Code of Conduct: http://python.org/psf/codeofconduct/
-- Software Development Engineer at Kakao corp. Tel: +82 010-3353-9127 Email: donghee.na92@gmail.com | denny.i@kakaocorp.com Linkedin: https://www.linkedin.com/in/dong-hee-na-2b713b49/
Hi, I proposed a second change to restore the "U" open() mode (universal newlines): https://github.com/python/cpython/pull/18767 Again, it's only kept in Python 3.9 to ease migration from Python 2, but will be removed in Python 3.10. By the way, I also added a new "You should check for DeprecationWarning in your code" section to "What's New In Python 3.9" document, to strongly advice users to check for deprecation warnings in their code: https://docs.python.org/dev/whatsnew/3.9.html#you-should-check-for-deprecati... Victor Le mar. 18 févr. 2020 à 12:37, Victor Stinner <vstinner@python.org> a écrit :
Hi,
I created an issue:
"Keep deprecated features in Python 3.9 to ease migration from Python 2.7, but remove in Python 3.10" https://bugs.python.org/issue39674
And proposed a first pull request to add again collections.Mapping:
https://github.com/python/cpython/pull/18545
Victor
On Wed, 4 Mar 2020 at 23:09, Victor Stinner <vstinner@python.org> wrote:
By the way, I also added a new "You should check for DeprecationWarning in your code" section to "What's New In Python 3.9" document, to strongly advice users to check for deprecation warnings in their code: https://docs.python.org/dev/whatsnew/3.9.html#you-should-check-for-deprecati...
That's a good idea. Having a section explaining how to check for deprecation warnings, and listing out some notable ones that are expected to become errors in the next release would actually make sense as a routine feature of the What's New document. Cheers, Nick. -- Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia
participants (29)
-
Antoine Pitrou -
Barry Scott -
Barry Warsaw -
Brett Cannon -
Chris Angelico -
Dima Tisnek -
Dong-hee Na -
Eric V. Smith -
Ethan Furman -
Guido van Rossum -
Ivan Levkivskyi -
Ivan Pozdeev -
Matthias Bussonnier -
Matěj Cepl -
Mike Miller -
Miro Hrončok -
MRAB -
Nick Coghlan -
Paul Moore -
Petr Viktorin -
Philipp A. -
Rhodri James -
Robert Collins -
Serhiy Storchaka -
Stephen J. Turnbull -
Steve Dower -
Terry Reedy -
Thomas Wouters -
Victor Stinner