Hi all, after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi. I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0 Cheers Oliver
Hi, test_alwayscopy failed on OS X only with py27 today with "Operation not permitted" and I did not have time to look into it yet, so I rather postpone the release until I found out whether that is a real problem or just some testing glitch on my end. Cheers Oliver On Mon, 30 Jan 2017 at 00:10 Oliver Bestwalter <oliver@bestwalter.de> wrote: Hi all, after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi. I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0 Cheers Oliver
Could you please add root/pypi to the bases of your index? Otherwise "devpi test" doesn't work, because it can't find any other package, like pytest. Regards, Florian Schulze On 30 Jan 2017, at 19:54, Oliver Bestwalter wrote:
Hi,
test_alwayscopy failed on OS X only with py27 today with "Operation not permitted" and I did not have time to look into it yet, so I rather postpone the release until I found out whether that is a real problem or just some testing glitch on my end.
Cheers Oliver
On Mon, 30 Jan 2017 at 00:10 Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hi all,
after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi.
I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0
Cheers Oliver
_______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
Hi Florian, oh right - sorry, it's fixed now. Cheers Oliver On Wed, 1 Feb 2017 at 11:49 Florian Schulze <mail@florian-schulze.net> wrote:
Could you please add root/pypi to the bases of your index? Otherwise "devpi test" doesn't work, because it can't find any other package, like pytest.
Regards, Florian Schulze
On 30 Jan 2017, at 19:54, Oliver Bestwalter wrote:
Hi,
test_alwayscopy failed on OS X only with py27 today with "Operation not permitted" and I did not have time to look into it yet, so I rather postpone the release until I found out whether that is a real problem or just some testing glitch on my end.
Cheers Oliver
On Mon, 30 Jan 2017 at 00:10 Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hi all,
after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi.
I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0
Cheers Oliver
_______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
2.6.0.dev1 tests seem to be fine for me on OS X. The Py 3.6 error is most likely a problem with my installation, I have that elsewhere as well. Regards, Florian Schulze On 1 Feb 2017, at 20:10, Oliver Bestwalter wrote:
Hi Florian,
oh right - sorry, it's fixed now.
Cheers Oliver
On Wed, 1 Feb 2017 at 11:49 Florian Schulze <mail@florian-schulze.net> wrote:
Could you please add root/pypi to the bases of your index? Otherwise "devpi test" doesn't work, because it can't find any other package, like pytest.
Regards, Florian Schulze
On 30 Jan 2017, at 19:54, Oliver Bestwalter wrote:
Hi,
test_alwayscopy failed on OS X only with py27 today with "Operation not permitted" and I did not have time to look into it yet, so I rather postpone the release until I found out whether that is a real problem or just some testing glitch on my end.
Cheers Oliver
On Mon, 30 Jan 2017 at 00:10 Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hi all,
after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi.
I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0
Cheers Oliver
_______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
o.k. then the py27 error must have been something weird on my end. Thanks Oliver On Wed, 1 Feb 2017 at 21:23 Florian Schulze <mail@florian-schulze.net> wrote:
2.6.0.dev1 tests seem to be fine for me on OS X. The Py 3.6 error is most likely a problem with my installation, I have that elsewhere as well.
Regards, Florian Schulze
On 1 Feb 2017, at 20:10, Oliver Bestwalter wrote:
Hi Florian,
oh right - sorry, it's fixed now.
Cheers Oliver
On Wed, 1 Feb 2017 at 11:49 Florian Schulze <mail@florian-schulze.net> wrote:
Could you please add root/pypi to the bases of your index? Otherwise "devpi test" doesn't work, because it can't find any other package, like pytest.
Regards, Florian Schulze
On 30 Jan 2017, at 19:54, Oliver Bestwalter wrote:
Hi,
test_alwayscopy failed on OS X only with py27 today with "Operation not permitted" and I did not have time to look into it yet, so I rather postpone the release until I found out whether that is a real problem or just some testing glitch on my end.
Cheers Oliver
On Mon, 30 Jan 2017 at 00:10 Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hi all,
after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi.
I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0
Cheers Oliver
_______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
Hi Walter, I've triggered testing of all Python versions on AppVeyor and Travis: AppVeyor: https://ci.appveyor.com/project/nicoddemus/devpi-cloud-tester/build/1.0.31 Travis: https://travis-ci.org/nicoddemus/devpi-cloud-tester/builds/197467586 If you are curious, I use https://github.com/nicoddemus/devpi-cloud-tester. I always use it for pytest releases to get clean test results that don't depend on local configurations. Feel free to fork it and use that for future releases if you like. Cheers! Bruno. On Wed, Feb 1, 2017 at 4:19 AM Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hi all,
after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi.
I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0
Cheers Oliver _______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
Duh, that is such a simple and in hindsight obvious hack! You could add OS X to Travis. Regards, Florian Schulze On 1 Feb 2017, at 23:37, Bruno Oliveira wrote:
Hi Walter,
I've triggered testing of all Python versions on AppVeyor and Travis:
AppVeyor: https://ci.appveyor.com/project/nicoddemus/devpi-cloud-tester/build/1.0.31 Travis: https://travis-ci.org/nicoddemus/devpi-cloud-tester/builds/197467586
If you are curious, I use https://github.com/nicoddemus/devpi-cloud-tester.
I always use it for pytest releases to get clean test results that don't depend on local configurations. Feel free to fork it and use that for future releases if you like.
Cheers! Bruno.
On Wed, Feb 1, 2017 at 4:19 AM Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hi all,
after some teething problems I am now all set up to do releases and will do the first one tomorrow. All tests are green on Linux and Windows. Will try to run tests also on a Mac I have lying around at work tomorrow and then push it to pypi.
I would appreciate some extra eye balls, so if someone wants to look at the package, it is here: https://devpi.net/obestwalter/dev/tox/2.6.0
Cheers Oliver _______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
_______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
On Thu, Feb 2, 2017 at 7:46 AM holger krekel <holger@merlinux.eu> wrote:
heh, i can see how you evolved a new prename for oliver :)
Oops, sorry Oliver! I know Oliver's name of course, it is just that "Walter" always jumps at me, and yesterday I was really tired and did not notice the slip-up. :) On Thu, Feb 2, 2017 at 7:36 AM Florian Schulze <mail@florian-schulze.net> wrote:
Duh, that is such a simple and in hindsight obvious hack!
You could add OS X to Travis.
I thought about it, but I seem to recall Travis doesn't support both OS X and Linux in the same .travis.yaml file (I think I've seen this somewhere in @conda-forge's docs, not sure where though). Alternatively, I guess I could also follow @conda-forge's and use Travis for OS X, CircleCI for Linux and AppVeyor for Windows. Cheers, Bruno.
* Bruno Oliveira <nicoddemus@gmail.com> [2017-02-02 11:05:43 +0000]:
You could add OS X to Travis.
I thought about it, but I seem to recall Travis doesn't support both OS X and Linux in the same .travis.yaml file (I think I've seen this somewhere in @conda-forge's docs, not sure where though).
That's wrong ;) https://github.com/The-Compiler/qutebrowser/blob/master/.travis.yml#L5-L23 Florian -- http://www.the-compiler.org | me@the-compiler.org (Mail/XMPP) GPG: 916E B0C8 FD55 A072 | http://the-compiler.org/pubkey.asc I love long mails! | http://email.is-not-s.ms/
On Thu, Feb 2, 2017 at 9:25 AM Florian Bruhin <me@the-compiler.org> wrote:
That's wrong ;) https://github.com/The-Compiler/qutebrowser/blob/master/.travis.yml#L5-L23
Ahh cool, thanks for the link! So it is simple to implement it directly in Travis. :) Cheers, Bruno.
Hey, I started thinking along those lines while preparing the release, but wanted to get the first release out old skool like Holger sketched out for me. I did already set up vagrant boxes for (arch)linux and Windows(10) with all interpreters installed which would add portable dev and test envs to the project. I did not add the Vagrantfile to the project yet, because I wanted to hear first, if this is considered a good idea. I like having a Vagrantfile in a project for tests and to help others getting started quickly, but it obviously has not only upsides and carries a (in my view small, one it's set up) maintenance burden. @Bruno that's very cool indeed, I will have a closer look at that as soon as I can and add it to our CI. Preparing a release is should be basically preparation free in my ideal world, so that every successfull built can be released right away without any extra testing necessary. I was also wondering if there is something speaking against introducing Ronnies https://github.com/pypa/setuptools_scm for tag based versioning? Love that tool - one less thing to worry about, just tag the thing and off you go :) Cheers Oliver P.S. @Bruno my "maiden" name is Walter and my friends in school used to call me that, you must have channeled that I guess ... I also think it's funny that your last name sounds a bit like my first name, so I might start calling you Oliveira and you can call me Walter :D On Thu, 2 Feb 2017 at 12:25 Florian Bruhin <me@the-compiler.org> wrote:
* Bruno Oliveira <nicoddemus@gmail.com> [2017-02-02 11:05:43 +0000]:
You could add OS X to Travis.
I thought about it, but I seem to recall Travis doesn't support both OS X and Linux in the same .travis.yaml file (I think I've seen this somewhere in @conda-forge's docs, not sure where though).
That's wrong ;) https://github.com/The-Compiler/qutebrowser/blob/master/.travis.yml#L5-L23
Florian
-- http://www.the-compiler.org | me@the-compiler.org (Mail/XMPP) GPG: 916E B0C8 FD55 A072 | http://the-compiler.org/pubkey.asc I love long mails! | http://email.is-not-s.ms/ _______________________________________________ tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
On Thu, Feb 2, 2017 at 11:01 AM Oliver Bestwalter <oliver@bestwalter.de> wrote:
@Bruno that's very cool indeed, I will have a closer look at that as soon as I can and add it to our CI.
Cool, let me know if I can help. I was also wondering if there is something speaking against introducing
Ronnies https://github.com/pypa/setuptools_scm for tag based versioning? Love that tool - one less thing to worry about, just tag the thing and off you go :)
I also like it and use it in all my plugins. You have to trust your CI dutifully tests the package, which when testing with tox is usually the case. I recall Holger is not a big fan of automatic releases though. P.S. @Bruno my "maiden" name is Walter and my friends in school used to
call me that, you must have channeled that I guess ... I also think it's funny that your last name sounds a bit like my first name, so I might start calling you Oliveira and you can call me Walter :D
Heh deal! :D Cheers, Bruno.
On Thu, Feb 02, 2017 at 13:17 +0000, Bruno Oliveira wrote:
On Thu, Feb 2, 2017 at 11:01 AM Oliver Bestwalter <oliver@bestwalter.de> wrote:
I was also wondering if there is something speaking against introducing
Ronnies https://github.com/pypa/setuptools_scm for tag based versioning? Love that tool - one less thing to worry about, just tag the thing and off you go :)
I also like it and use it in all my plugins. You have to trust your CI dutifully tests the package, which when testing with tox is usually the case. I recall Holger is not a big fan of automatic releases though.
i am fine with automating releases. Just wasn't convinced yet that a setting a VCS tag should trigger a release. I am fine if you want to try it out with this or maybe better later releases. holger
On Thu, 2 Feb 2017 at 14:20 holger krekel <holger@merlinux.eu> wrote:
i am fine with automating releases. Just wasn't convinced yet that a setting a VCS tag should trigger a release. I am fine if you want to try it out with this or maybe better later releases.
I'll release 2.6.0 old skool tomorrow (this time for real!) and will phase these things in over time to make my own work more pleasant and to make myself superfluous in the long run :) - I don't believe in automatic releases either, but I like one-click releases. @all Thanks for all the help and input, very appreciated! Cheers Oliver
Hello all, 2.6.0 is released. I was not able to push from my devpi index though. I got: devpi --debug push tox==2.6.0 pypi:pypi using pypirc /home/oliver/.pypirc PUSH https://devpi.net/obestwalter/dev 502 Bad Gateway: { "type": "actionlog", "result": [ [ 500, "register", "tox", "2.6.0" ] ] } Which does not tell me what's wrong (the register in the result surprises me though). Does anyone know what could be the problem or how I can get better debug info? I guess it's something obvious because my experience with devpi is still very limited. I did not want to delay the release any further though, so I did the old python setup.py sdist bdist_wheel upload -r pypi with the packages I had pushed to my index. If you encounter any problems let me know. Cheers Oliver On Thu, 2 Feb 2017 at 14:30 Oliver Bestwalter <oliver@bestwalter.de> wrote:
On Thu, 2 Feb 2017 at 14:20 holger krekel <holger@merlinux.eu> wrote:
i am fine with automating releases. Just wasn't convinced yet that a setting a VCS tag should trigger a release. I am fine if you want to try it out with this or maybe better later releases.
I'll release 2.6.0 old skool tomorrow (this time for real!) and will phase these things in over time to make my own work more pleasant and to make myself superfluous in the long run :) - I don't believe in automatic releases either, but I like one-click releases.
@all Thanks for all the help and input, very appreciated!
Cheers Oliver
Hi Oliver, Not sure about the error on devpi upload, but regarding uploading the packages manually I think you are better off using twine in the future. Twine uploads pre-built packages and uses SSL to communicate with PyPI. I think running "python setup.py sdist bdist_wheel upload -r pypi" you have built new packages and uploaded them instead (which is no big deal). Cheers, Bruno. On Fri, Feb 3, 2017 at 1:02 PM Oliver Bestwalter <oliver@bestwalter.de> wrote:
Hello all,
2.6.0 is released.
I was not able to push from my devpi index though. I got:
devpi --debug push tox==2.6.0 pypi:pypi using pypirc /home/oliver/.pypirc PUSH https://devpi.net/obestwalter/dev 502 Bad Gateway: { "type": "actionlog", "result": [ [ 500, "register", "tox", "2.6.0" ] ] }
Which does not tell me what's wrong (the register in the result surprises me though). Does anyone know what could be the problem or how I can get better debug info? I guess it's something obvious because my experience with devpi is still very limited.
I did not want to delay the release any further though, so I did the old
python setup.py sdist bdist_wheel upload -r pypi
with the packages I had pushed to my index.
If you encounter any problems let me know.
Cheers Oliver
On Thu, 2 Feb 2017 at 14:30 Oliver Bestwalter <oliver@bestwalter.de> wrote:
On Thu, 2 Feb 2017 at 14:20 holger krekel <holger@merlinux.eu> wrote:
i am fine with automating releases. Just wasn't convinced yet that a setting a VCS tag should trigger a release. I am fine if you want to try it out with this or maybe better later releases.
I'll release 2.6.0 old skool tomorrow (this time for real!) and will phase these things in over time to make my own work more pleasant and to make myself superfluous in the long run :) - I don't believe in automatic releases either, but I like one-click releases.
@all Thanks for all the help and input, very appreciated!
Cheers Oliver
On Fri, Feb 03, 2017 at 15:02 +0000, Oliver Bestwalter wrote:
Hello all,
2.6.0 is released.
I was not able to push from my devpi index though. I got:
devpi --debug push tox==2.6.0 pypi:pypi using pypirc /home/oliver/.pypirc PUSH https://devpi.net/obestwalter/dev 502 Bad Gateway: { "type": "actionlog", "result": [ [ 500,
this means that pypi.python.org responded with a 500 internal server error when trying to register. Not sure why. can't remember that this happened to me yet. hogler
"register", "tox", "2.6.0" ] ] }
Which does not tell me what's wrong (the register in the result surprises me though). Does anyone know what could be the problem or how I can get better debug info? I guess it's something obvious because my experience with devpi is still very limited.
I did not want to delay the release any further though, so I did the old
python setup.py sdist bdist_wheel upload -r pypi
with the packages I had pushed to my index.
If you encounter any problems let me know.
Cheers Oliver
On Thu, 2 Feb 2017 at 14:30 Oliver Bestwalter <oliver@bestwalter.de> wrote:
On Thu, 2 Feb 2017 at 14:20 holger krekel <holger@merlinux.eu> wrote:
i am fine with automating releases. Just wasn't convinced yet that a setting a VCS tag should trigger a release. I am fine if you want to try it out with this or maybe better later releases.
I'll release 2.6.0 old skool tomorrow (this time for real!) and will phase these things in over time to make my own work more pleasant and to make myself superfluous in the long run :) - I don't believe in automatic releases either, but I like one-click releases.
@all Thanks for all the help and input, very appreciated!
Cheers Oliver
On Fri, Feb 3, 2017 at 2:16 PM holger krekel <holger@merlinux.eu> wrote:
this means that pypi.python.org responded with a 500 internal server error when trying to register. Not sure why. can't remember that this happened to me yet.
Just this week I tried to use "python setup.py register" to update some classifiers on some of my plugins and it failed saying the "register" command is now obsolete. Didn't investigate much though. Cheers, Bruno.
On Fri, Feb 03, 2017 at 16:51 +0000, Bruno Oliveira wrote:
On Fri, Feb 3, 2017 at 2:16 PM holger krekel <holger@merlinux.eu> wrote:
this means that pypi.python.org responded with a 500 internal server error when trying to register. Not sure why. can't remember that this happened to me yet.
Just this week I tried to use "python setup.py register" to update some classifiers on some of my plugins and it failed saying the "register" command is now obsolete. Didn't investigate much though.
fixing metadata is also why i used register only sometimes. Just tried with one of my less known projects to "setup.py register" and it worked for pypi.python.org. hum ... holger
Cheers, Bruno.
PyPI doesn't support "register" anymore. The new upload location is warehouse. This is from my ~/.pypirc: [warehouse] repository:https://upload.pypi.io/legacy/ you can add the repository line to your [pypi] section or whatever you use for "devpi push" after the "pypi:...", then it should work. I pushed several releases from devpi.net and my local devpi recently. Regards, Florian Schulze On 3 Feb 2017, at 18:05, holger krekel wrote:
On Fri, Feb 03, 2017 at 16:51 +0000, Bruno Oliveira wrote:
On Fri, Feb 3, 2017 at 2:16 PM holger krekel <holger@merlinux.eu> wrote:
this means that pypi.python.org responded with a 500 internal server error when trying to register. Not sure why. can't remember that this happened to me yet.
Just this week I tried to use "python setup.py register" to update some classifiers on some of my plugins and it failed saying the "register" command is now obsolete. Didn't investigate much though.
fixing metadata is also why i used register only sometimes.
Just tried with one of my less known projects to "setup.py register" and it worked for pypi.python.org.
hum ...
holger
Cheers, Bruno.
tox-dev mailing list tox-dev@python.org https://mail.python.org/mm3/mailman3/lists/tox-dev.python.org/
On Wed, Feb 1, 2017 at 8:37 PM Bruno Oliveira <nicoddemus@gmail.com> wrote:
AppVeyor: https://ci.appveyor.com/project/nicoddemus/devpi-cloud-tester/build/1.0.31 Travis: https://travis-ci.org/nicoddemus/devpi-cloud-tester/builds/197467586
Just a few comments and a question: Travis failed because it didn't have a "3.5" interpreter; that's because I chose the "3.6" environment, which I think is not part of the official images yet because it downloads and installs a 3.6 interpreter on the spot, and that image for some reason doesn't contain "3.5". AppVeyor failed due to reaching the 1h timeout. :( But it tested all environments successfully for the sdist package, and was just finishing with the wheel. Question: is there a way to make "devpi test" use only sdist or wheel? If that is possible, I could split the testing in two jobs, dodging the time out issue and speeding up things. Cheers, Bruno.
On 2 Feb 2017, at 12:10, Bruno Oliveira wrote:
Question: is there a way to make "devpi test" use only sdist or wheel? If that is possible, I could split the testing in two jobs, dodging the time out issue and speeding up things.
Not yet, but I had the same issue before, so it will come :) Regards, Florian Schulze
participants (5)
-
Bruno Oliveira
-
Florian Bruhin
-
Florian Schulze
-
holger krekel
-
Oliver Bestwalter