The PEP process, while *miles* better then anything I've seen in any other development process I've ever seen, has some deficiencies. I'll try to point them out, and to suggest some partial solution: 1. Users are not sure who to post PEP questions/remarks to: Python dev? the original author? The Python devver they know best? 2. It is the responsiblity of the PEP author to add open questions/pertinent remarks to the PEP. 3. Mail about the PEP which contains important discussion is lost. Proposal: (still some shaky areas though): Why not use Squishdot for PEPs? The PEP would be the "article", and remarks could be comments. I think (but I'm not sure) Squishdot has the ability to edit/moderate comments, and retroactively change the article. If not, then my second best hope is some kind of Wiki system, with self imposed article/comments-like structure. Immediate problem I'm trying to solve: PEP 228, the numeric model PEP, has inspired a lot of discussion, and very interesting comments. However for someone who doesn't remember all the e-mails by heart, wading through the archives is much too time consuming. In addition, personal mail sent to me is not available for public discussion. Problems with proposed solution: no e-mail notification. I think Zwiki does have an option for e-mail notification, though. Don't know about Squishdot. -- Moshe Zadka <sig@zadka.site.co.il> This is a signature anti-virus. Please stop the spread of signature viruses!
Moshe Zadka wrote:
Perhaps the PEP should specify this by e.g. using a header Comments-To: email-address The email address could then be made "usable" on the PEP web pages.
The original idea was for PEP authors to include the important parts of those discussions in summary in the PEP -- AFAIK, this is not done :-(
Hmm, wouldn't something like the SF bug manager be the ideal platform ? Perhaps someone could get the SF guys to add something like an "idea" manager with similar functionality but different content to SF ?! In the meantime, I think PEP discussions ought to taken to the sig mailing lists using python-dev as fallback solution. -- Marc-Andre Lemburg ______________________________________________________________________ Company: http://www.egenix.com/ Consulting: http://www.lemburg.com/ Python Pages: http://www.lemburg.com/python/
On Wed, 22 Nov 2000, M.-A. Lemburg wrote:
I'm trying to do this. However, summarising every e-mail sent to me about this PEP is quite a burden.
Yeah, that would work to.
In the meantime, I think PEP discussions ought to taken to the sig mailing lists using python-dev as fallback solution.
I think most of the PEPs are not in the charter of any new sig, and python-dev is invitation only forum. Perhaps having pep-xxx-sig would solve it too? -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
Moshe Zadka wrote:
I don't think people would sign up for each and every PEP... and you know how hard it is to even move a discussion from one forum to another -- this either doesn't work out or kills discussion. I think the simplest solution would be something like the SF bug manager. I'm pretty sure that other projects on SF have similar problems, so this might be worthwhile dicussing with the SF folks. The technology and user interface are there, it would only take a few new tables in the database. -- Marc-Andre Lemburg ______________________________________________________________________ Company: http://www.egenix.com/ Consulting: http://www.lemburg.com/ Python Pages: http://www.lemburg.com/python/
On Wed, 22 Nov 2000, Trent Mick wrote:
A Roundup mailing list with a category for each PEP?
OK, when we get over the make-Moshe-feel-stupid-for-not-proposing-roundup part, just let me know <wink>. Yep, Roundup is perfect for this! -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
"MZ" == Moshe Zadka <moshez@zadka.site.co.il> writes:
MZ> The PEP process, while *miles* better then anything I've seen in MZ> any other development process I've ever seen, has some (Not that we've used the PEP process for much of anything yet. Except, I guess, extended print, which I've grown quite fond of.) MZ> deficiencies. I'll try to point them out, and to suggest some MZ> partial solution: I don't like your partial solution, so I'll try to pick about the problems with the process <0.8 wink>. Put another way: I can't envision Slashdot as a useful design forum, but am interested in improving the PEP process. MZ> 1. Users are not sure who to post PEP questions/remarks to: MZ> Python dev? the original author? The Python devver they know MZ> best? In the absence of a reason to post elsewhere, the comments ought to be sent to the PEP author. If the message is intended to provoke wider discussion, then the user can post it on any relevent forum (and cc the PEP author). A Unicode PEP might be discussed on the i18n-sig; a Web PEP on the python-web-modules list; a change to core Python on python-dev. I don't think there will ever be a rule that says: "Comments on PEP 1812 must be posted to the pep-1812 web forum." Discussion should occur in the forum in which it is most relevent. Perhaps we could all send our comments to Andrew, and he could write a bi-weekly pep-comment summary <wink>. Is the problem that you don't know where to post comments on someone else's PEP, or that you are having trouble keeping track of discussions in multiple places? MZ> 2. It is the responsiblity of the PEP author to add open MZ> questions/pertinent remarks to the PEP. What is the deficiency with this approach? The PEP is supposed to present a coherent proposal and design for a new language feature. It is the designer's responsibility to write a good design document; several people can share responsibility for the design. Are you saying it's a problem that the PEPs aren't open to modification by anyone? (The last thing we need is someone who doesn't understand what a first-class function is messing with the scoping PEP <0.2 wink>.) The designer is responsible for discussing trade-offs and alternatives to justify her design. This is where the responsibility to address questions and comments comes from. MZ> 3. Mail about the PEP which contains important discussion is MZ> lost. How? I've got all the mail on the nested static scopes PEP. Is the problem just the personal burden of keeping track of lots of mail messages discussing a PEP? Any discussion that occurs on a mailing list will be preserved in the mailing list archive. That's not lost. So the only problem would be with discussion that occurs in private email, where everyone deletes copies of the mail. I wonder if the key problem is not having an authoritative mail archive for the PEP. One part of the Scheme RFI process that we did not copy was having a mail archive for each document. Perhaps this would be sufficient to address your concerns. If you, the PEP author, receive a comment by private email, you could bounce it to the archive. the-biggest-problem-with-the-PEP-process-is-no-one-writing-PEPs-ly y'rs, Jeremy
The key problem, as Jeremy noted, was that there is no archive of discussion about the PEP. I had a much more long winded and roundabout way of saying that. Also, a minor related problem is that when the only address on the PEP points to me (for example), I get all the mail about the PEP personally -- and I'm loath to bounce it to a public forum. *Having* a public forum which is *indicated in the PEP* would solve that. I could phrase that as a point-by-point answer to Jeremy (and originally did) but I decided that that would be idiotic. -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
Moshe Zadka writes:
Having a public forum isn't quite what's needed (though useful); much of the time, it's more important to have a public *archive*. I'm sure there's a way to set things up a mail handler that maintains an mbox file with collected mails on a topic, and a pipermail browsable version of that should be available, and each PEP should point to the appropriate archive. (It should also clearly state that relevant emails may be added to the archive at the recipient's discretion!) Using a mail handler for this would allow the handler to be added to the CC: on the emails and generally collect replies without any additional work on the part of the PEP author. The handler could be CC'd when the PEP is posted so that replies to the posting go there as well. Now we just need someone interested in writing a handler. ;-{ -Fred -- Fred L. Drake, Jr. <fdrake at acm.org> PythonLabs at Digital Creations
On Wed, 22 Nov 2000, Fred L. Drake, Jr. wrote:
Ummmmm....couldn't that be implemented as a mailman list with *no* subscribers (subscription requests auto-bounce) I probably missed something? -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
Moshe Zadka writes:
Definately, but at that point, why disallow subscribers unless there's already a forum for the PEP? I don't object to setting up a mailing list if there isn't one that fits the bill, but I think Mailman is a fairly heavy way to do the job, and server resources definately seem scarce these days, so we don't want to overload what we have! (We'd probably also want to add checks that a message hasn't already been sent along, and deal with forwards a little differently -- we only want the forwarded message if someone is just archiving a message received in private email. Hmmm... this is starting to sound like an AI problem!) -Fred -- Fred L. Drake, Jr. <fdrake at acm.org> PythonLabs at Digital Creations
Moshe Zadka wrote:
Perhaps the PEP should specify this by e.g. using a header Comments-To: email-address The email address could then be made "usable" on the PEP web pages.
The original idea was for PEP authors to include the important parts of those discussions in summary in the PEP -- AFAIK, this is not done :-(
Hmm, wouldn't something like the SF bug manager be the ideal platform ? Perhaps someone could get the SF guys to add something like an "idea" manager with similar functionality but different content to SF ?! In the meantime, I think PEP discussions ought to taken to the sig mailing lists using python-dev as fallback solution. -- Marc-Andre Lemburg ______________________________________________________________________ Company: http://www.egenix.com/ Consulting: http://www.lemburg.com/ Python Pages: http://www.lemburg.com/python/
On Wed, 22 Nov 2000, M.-A. Lemburg wrote:
I'm trying to do this. However, summarising every e-mail sent to me about this PEP is quite a burden.
Yeah, that would work to.
In the meantime, I think PEP discussions ought to taken to the sig mailing lists using python-dev as fallback solution.
I think most of the PEPs are not in the charter of any new sig, and python-dev is invitation only forum. Perhaps having pep-xxx-sig would solve it too? -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
Moshe Zadka wrote:
I don't think people would sign up for each and every PEP... and you know how hard it is to even move a discussion from one forum to another -- this either doesn't work out or kills discussion. I think the simplest solution would be something like the SF bug manager. I'm pretty sure that other projects on SF have similar problems, so this might be worthwhile dicussing with the SF folks. The technology and user interface are there, it would only take a few new tables in the database. -- Marc-Andre Lemburg ______________________________________________________________________ Company: http://www.egenix.com/ Consulting: http://www.lemburg.com/ Python Pages: http://www.lemburg.com/python/
On Wed, 22 Nov 2000, Trent Mick wrote:
A Roundup mailing list with a category for each PEP?
OK, when we get over the make-Moshe-feel-stupid-for-not-proposing-roundup part, just let me know <wink>. Yep, Roundup is perfect for this! -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
"MZ" == Moshe Zadka <moshez@zadka.site.co.il> writes:
MZ> The PEP process, while *miles* better then anything I've seen in MZ> any other development process I've ever seen, has some (Not that we've used the PEP process for much of anything yet. Except, I guess, extended print, which I've grown quite fond of.) MZ> deficiencies. I'll try to point them out, and to suggest some MZ> partial solution: I don't like your partial solution, so I'll try to pick about the problems with the process <0.8 wink>. Put another way: I can't envision Slashdot as a useful design forum, but am interested in improving the PEP process. MZ> 1. Users are not sure who to post PEP questions/remarks to: MZ> Python dev? the original author? The Python devver they know MZ> best? In the absence of a reason to post elsewhere, the comments ought to be sent to the PEP author. If the message is intended to provoke wider discussion, then the user can post it on any relevent forum (and cc the PEP author). A Unicode PEP might be discussed on the i18n-sig; a Web PEP on the python-web-modules list; a change to core Python on python-dev. I don't think there will ever be a rule that says: "Comments on PEP 1812 must be posted to the pep-1812 web forum." Discussion should occur in the forum in which it is most relevent. Perhaps we could all send our comments to Andrew, and he could write a bi-weekly pep-comment summary <wink>. Is the problem that you don't know where to post comments on someone else's PEP, or that you are having trouble keeping track of discussions in multiple places? MZ> 2. It is the responsiblity of the PEP author to add open MZ> questions/pertinent remarks to the PEP. What is the deficiency with this approach? The PEP is supposed to present a coherent proposal and design for a new language feature. It is the designer's responsibility to write a good design document; several people can share responsibility for the design. Are you saying it's a problem that the PEPs aren't open to modification by anyone? (The last thing we need is someone who doesn't understand what a first-class function is messing with the scoping PEP <0.2 wink>.) The designer is responsible for discussing trade-offs and alternatives to justify her design. This is where the responsibility to address questions and comments comes from. MZ> 3. Mail about the PEP which contains important discussion is MZ> lost. How? I've got all the mail on the nested static scopes PEP. Is the problem just the personal burden of keeping track of lots of mail messages discussing a PEP? Any discussion that occurs on a mailing list will be preserved in the mailing list archive. That's not lost. So the only problem would be with discussion that occurs in private email, where everyone deletes copies of the mail. I wonder if the key problem is not having an authoritative mail archive for the PEP. One part of the Scheme RFI process that we did not copy was having a mail archive for each document. Perhaps this would be sufficient to address your concerns. If you, the PEP author, receive a comment by private email, you could bounce it to the archive. the-biggest-problem-with-the-PEP-process-is-no-one-writing-PEPs-ly y'rs, Jeremy
The key problem, as Jeremy noted, was that there is no archive of discussion about the PEP. I had a much more long winded and roundabout way of saying that. Also, a minor related problem is that when the only address on the PEP points to me (for example), I get all the mail about the PEP personally -- and I'm loath to bounce it to a public forum. *Having* a public forum which is *indicated in the PEP* would solve that. I could phrase that as a point-by-point answer to Jeremy (and originally did) but I decided that that would be idiotic. -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
Moshe Zadka writes:
Having a public forum isn't quite what's needed (though useful); much of the time, it's more important to have a public *archive*. I'm sure there's a way to set things up a mail handler that maintains an mbox file with collected mails on a topic, and a pipermail browsable version of that should be available, and each PEP should point to the appropriate archive. (It should also clearly state that relevant emails may be added to the archive at the recipient's discretion!) Using a mail handler for this would allow the handler to be added to the CC: on the emails and generally collect replies without any additional work on the part of the PEP author. The handler could be CC'd when the PEP is posted so that replies to the posting go there as well. Now we just need someone interested in writing a handler. ;-{ -Fred -- Fred L. Drake, Jr. <fdrake at acm.org> PythonLabs at Digital Creations
On Wed, 22 Nov 2000, Fred L. Drake, Jr. wrote:
Ummmmm....couldn't that be implemented as a mailman list with *no* subscribers (subscription requests auto-bounce) I probably missed something? -- Moshe Zadka <moshez@math.huji.ac.il> -- 95855124 http://advogato.org/person/moshez
Moshe Zadka writes:
Definately, but at that point, why disallow subscribers unless there's already a forum for the PEP? I don't object to setting up a mailing list if there isn't one that fits the bill, but I think Mailman is a fairly heavy way to do the job, and server resources definately seem scarce these days, so we don't want to overload what we have! (We'd probably also want to add checks that a message hasn't already been sent along, and deal with forwards a little differently -- we only want the forwarded message if someone is just archiving a message received in private email. Hmmm... this is starting to sound like an AI problem!) -Fred -- Fred L. Drake, Jr. <fdrake at acm.org> PythonLabs at Digital Creations
participants (8)
-
Andrew Kuchling -
Fred L. Drake, Jr. -
Fredrik Lundh -
Jeremy Hylton -
M.-A. Lemburg -
Moshe Zadka -
Moshe Zadka -
Trent Mick