On Mar 7, 2019, at 14:36, Mariatta
I was not involved in core Python development back then, so if it is really important and if people think such paragraph needs to be part of the PEP, then perhaps someone else more knowledgeable will need to help with this.
Personally, I don't think it was a horrible mistake. I believe the core devs back then carefully considered all options and decided that bpo/roundup was the way to go. And I really don't want to give that impression to the readers of this PEP that "I" or "core devs" now think it was a horrible mistake. If there is specific parts of the PEP that gives people that impression, then I'd definitely want to work and improve that.
I did a little bit of archive archeology (always a frightening and humbling black hole spelunking expedition), and here’s a brief history AFAICT. Dates are approximate. 5/2000 - we move all development (CVS at the time, and bug tracking) to SourceForge. This roughly coincided with PythonLabs leaving CNRI, so clearly we couldn’t continue running infra off of their systems. 10/2005 - we move to Subversion 9/2006 - we begin to discuss moving off of the SF bug tracker. I believe that Thomas Wouters, Martin von Loewis, Brett Cannon (big surprise! :), and myself were involved in that effort, with Richard Jones (original author of Roundup) recusing himself. The candidates were Roundup, Trac, Jira, and Launchpad. I think Brett did the first round of feature reviews and comparisons. David Goodger was also involved. We did want it to be written in Python and we preferred running it on python.org infra, but neither of these were required criteria. Jira and Roundup made the first cuts, with Launchpad and Trac being discarded as “having issues” (I don’t have access in memory or emails to any of those details). Jira was deemed pretty complex, but Atlassian offered hosting. Roundup was “not as polished" back then, but that wasn’t necessarily a negative. It was easy to use and host, and had a complimentary feature set, but we felt like we needed volunteers to help us keep it going. Richard Jones of course did fantastic work on the software itself, and we did manage to, um, round up enough volunteers to make it a viable choice. 10/2006 - the decision was made to move to Roundup, and we decided to accept Upfront’s offer to host the instance. 3/2007 - new-bugs-announce was created and notifications of new bugs was redirected to that mailing list. I’ll disappear down that archive rabbit hole now, which in some cases goes back to 1995. There are so many fun and scary paths to explore. See you in 6 months. jeremy-is-salty-ly y’rs, -Barry