Stefan Behnel wrote:

Paul Boddie, 01.07.2012 02:22:

...

Is there any reason why the compiler-sig mailing list wasn't chosen as a venue

Even I didn't know that this list even existed. And looking at the archives now, it's hard to see any relevant discussion in all the spam it received until it apparently died away in (almost) silence a couple of years ago.

Yes, although the mailing lists for special interest groups are advertised on python.org, there is no longer the focus on steering discussion to those lists. And I see that the compiler SIG is "retired", as is the related types SIG: http://www.python.org/community/sigs/retired/ I seem to remember various procedures about SIGs and their retirement, but I don't really recall much discussion of such things recently. Still, the compiler SIG matches the scope of the Google group pretty well: http://www.python.org/community/sigs/retired/compiler-sig/ There's even a link to discussion of some tools you may be familiar with.

...

It's obviously your choice where you host discussions and who you invite, and I know that the special interest group mailing lists aren't exactly well advertised these days

True, but many (most?) of them are simply not very well frequented, which reduces the interest in joining them even further. Both SIG mailing lists that I read only receive a mail every so many months, often enough without any reply. And almost all of these mails deal with questions that would better be discussed on python-list to leverage the substantially higher number of eyeballs there.

Special interest group lists were always meant to be used as focused channels of communication where people are actively trying to get stuff done. The unfortunate thing is that they aren't as well known as they were. Another unfortunate thing is that getting stuff done of mutual benefit is frequently something that takes second place to whatever other motivations and goals people have, for whatever reason, good or bad. Thus, traffic drops away as people either do other things entirely or instead promote any related work in other channels instead.

I think that's the basic problem: as long as more experts are lurking on python-list than on the dedicated SIG-ML, it's better not to use the SIG-ML for discussions but to go to python-list (or maybe python-ideas or python-dev) straight away.

I think we really have to sort out what python-dev is for, because currently there's a tendency to target the list when any kind of "expert" discussion is required, but there are a number of people who would rather see only CPython-related discussion here. Another matter is that static analysis of Python is a topic that frequently hits the end of the road when one cannot, by definition, analyze Python in its most dynamic form, and when people refuse to accept that such analysis has anything to do with Python in its most pure, undiluted (and most contrived) form. But as I wrote, I still intend to follow the newly created group and see what people have to say. Paul