On 10/30/2020 5:08 PM, Garrett D'Amore via Python-Dev wrote:
I’m not on this list. But I have offered to help - if there are tasks that need to be done to help this I can help put the weight of a commercial entity behind it whether that involves assigning our developers to work on this, helping pay for external developers to do so, or assisting with access to machine resources.
What’s more likely is that some group of developers aren’t interested in supporting stuff they don’t actively use.
[snip] I think it is more of a matter of pride in putting out a quality product and not making claims that might be false. Anyway, here is what I posted on the issue. https://bugs.python.org/issue42173 """ Dear Solaris Python fans (about 20 so far here): Here is the situation. There are 1000s of open issues on this tracker and at least 100s of open cpython PRs, and only 20-30 core developers, mostly volunteers, actively (depending on definition) merging PRs and maybe another 10 triagers helping to manage issues.
You all can help on issues by checking whether reported bugs still exist with current Python (3.8+) on current 'Solaris' (which includes what?). Searching open issues for 'Solaris' in 'All text' just returned 114 hits. About half were last touched over two years ago, and sometimes the last touch was inconsequential (a version or nosy change). I suspect many of the 114 are obsolete. For example, the last comment, five years ago, on #1471934, says the problem was fixed in Solaris 11.2. Does this mean the issue should be closed?
The bottleneck for merging PRs is reviewing PRs. We coredevs cannot do enough reviews ourselves. But any competent user can help. Reviewing has two components. First, does the patch fix the problem? Testing this requires a Github account and a local clone and ability to build a test binary. See devguide.python.org for more. Second, does the patch meet our standards of code quality. Solaris-specific patches likely change the C part of cpython, so C competence and understanding of PEP 7 is needed here. """
Of the 114 issues mentioned above, not all are Solaris specific, and at least one says not a problem on Solaris. Any at least some are obsolete. Anyway, 67 are marked as having a patch. Of those, about 25 have a github PR, the others should have an uploaded .patch or .diff that could be made into a PR.
Slicing another way, 20 and 15 are marked as type 'behavior' and 'compile error'. A few errors are possible. Another 11 have no type selected. So I guess there might be say 30 bug issues that affect Solaris and that have a patch to review and improve.
So one or more of your people could select issues of interest and help get a patch to where they think is it ready to merge. Then ask if some coredev that can do so will do a final review and possible merge.
In message msg380008 yesterday, Victor listed steps toward a buildbot, starting with a census of failing tests to discover the current state of CPython 3.10 on Solaris.