> Feel free to do that. I've used up all my open source time for today, sorry. Would love to do that, but the list you are mentioning " virtualenv-users" doesn't seem to exist. At least google group only finds "virtualenv-api users" and it's invitation only
Jakub Bocheński On Thu, Sep 28, 2017 at 10:00 PM, Paul Moore <p.f.mo...@gmail.com> wrote: > On 28 September 2017 at 19:55, Jakub Bocheński <kuba.bochen...@gmail.com> > wrote: > > I think keeping the discussion in one place might save everybody's time. > > It's cool if you prefer mailing list to PR/issue thread but at least > linking > > the discussion there would go a long way. > > Feel free to do that. I've used up all my open source time for today, > sorry. > > >> Or would you have been happy if I'd > > added a one-word "noted" comment to the issue and left it at that? > > (Which would probably be about the same effort as many of the "me too" > > comments cost their authors). > > > > No. But if you quickly explained why see the value/effort ratio is low > here > > that would be fine. > > If I could have done that quickly, I would have. > > > Now: on the actual issue. > > This is not some random bash setting. Setting it on is bash best > practice -- > > don't take my word for it just google it. > > No need. See below. > > >> The use > > of the "${PS1:-}" construct (again, sorry if I got that wrong) may not > > be supported on all of those - so we risk breaking the scripts for > > some of our users in order to make them work for users who can easily > > enough switch off the undeclared variable setting. > > > > I understand the concern, but it's not as bad as you think. > > The ${PS1:-} construct is not a bash extension. It's in the POSIX > standard > > http://pubs.opengroup.org/onlinepubs/9699919799/ > utilities/V3_chap02.html#tag_18_06_02 > > Any POSIX-compatibile shell will accept this (which is basically any > shell > > in use). > > I can't comment on whether this is true. As I said, I'm not a Unix > user, and more specifically I have no feel for what's common Unix > practice (at my work, a major proportion of the servers I see use RHEL > 5, which as I understand it is ancient, and some use Solaris and AIX, > with shells whose vintage I don't know, but they *certainly* aren't > bash). Luckily for me, I don't need to use Python on those servers... > > Let me just be 100% clear here. I will not personally commit this > change. It is not in my area of expertise, and I'm not willing to be > browbeaten into doing so just because people keep telling me it's > fine. We have PyPA members who *are* Unix users, and whose judgement I > will trust on this. But they are very busy with other issues, so they > haven't had the time to look at this ticket. Sorry, but that's the > nature of volunteer-run open source. In the meantime, I've tried to > help give some perspective, by explaining the situation. I could have > just ignored the issue as not in my area of expertise (and indeed > that's what I did for some time). But I thought people might > appreciate at least getting a summary of the position. Maybe I was > wrong - you certainly don't seem pleased that I bothered. > > Repeating your comment from above: > > > No. But if you quickly explained why see the value/effort ratio is low > here > > that would be fine. > > Sure doesn't feel like you think it's "fine" that I've spent a number > of hours on this for you. > > I'm done with this issue. Sorry you're not happy. > Paul >