On Sun, Jan 4, 2015 at 11:01 AM, Matthew Brett <[email protected]> wrote:
> On Sun, Jan 4, 2015 at 5:43 PM, Joachim Durchholz <[email protected]> > wrote: > > Am 04.01.2015 um 14:51 schrieb Matthew Brett: > >> > >> As far as I can see pip has had uninstall since version 0.6, about 5 > >> years ago. The stackoverflow post is someone trying to uninstall > >> using the wrong command line. > > > > > > LOL. That's what you get from just skimming a problem report. > > > > So, pip is in the clear about uninstalling. > > Are there other problems with it that would affect us? > > For sympy / mpmath, no, I don't think so. I'm happy to be corrected > if someone can think of something. > I think you're right. The proglems I know of with pip are - packages that have bad metadata in setup.py (something we can easily avoid) - combinations of packages that have some version restrictions. I doubt people will want anything but the latest mpmath. The only issue might be if mpmath breaks compatibility in some subtle way that breaks an old version of SymPy, and someone needs to use an old version of SymPy. In that case, they will need to know what version of mpmath to install in addition to SymPy. Conda does help here because in addition to having a true dependency solver, it lets you add dependency restrictions to old versions of packages retroactively. But honestly, with mpmath, it's a non-issue, and I doubt it will come up in any serious way. - packages that require compilation. This is pip's worst sore, because it compiles packages from source, unless there are wheels, and compiling from source on a user's machine is destined to fail. But SymPy is and will remain pure Python. This is more an issue with CSymPy, and again, you can work around it by building wheels. (by the way, in case you didn't know, I work on conda for my day job, so I'm very biased in that direction) Aaron Meurer > >> I suspect the new tool that gets really widespread adoption will first > >> need to persuade the Python Packaging Authority [3]. > > > > > > I'd be very happy if that's truly the case, because their policy sounds > very > > much like what we need (very careful not to break backwards compatibility > > etc.) > > Can we verify that the PPA is really authoritative? At least in those > ways > > that count, i.e. if what they decide is quickly and widely adopted, > that's > > good enough for me, authoritative or not. > > I think PPA is authoritative on general Python packaging. I think that > conda and so on are trying to establish themselves as de-facto > standards in scientific Python. > > The route that most projects I know have taken is to build things that > work with pip (like wheels for matplotlib etc) and let Continuum and > the conda team develop conda packages. > > Cheers, > > Matthew > > -- > You received this message because you are subscribed to the Google Groups > "sympy" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/sympy. > To view this discussion on the web visit > https://groups.google.com/d/msgid/sympy/CAH6Pt5p8b46L0ehP%2B_SGjnGem%3DAPR5qN11EeNLqygunWuP9EYA%40mail.gmail.com > . > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "sympy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/sympy. To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/CAKgW%3D6JERR9%2BFA_-O1Zmw%2Bh8N4bc95ek4Lh6hFzPdLbdbasgiw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
