How to install Python package from source on Windows

Paul Moore p.f.moore at gmail.com
Tue May 30 14:00:47 EDT 2017


On 30 May 2017 at 16:56, Deborah Swanson <python at deborahswanson.net> wrote:
> I'm sorry people on the list somehow got the idea I was asking for help
> with this. Originally I only asked how to build recordclass from source,
> and apparently I made the mistake of chiming in with my experiences with
> Anaconda3, Python 3.4.3, on XP SP2, when the conversation turned to
> Python's use of Visual C++ and Visual Studio in the installation of new
> builds. I thought you might find my experiences interesting, but at no
> point did I ask for help that I'm aware of. If I did somehow give that
> impression it was a miscommunication with no intentional asking for
> help.

It certainly wasn't clear to me that once you'd got beyond the
installation of recordclass, you were not asking for help.

>From my perspective, as one of the authors of pip, you seemed to be
hitting a number of issues, all of which I would consider to be pretty
serious problems with the usability of pip. And I'd want to get to the
bottom of them so I could get them fixed for other users, as much as
for yourself.

1. That pip (or maybe Python?) attempts to install Visual Studio. If
true, this would be a pretty serious security issue, as you clearly
didn't intend to request an install of VS.
2. That upgrading pip via "python -m pip install -U pip" didn't work.
That command's already far messier than we would like, so we take any
further issues making it hard for people to keep up to date with pip
pretty seriously.
3. That you even thought that pip should want to install VS - which is
a problem with our documentation.

Obviously, you aren't required to do any diagnosis beyond that needed
to resolve your own issues, but the fact that you were still
participating in the thread seemed like you were willing to help - so
people asked questions to try to pin down the issue. Part of that is
trying to reproduce your problems, and identify what's particular
about your case, so understanding details of your system is important.
Things got a little out of hand at this point because it appears that
you got the impression that people were criticising your choice of
system. I don't think they were, but I didn't read the thread in full
(and I know that it's sometimes easy to see such criticism - as a
Windows user myself, I'm prone to over-reacting to "you wouldn't get
that with Linux" types of comment!) Nevertheless, I don't recall
seeing you ever saying that you'd resolved your issues to your
satisfaction, and you didn't think that the problems you'd mentioned
were going to happen to anyone else, as they were specific to your
setup. Maybe I missed it (it's a long thread and I did skim rather
than reading in detail) - apologies if that was the case.

> My strong suspicion is that the reason I'm actually seeing failed
> attempts to install Visual C++ and Visual Studio 2015 while none of you
> others are seeing any use of Visual C++/Studio is because it shows up in
> my tracebacks when it fails. In an XP SP3 or later Windows these minor
> installations would succeed and the user would never know they occurred,
> or perhaps they aren't even needed on later versions of Windows.

Here, you still seem to be asserting that Python is trying to install
something (VS). That simply isn't true. It may be that you're just
saying that *something* caused an install attempt, and aren't
interested in what it was. But for the sake of others finding this
thread, I'd ask you to confirm that you understand that we're telling
you that it isn't Python - or assist the rest of us to find out the
truth, because we'd be very concerned if a user's attempts to use
Python normally could cause attempted installs of *anything*. In
theory, that could be a pretty serious security issue. (I'm sure it
isn't, though - because I don't think that's what's happening).

> Unfortunately I didn't keep my tracebacks, having thought that they were
> no longer of any interest to the group at the time I had them, and I'm
> unwilling to destabilize my system and risk any further loss of
> functionality in any more attempts to reproduce them.

That's fine. But a simple statement to the effect that you have things
working as you want, and that it appears that there's no evidence that
Python was the cause of the VS installation attempts, and you don't
need to take things any further, would make things much clearer.

> I hope I am being perfectly clear now. But I certainly wasn't willing to
> go up against an angry mob intent on "helping" me or satisfying their
> curiosity at my my expense (or whatever they wanted), when I neither
> wanted nor needed (nor was aware of any explicit request for) any help.
> It is what it is, and I'm progressing just fine with what I've got.

Feelings certainly were running high, and hopefully they can calm down
now. But from my perspective, there's no "angry mob", simply people
who thought you wanted help, and were confused and frustrated by your
responses. An "it's OK I got things sorted, thanks for the help"
posting, followed by maybe a couple of replies along the lines of "no
idea, I don't have any of the evidence any more so I guess it wasn't
what I thought it was" to the inevitable "so what was the problem"
questions, should clear things up.

Paul



More information about the Python-list mailing list