How to install Python package from source on Windows

[Deborah Swanson]

Chris Angelico wrote, on Monday, May 15, 2017 1:02 AM
> 
> On Mon, May 15, 2017 at 4:48 PM, Deborah Swanson 
> <python at deborahswanson.net> wrote:
> > Chris Angelico wrote on Sunday, May 14, 2017 11:30 PM
> >>
> >> On Mon, May 15, 2017 at 4:28 PM, Deborah Swanson 
> >> <python at deborahswanson.net> wrote:
> >> >
> >> > Again, maybe I should go back to Python 2 for this.
> >>
> >> Won't help. The same problems will exist.
> >>
> >> ChrisA
> >>
> >
> > Good to know.  Maybe I just won't be able to try 
> recordclass until I 
> > get Linux set up one way or another. Right now it's more 
> important to 
> > me to spend the few hours a day I have making progress in Python.
> >
> > (Now, if I can figure out how to make gettattr work in this one 
> > namedtuple problem, I won't need anything like recordclass. I've 
> > converted three more Excel spreadsheets to Python, but it's tricky 
> > working with namedtuples when you want to group rows or 
> take n colomns 
> > at a time. I've done it, but this one is really tough.)
> 
> Hmm. If you're working heavily with tabular data, you may 
> want to consider numpy or pandas, both of which are designed 
> for major computational work and will happily manipulate rows 
> or columns of data. Both are well supported on Windows (and 
> everywhere). They do have their own learning curve though.
> 
> ChrisA
> -- 

Yes. I want to spend more time with numpy and learn pandas, but right
now I do have to consider the learning curves involved. Maybe in 4-5
months I'll pull out of this wretched state I'm in now, and I'll spread
my wings in dozens of directions then - for about a year, don't ask me
how I know. (I haven't calculated out beyond 2017 yet.)

But there's another issue too. Contrary to popular belief, just because
spreadsheets are what you build on a rectangular grid of cells, not many
advanced spreadsheet applications are straight-up tabular.  Most
beginners start out tabular (though I didn't, a couple decades ago), and
it seems that most never get out of the tabular framework.

I've been finding that namedtuples covers a good 90-95% of my "not
exactly tabular" (and "not at all tabular") spreadsheet applications,
and I'm quite curious to keep extending my knowledge about what you can
do with namedtuples. And it probably would be surprising to some how
much that is. It's mainly a few awkward corner cases, like grouping rows
and taking columns n at a time that can really be tricky. (But I've done
both now.)

Eventually I plan to know several tools to convert spreadsheets to
Python with, including numpy and pandas. But considering the "tabular"
discipline both of those have, I doubt I'd be saving myself any grief by
switching to them. 

Learning curves are a bitch in their own right, and I can think of worse
ways to spend a few broken months than getting to know one way really
well. It's my fallback position when things get bad, to just stick with
what I don't have to learn too much about and find out what I can do
with it. You'd be surprised!

Deborah