Splitting up large python module impact on performance?

[Rick Johnson]

On Wednesday, June 13, 2018 at 4:27:35 AM UTC-5, Chris Angelico wrote:
> It's more his definition of "large" and "small" that I was
> disagreeing with. You're absolutely right that a dense
> global scope is a problem; but a "one class per file" rule
> is a terrible idea.

What if the "one class" spans thousands of lines? Surely you
don't believe that class definitions (outside of toy classes
and academic examples, that is) are limited to a couple
dozen lines, do you?

> A hundred tiny files is far harder to work with than ten
> medium-sized files,

I doubt anyone here is suggesting that modules should be
limited to 10 lines each, or something ridiculous like that.
Heck, the minimum class definition _alone_ (at least, if
you're going to properly utilize vertical whitespace)
requires around 2 dozen lines.

    ## BEGIN MODULE ##
    01#
    02# import something here
    03#
    04# define constants here
    05#
    06#
    07# (WHITE SPACE BUFFER)
    08#
    09#
    10class MyClass(object):
    11    def __init__(self):
    12        pass
    13
    14    def method(self):
    15        pass
    16#
    17#
    18# (WHITE SPACE BUFFER)
    19#
    20#
    21# (INITIALIZATION/TESTING/LOGIC/ETC)
    22#
    23#
    ## END MODULE ##

That's 23 lines of code to define whitespace and boilerplate
_alone_.

> and IMO a single file with all the code in it is only
> slightly worse. That is to say, I would prefer to work with
> a single gigantic file than a directory with lots and lots
> of tiny interdependent files, each one importing six or
> seven others.

Blame the import mess on a missing feature, not on those of
us who prefer reasonably sized source files.

To alleviate a large portion of the circular import mess
(and depending on the coding style employed, possibly all of
it), python's packaging system could have leveraged the
power of "componentized modules" -- much the way that
classes share variables between them using class attributes
(no import required!) -- but unfortunately, this powerful
feature was overlooked. And packages are nothing more than
some kind of quasi directory tree which masquerades itself
as a namespace so that a programmer can replace a long
import path with a _dot_.

WELL, WHOOP-DEE-DO!

That's like a superhero without a superpower.

Hmm...

After considering the blinkered state of Python packages and
the resulting pitfalls of the python import mechanism, we
can only assume the Python gods found inspiration for this
"hot mess" in the title of GnR's fifth studio album.