[python-win32] python-win32 Digest, Vol 205, Issue 5

[Tim Roberts]

On May 24, 2020, at 9:21 AM, Markus Kramer <markuskramerigitt at gmail.com> wrote:
> 
> The points are:
> - Reduce vulnerability. The MFC library is large and (earlier? versions) contained CVE's. 

Whatever vulnerabilities there might be aren’t a factor if you aren’t using them.


> - Reduce maintenance cost. By shipping less, you need to observe less, and patch less.

You are missing the point.  Pywin32 simply makes all those thing available to you.  You don’t have to use them.  And if you aren’t using the graphical interfaces, then you aren’t shipping them.  The Python installer apps only include the things your script needs.  If you don’t need graphics, then graphics DLLs won’t be included in your package.

And your proposal will, of course, INCREASE the maintenance cost for the Pywin32 programmers, who now have to deal with multiple packages.


> - Reduce resources. Less size mean less disk space and less network bandwidth.

The MFC DLLs are about 100MB.  At today’s mass storage prices, that’s 3/10 of one American penny.  Network bandwidth will not be an issue for your applications, because MFC won’t be included in your application.


> Then you wrote
> > If you ship a package that doesn’t need the UI components, then your package will not include the components.
> 
> This is my goal. 
> I don't get what you refer to with "package". 
> Currently, `pip install pywin32` will include the UI components, even if they are not needed.

When I say “package”, I mean an application package that you want to distribute, created by something like pyinstaller.  It will only include the components your application needs.

I just don’t see that there is anything to gain in making such a substantial change.
— 
Tim Roberts, timr at probo.com
Providenza & Boekelheide, Inc.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/python-win32/attachments/20200524/3a7255b9/attachment.html>