Last call regarding Python 2 removal

[Fred Wright]

On Tue, 3 Nov 2026, Matt Selsky via devel wrote:

> I plan to merge https://gitlab.com/NTPsec/ntpsec/-/merge_requests/1510 
> after Saturday November 11 unless I hear any objections.

Is this meant for 2028?  That's the next time November 11 occurs on a 
Saturday. :-)

> Please speak up if there's any reason to further delay removing Python 2 
> support from our codebase.

I've previously spoken up against this.

Aside from impact on users, there's the matter of the polyXXX() wrappers. 
Some people mistakenly believe that those are there to provide Python 2 
compatibility.  They're not.  They're there to provide *Python 3* 
compatibility.  They're a kludge that Eric created to allow Python 2 code 
to work in Python 3 without having to be rewritten to conform to the 
Python 3 datatype model.  Eliminating them would require said rewriting to 
happen, along with ensuring that the new code is adequately tested.  Such 
rewrites do *not* render the code incompatible with Python 2, and in fact 
any errors in this rewriting are far more likely to break the Python 3 
case than the Python 2 case.  If that happens, using Python 2 is a 
workaround, provided that the option hasn't been removed.

Hence, people who think removing Python 2 is a prerequisite for removing 
the polyXXX() wrappers have it exactly backwards; the latter should happen 
*before* the former, lest it risk succumbing to "SCons disease", where 
Python 2 support was removed before all the Python 3 bugs were fixed, and 
no workaround was available.

Fred Wright