[stdlib-sig] Breaking out the stdlib

[Antoine Pitrou]

Le mardi 15 septembre 2009 à 13:06 +0000, Collin Winter a écrit :
> 
> On the contrary: it's trivial to put numbers to "popular". Which
> number would you prefer:
[snip]

Well, the fact that there are so many different numbers to choose from
makes them collectively useless, unless you have a magic recipe which
yields an "objective" and irrefutable truth by combining them :)

My point, more generally, was that most python-core discussions don't
really bother with objective numbers. They try to conflate perceptions
and see what comes out. When Marc-André and others pointed out that
optparse and getopt were popular, noone tried to challenge them by
asking for numbers. Therefore, the view that these modules are popular
is accepted as truth within our small stdlib-sig community, even though
no objective proof was ever given.

IMO, most discussions about "outdated" modules would follow the same
path along the "popularity" issue: either someone points out that the
module is still popular and he is believed a priori, or nobody points it
out and chances are the module can be considered non-popular.