[Catalog-sig] PEP 345 Update

[P.J. Eby]

At 10:16 AM 8/23/2010 -0700, Tarek Ziadé wrote:
>Well the only nuance is that --depending on the trust you have in the
>project of course--
>You know that you will get Bar, if you install Foo which Obsoletes it,
>and that you don't need
>to install Bar anymore. e.g. It just become redundant.

Right - this is the place where things potentially get tricky.  If 
Foo obsoletes Bar, and Baz depends on Bar, then a tool shouldn't 
silently substitute Foo without user intervention (but it could 
certainly store the user's preference so as not to need that 
preference to be repeatedly expressed).

However, even if it is doing the substitution per the user's saved 
preference, it should still *mention* (at least during Baz's 
installation) that it's substituting Foo for Bar to satisfy Baz's request.

Otherwise, if Baz has only been tested with Bar, the author of Baz 
may get complaints that actually fall to regressions or changes in 
Foo, wasting everybody's time on debugging.  If the user is 
warned/reminded of the substitution, there is at least a *chance* 
they'll remember to tell Baz's author about it.  ;-)

(Obviously, this is all suggestion/recommendation rather than fiat; 
an installer author is of course going to do whatever they can or 
whatever they want.  It's just a matter of clarifying intent.)