Re [Svnmerge] Noisy pychecker on svnmerge.py

[Blair Zajac]

Giovanni Bajo wrote:
> Madan U S <madan at collab.net> wrote:
> 
> 
>>>Last time I checked, many were bogus ones. For instance, I'm opposed to
>>>avoid the "is [not] None" paradigm.
>>
>>Well, now, I dont see a single "is [not] None" error from pychecker.
>>Anyways, why were you oppposed to it?
> 
> 
> None is interned in CPython. There's no reason to go through the ==
> semantic.
> 
> 
>>I will start working on the rest... will send multiple fixes in every
>>patch. I hope this is fine.
> 
> 
> What are the other issues? I don't like patches to shut up warnings caused
> by bugs/imperfections in external tools. If the tools help finding out real
> bugs, that's fine, let's fix them. But I'm -1 on any patch that tries to
> adjust code so to shut down non-issues and/or change our code to follow some
> coding convention (like the "is None" issue) the tools unilaterally decided
> it's the Good One(TM).

The Subversion teams takes care to remove all gcc -Wall warnings from the C 
source code and I don't see why that shouldn't extend to svnmerge.py.

Some of the warnings are bogus with the older version of pychecker and others 
pychecker warns about since we support older versions of Python.  However, I do 
feel that it's good practice to have as few warnings as possible, as when a new 
warning does appear, it may be an actual bug or something to address.

Regards,
Blair