[Python-projects] Bug in pylint ? Looks like pylint 0.14.0 can't detect mismatch in function arguments!

[Pierre Rouleau]

On 9-Jul-08, at 7:40 AM, Sylvain Thénault wrote:

> On Wed, Jul 09, 2008 at 07:34:03AM -0400, Pierre Rouleau wrote:
>> Sylvain Thénault wrote:
>>> On Sat, Jul 05, 2008 at 05:21:20PM -0400, Pierre Rouleau wrote:
>>>
>>>> Hi.  I think I'v found a bug in Pylint 0.14.0
>>>>
>>>
>>> Hi Pierre!
>>>
>> Bonjour!
>>>
>>>> In the test file included below, Pylint 0.14.0 fails to report  
>>>> calls
>>>> with invalid number of arguments!  I tried it on a directory  
>>>> with no
>>>> rcfile just to be sure.  That's got to be a new bug, I'm sure I'v
>>>> seen  report of invalid number of arguments before with Pylint.
>>>>
>>>> At first I ran it with logilab common 0.29.1.  Then I noticed that
>>>> common 0.33.0 was available; I downloaded it and tried the test  
>>>> gain:
>>>> no  change.
>>>>
>>>> This is got to be a *major* bug.  Am I missing something here? I  
>>>> am a
>>>>  little tired, having worked several nights to complete  
>>>> something at
>>>> work.  But I was sure pylint was able to detect these mismatch
>>>> before...  Somehow, now, pylint 0.14.0 accepts invalid calls...
>>>>
>>>
>>> huum, I'm afraid that this has never been implemented in pylint  
>>> (yet).
>>> The only thing that has been done is checking of argument  
>>> consistency
>>> when defining/overriding *methods*. Function/methods calls aren't
>>> checked, though this could be done quite easily now that we have a
>>> better inference mecanism. The problem is that, as you may have
>>> noticed, i'm really missing some time to work on pylint those  
>>> days. For
>>> instance I've started a few months ago some work on astng to have it
>>> working using
>>> either the compiler module or the new _ast of python 2.5 but I've  
>>> not
>>> yet found the time to finalize it. Hope I'll get some time back in a
>>> near future for the pending stuff I've in mind for pylint. I'll  
>>> include
>>> function calls checking in this, I've created a ticket for this so I
>>> wont forget: http://www.logilab.org/ticket/5561.
>>>
>>> Thank you for the already written test file ;)
>>> Cheers,
>>>
>> Ok. Thanks.  I never realized there was a difference in treatment of
>> methods and functions in pylint.  I think it would indeed be a good
>> thing to be able to detect such discrepancies.
>
> I didn't meant that methods calls where checked while functions calls
> weren't. I juste meant that pylint is checking arguments when a method
> overrides another method, which doesn't make sense on functions.
>
Understood.  Hope you get some free time to work on pylint.

--

Pierre Rouleau