from
Stefan Rank <stefan.rank at ofai dot at>
to
Sylvain Thénault <sylvain.thenault at logilab dot fr>
subject
Re: [Python-projects] Re: bug report for pylint - line too long on
docstring
date
on 11.08.2005 12:14 Sylvain Thénault said the following:2005/08/11 12:55
> On Monday 08 August à 16:44, Stefan Rank wrote:
>
>>and of course... i did not attach the file ;-)
>>here it is
>>
>>on 08.08.2005 16:31 Stefan Rank said the following:
>>>hi, small bug report coming:
>>>
>>>the attached file produces a::
>>>
>>> test.py:2: [C] Line too long (130/100)
>>>
>>>for me.
>>>(100 is my line length limit, but the docstring in the module is
>>>split over several -shorter- lines)
>>>
>>>google knows about one message with a similar report::
>>>
>>>http://lists.logilab.org/pipermail/python-projects/2004-August/000138.html
>
> Hum, still unable to reproduce this on my linux box. I'll definitly have
> to find a windows box to test this :(.
>
i had a look at the source:
the culprit seems to be the (use of the) tokenize.generate_tokens
function. it returns the whole triple-quoted string as one token, but
where pylint.checkers.format.FormatChecker.process_tokens calls
self.new_line it assumes that the token only brought one new line with it...
i have no idea why this does not happen on Linux too (and i am too lazy
to reboot now ;-)
à propos: another cross-platform issue: i saw that FormatChecker uses
os.linesep in check_lines, which is not a good idea, firstly because not
all files have to use the platform's standard linesep (none of my python
files on win have a \r in them) and secondly because it gets the lines
from tokenize which probably uses 'universal newlines support' and
always returns \n - separated lines anyway.
cheers,
stefan
