[Python-projects] regression(?) between 0.9.0 and 0.10.0 (false positive)

[Sylvain Thénault]

On Saturday 03 June à 18:08, Tobias Klausmann wrote:
> Hi!

Hi,
 
> Consider the following program:
> 
> import sys
> 
> try:
>     for line in sys.stdin:
>         print "Got a line"
> 
>     print "The last line was:"
>     print line # culprit
> 
> except KeyboardInterrupt:
>     sys.stderr.write("Interrupted. Probably your fault.\n")
>     sys.exit(1
> 
> PyLint rightly complains (in 0.10.0+) that line might be
> undefined where it's marked "culprit".
> 
> However, if I add "line = None" before the try: or between try:
> and the head of the for-loop, thwarning does not go away, as it
> should. v0.9.0 does not complain at all (it doesn't check for
> this error, as it seems).

yep, 0.9 wasn't checking for this. 
I'll try to catch this until the next release (but sadly I've no
time at all for pylint now...) 
 
> PS: I'd also be happy about suggestions of fixing this in a
> different way that makes Pylint happy.

hum, there are probably some but nothing  at the top of my head right
now...

-- 
Sylvain Thénault                               LOGILAB, Paris (France)
Formations Python, Zope, Plone, Debian:  http://www.logilab.fr/formations
Développement logiciel sur mesure:       http://www.logilab.fr/services
Python et calcul scientifique:           http://www.logilab.fr/science