I got some lint/splint output off-list (thanks!) but couldn't
find anything with related to the problem.
I did find (and fixed) an unrelated error in the python version
that prevented it to work on the annonynized input files though.
Anyone other ideas on this problem or what can be done to track
it down?
Yours,
Emil
--On måndag, juli 21, 2008 22.59.47 +0200 Emil Pedersen
<emil.pedersen@xxxxxxxxx> wrote:
Hello list!
I'll try to keep this short, if any more info is needed just ask.
I have a program that is supposed to parse standard log files
and act on specific lines.
It mostly does what it's supposed to, but if I run it a couple
of time, using the same input data, some of the results differ.
If someone could find out what is wrong with them I would be most
greatefull. Due to the length (c version is ~230 lines, python
version ~130) I didn't know if it was ok to post them, so I've put
them and the test data on
http://83.253.255.25/tmp/
[Some more background]
I wrote a python version with psycopg2 as "glue" to postgres and
this is where I noticed the behavior. Since I couldn't spot the
error in my program but did find a bug ubuntu reports regarding
psycopg2 I suspected the debian version might be affected to and
asked on debian-python[1].
While trying to find out what was wrong I re-wrote the program in
c/libpq, and it still behaves the same way so the fault has to be
in my program.
I don't know if I am using libpq in some wrong way (and I know
that the program is a mess), but I would expect the program to
produce the same result every time given its simplicity.
I have tested on two different computers, one debian/etch i386
and one debian/etch amd64. I've also tested with two different
versions os psycopg2 while trying to rule it out or prove it
broken.
[1]
http://teams.debian.net/lurker/message/20080717.145350.2ba26912.en.ht
ml
Sincerely yours,
Emil
