Hi Stephan,
My apologies for the long delay before the answer, but I'm just back at
home.
Stephan Bergmann a écrit :
<http://eric.bachard.free.fr/news/2008/06/wae4binfilter01.html>: "I think we can start to fix warnings in binfilter." When we started the
warning-free endeavor, we deliberately left binfilter "unfixed."
Sorry, what means "we" ? I was not aware something was decided
"officially" about binfilter module.
Do you have a link ?
The main motivation for making the code warning free was to offer a quality control tool for future code contributions/modifications (locating existing errors was only of subordinate interest). But as binfilter is
more-or-less dead code, it was deemed unrewarding to make that code warning free (a process with a non-trivial risk of introducing new
errors, by the way).
Well, there are several things :
First, the idea was pedagogical. Means, I think this is helpfull for the
Education Project (thus undirectly for OpenOffice.org Project), e.g.
other could say fix warnings is one simple way to attract people
interested to join, because :
- build OpenOffice.org is required (important prerequisite)
- create, apply/unapply patches is a daily task for all
- understand the code is mandatory
- discuss some warnings fixes on IRC is helpfull (and completmentary),
- this is good practice, and opens doors for other questions, ..etc
If I can, with the little experience I have, this is motivating:
students have not an important experience with OpenOffice.org source
code, and start slowly with a module like this one cannot cause too much
of trouble. After some experiences like that, they become familarized
with OpenOffice.org tree, and are less afraid with it's size.
Don't forget the main complain from the beginners: too complicated and
dark.
Continuying with the same goal, the idea to create cws is to show how
things happen for true in the project, with a limited risk.
Second, I must admit 3MB of warnings at every build is a bit boring me
(and I'm not alone), mainly when I know most of the problems do have a
trivial fix ( say at least 95%), I think it cannot be bad to fix them.
I never said fix all ( I discussed some with Philipp recently, and
obviously, some fixes will probably be tricky, or not possible), but
again, I'm pretty sure it won't hurt anybody to eliminate 95% of those
warnings.
Last but not least, Jan Holesovsky (alias Kendy) intelligently proposed
to create patches with one objective only. e.g. one patch for
uninitialized variables, one for bad Ctor initializations .. etc. More
easy to verify the fixes are ok.
I hope you get what I thought starting such task.
Regards,
Eric
--
Education Project:
http://wiki.services.openoffice.org/wiki/Education_Project
Blog : http://eric.bachard.free.fr/news
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
