Completion of error handling

[Markus Elfring]

Pino Toscano schrieb:
> I'd prefer promoting further improvements in KDE software,

I am trying this for various issues, too.


> given that there are many things to do, so caring also about 3rd parties resources
> that we  don't use (ie more than the ones we use for one reason or another)
> is simply not doable, given also the (not so extended) resources KDE has.

The requested changes need just a bit more work. I hope that coding efforts can
be saved after corresponding adjustments.
Other tools like approaches for static source code analysis can also help to
identify update candidates.


> And I would like a bit more of common sense, especially for simple cases.

I would be more careful with the use of such an ambiguous and fuzzy term where
software developers prefer strict programming rules.


> It does not make sense using aspect orientente programming just to check the 
> return value of a single function, especially when there are so few cases in 
> the code. This is simply overdesign.

Return code checking is a cross-cutting concern like several other design
patterns that are scattered around in the source code base.
http://en.wikipedia.org/wiki/Cross-cutting_concern

The main point is to specify in a single source file which advices should be
applied to a lot of function calls. The traditional attempt to cope with this
situation is to create wrapper libraries.


> Do you agree on that?

No - I hope that more participants can be convinced of the benefits and the
applicability in multidimensional software design.

I see it as a move from manual application of patterns to an automated weaving
step during the build process.

Regards,
Markus
 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<