Mailinglist Archive: zypp-devel (57 mails)

[Klaus Kaempf <kkaempf@xxxxxxx>]

* Michael Schroeder <mls@xxxxxxx> [Jan 08. 2008 18:07]:
> No, if an assertion fails it normally means data corruption. It
> really makes no sense to return something in that case.
> Even glibc aborts in that case.

Well ...

I partly agree and I partly disagree ;-)

My current problem with sat-solver, and why I stumbled over the
aborts, is reusablitity of the 'Solver' structure. It seems that it
cannot be used multiple times because some internal queues aren't
properly re-initialized.
The current way of simply aborting without giving a reason or even a
line number is just not acceptable (imho). (The actual error also
shows missing test coverage for the case in question.)

I do agree that in case of data corruption, 'drastic means' have to be
taken. But please not without some 'last words' ;-)

Klaus
-- 
To unsubscribe, e-mail: zypp-devel+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: zypp-devel+help@xxxxxxxxxxxx