Greg Hudson wrote:
>On Thu, 2004-05-13 at 21:00, Branko Čibej wrote:
>
>
>>No, it would make the code wrong IMHO because it would stop at the first
>>incompatibility instead of listing them all.
>>
>>
>
>Sorry, I missed that you were chaining the errors together. I continue
>to believe that it's a mistake to chain together unrelated errors
>(nested errors should have an inner-to-outer relationship, not a simple
>concatenation relationship), but we do so in other parts of the code
>already.
>
>
I've noticed this problem myself, and I think the long-term solution is
to have two-dimensional chaining -- i.e., each error can have both
siblings and children. I think that would cover all the possibilities.
>On the other hand, this error-generator callback strikes me as horribly
>over-engineered. We have no other functions which use callbacks to
>generate error messages. If the caller needs to add information to the
>error message generated by the library routine, it can do so using error
>chaining (of the more proper variety).
>
>
Yes, I think I agree... I'll remove that generator function. Simpler
life, etc. :-)
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri May 14 03:29:32 2004