Re: Questions about svn_error_clear().

"Max Bowsher" <maxb@ukf.net> writes:

> kfogel@collab.net wrote:
>>
>> Of these four, I think I favor (3), and I'll implement it if people
>> agree.
>
> I really really don't like the idea of (3).

I don't like it much either.

> I think the idiom it destroys is useful and elegant.
>
> (2) is messy, and (4) is not so good, because error cases are much
> less exercised than success cases.

While (4) is not a solution I'd like to see it as part of
SVN_DEBUG_ERROR.

> Therefore, my vote is for (1). Obviously, that means I'm volunteering
> to do some of the work.

We have evidence that the bug occurred once, but it may not occur
anywhere else and I'd be surprised if it is widespread. I don't think
any such review is urgent (and the code paths covered by the testsuite
can mostly be checked by the combination of pool-debugging and
valgrind as the error will often be detected as a freed memory read).

-- 
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org