On 11/24/2011 03:10 PM, Andreas Wehrmann wrote:
On 11/24/2011 03:01 PM, Christian Grothoff wrote:
Well, if creating the PP failed, why would you try to destroy it? ;-).
I don't think that any function has to work nicely when being called
with NULL (unless specified that it should accept NULL, of course).
Happy hacking,
Christian
That's easy: Because I continued to handle the request although creating
th PP failed...
Yes I know, my mistake :-)
But then:
Why do you check whether PP is not NULL in MHD_post_process()?
Same problem there.
It may be confusing to some users that some functions are "safe" and
some are not,
I think the library should be clear on this.
Consistency is a good argument, I'll add the check (removing checks
elsewhere is again a bad idea because technically it might break
existing, working code).
Happy hacking,
Christian
