> Is it really a performance problem? Avoiding this kind of pointer/buffer
> size juggling is one of the reasons for using strlcpy/strlcat.
>
> Without this, you have a quadratic behaviour for a function working with
paths, which I consider a really bad thing.
Also I don't see the link between strlcat and point juggling. My only point
was that if you know a string S already is N character long, you can strlcat
at S+N and avoid rescanning it and without any risks because if your buffer
size of size M then S+N is of size M-N.

As for the -1, I didn't pay attention but it seems to me that it is an
unrelated problem.
Received on 2011-06-21