Hi, I noticed that routines dealing with possible memory exhaustion just return an error value. This means that, for an unrecoverable error, we must send the message all the way to the top, resulting in the library informing it's user that not enough memory is available.
Also notice that a change in a routine that suddenly inserts memory management would imply that it's interface must change (and probably it's caller's and so on). I think this not only places an excessive burden on the programmer, but makes the library difficult if not impossible to modify. Ideas? Regards, Gustavo.
