Le mardi 20 mars 2012 12:44:25, grischka a écrit : > Thomas Preud'homme wrote: > >> The problem with examples is that they can show bugs that exist, but > >> not really the absence of bugs. To test such feature I'd rather use > >> some printf here or there. > > > > And valgrind to test you don't free too early. > > The "free too early" bug wasn't really your fault but a mean trap > made of bad balance and lack of isolation, though announced by the > suspicious char* arg. In that sense it is better not to have such > code than to rely on valgrind.
The thing is I saw it was char * so I inspected the method but only followed options variable. I failed to see the assignment in p, although I was expecting this kind of code considering the char * (when the link_option function has a const char *). > > >> Anyway, the char * argument to tcc_set_linker is not really tolerable > >> if we want to offer it as LIBTCCAPI to the user. Also, instead of > >> tcc_set_debug, tcc_set_warning, etc. it would be good to have one more > >> generic 'tcc_set_option' API, of course with const char* argument(s) > >> then. > > > > This could/should we done in a separate commit, don't you think? > > That you volunteer to do it is more than I could expect. In any case > be careful with your precious (spring-)time ;) I didn't realize I was volunteering. Anyway, why not, just not right now though. I guess it will wait a few weeks. > > --- grischka Thomas Preud'homme
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
