On 15 August 2010 22:11, Charles Wilson wrote: > On 1/5/2010 11:48 AM, Charles Wilson wrote: >> Personally, I dislike global static variables. It's also bad practice >> with popt, because a single program can have multiple parsing contexts. >> (Not that any of the cygutils tools do this; so if you want to globally >> modify all cygutils popt-using utilities to make optCon a static var, I >> would accept that patch; > > I've made this change throughout. > >> I think that, in general, --help output should go to stdout, not stderr. >> Also --usage, --license, and --version output. However, if an *error* >> is handled by printing the usage(), then it should go to stderr. >> >> That wasn't the case in mkshortcut. So, rather than removing the FILE * >> argument -- since, as you noticed, all of the affected functions were >> invariably called with stderr -- I'd prefer that the ?,u,v,l cases be >> updated to call the appropriate function with stdout, but that existing >> error handling cases call usage() with stderr. (Ick. It appears that >> most of the other popt-using cygutils tools ALSO send everything always >> to stderr. Well, this is an easy change. Let's punt on this, too, and >> I'll do it all-at-once) > > This too.
Thanks very much! > So, if you want to forward port your locale/charset changes and > cleanup-handling changes for mkshortcut (probably as two separate > patches) I'll try to act on them more rapidly this time. :-) So you'd be ok with not freeing stuff before exit()? That would make things quite a bit easier. (Belated apologies for being too strident about this back then.) Andy
