On Thu, Jan 27, 2011 at 04:18:15PM -0800, patrick keshishian wrote: > On Thu, Jan 27, 2011 at 3:58 PM, Federico G. Schwindt <[email protected]> > wrote: > > On Thu, Jan 27, 2011 at 11:12:43AM -0800, patrick keshishian wrote: > >> On Thu, Jan 27, 2011 at 1:37 AM, Federico G. Schwindt <[email protected]> > >> wrote: > >> > On Wed, Jan 26, 2011 at 05:13:44PM -0800, patrick keshishian wrote: > >> >> On Wed, Jan 26, 2011 at 4:45 PM, Federico G. Schwindt > >> >> <[email protected]> wrote: > >> >> > On Wed, Jan 26, 2011 at 03:15:11PM -0800, patrick keshishian wrote: > >> >> >> http://marc.info/?l=openbsd-ports&m=129372769129646&w=2 > >> >> >> > >> >> >> I never got any replies of "yes", "maybe", "you are out of your > >> >> >> freaking mind", etc. > >> >> > > >> >> > ?what about something like this: > >> >> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=300289? > >> >> > ?looks simpler imho. > >> >> > >> >> The libhfsp/src/swab.h source on debian link seems different than what > >> >> I see in hfsplus_1.0.4.src.tar.bz2 (just downloaded). > >> >> > >> >> MD5 (hfsplus_1.0.4.src.tar.bz2) = 18fa1efb5432469357ffa6bfa7c08fcd > >> >> size: 185461 > >> >> > >> >> > >> >> Either way, I don't think the dabian patch is correct; at least not > >> >> entirely. They propose: > >> >> > >> >> -#define bswabU16_inc(ptr) (*((UInt16*) (ptr))++) > >> >> +#define bswabU16_inc(ptr) ptr = (UInt16*)ptr + 1 > >> >> > >> >> which will change the following (from btree.c): > >> >> > >> >> ? ? node->num_rec ? = bswabU16_inc(p); > >> >> ? ? node->reserved ?= bswabU16_inc(p); > >> >> > >> >> to: > >> >> ? ? node->num_rec ? = p = (UInt16*)p + 1; > >> >> ? ? node->reserved ?= p = (UInt16*)p +1; > >> >> > >> >> I seriously doubt this is the intent of the original code. > >> > > >> > ?ok, assuming they wanted "*p; *p++" i believe you can do: > >> > >> No. The intent is: > >> > >> ? ? ? some_integer_type = *(some_integer_type*)p; > >> ? ? ? p = (some_interger_type*)p + sizeof(some_integer_type); > > > > ?sorry, that's what i meant (as opposed to p = p + 1) and the code below > > should do that. > > Nope. What your macro does is "*p; (*p)++".
ops, you're right. i've commited a version that hopefully is correct and should unbreak it for good. there are others issues that i've found while looking at this port but they'll be addressed post lock. f.-
