On 2010/08/06 8:21 PM, Chris Sutcliffe wrote: > On 6 August 2010 20:31, Steven Monai wrote: >> On 2010/08/06 11:48 AM, Chris Sutcliffe wrote: >>> I've decided to take a different approach and decided to implement it >>> as follows: >>> >>> #ifdef __CYGWIN__ >>> itr->d_fileno = entry->d_ino; >>> itr->d_reclen = strlen(entry->d_name); >>> #else >>> itr->d_fileno = entry->d_fileno; >>> itr->d_reclen = entry->d_reclen; >>> #endif >>> >>> I assume this is appropriate? >> >> No, not according to this: >> >> http://cygwin.com/ml/cygwin/2009-09/msg00031.html >> >> Quoting cgf from there: >> >> "Defining d_*rec*len as strlen(d_name) would not be correct since that >> is supposed to be the length of the record not the name." > > Interesting. From an rtorrent perspective it's working as expected, > but as I previously stated, although rtorrent grabs the value, it > doesn't actually seem to do anything with it. > > I'll leave it as is for now (I figure having something there is better > than nothing at all).
IMHO, this is an unsafe approach. If, in a future version, upstream decides to start using itr->d_reclen for its intended purpose, then the plausible-but-incorrect value you've put there could become the source of Cygwin-specific bugs. In principle, if you can't/won't put the correct value there (which I see Corinna has helpfully posted), then you ought to put an obviously incorrect value there instead, such as 0 or -1. Perhaps the safest approach, though, would be to remove (or comment-out) the d_reclen field from the itr struct. Then, if upstream *does* use that field in the future, you should get build errors to alert you of that. -SM -- -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
