meven added a comment.
In D25010#565421 <https://phabricator.kde.org/D25010#565421>, @dfaure wrote: > Hmm, I meant Name is always useful when *listing*. But when *stating*, we don't always need to get the name back. The NoDetails comment is about a stat that really can just succeed/fail. I added enum StatDetail { /// No field returned, useful to check if a file exists NoDetails = 0x0, Funny thing is that statx includes always the name, down to syscall we can save at most STATX_SIZE , STATX_TYPE fields in NoDetails case. I updated kdiroperator.cpp to use it line 748 since it only checks if a file exists. This code path is not run for local files anyway. > We can always add NoDetails later (with a value of 0), however, so I'll stop nitpicking here :-) I would not mind nitpicking/discussing a little more, this is easier to do now and since it is no small change. As long as we don't go off topic at least now I have a design standpoint validation. Another thing comes to mind If we allow not to fill KIO::UDSEntry::UDS_LINK_DEST but when KIO::ResolveSymlink is passed, we can save the second STAT by not passing AT_SYMLINK_NOFOLLOW to the first statx. So perhaps we want to expose a detail for this use case : that is add a IncludeLinkDest to KIO::StatDetail removing this field for StatDetail::Basic. I am not sure we have any use case currently though. INLINE COMMENTS > dfaure wrote in statjob.h:236 > missing the same #if as the above methods, no? > In a "clean" build it should disappear. It shares the same #if block with the stat function REPOSITORY R241 KIO BRANCH arcpatch-D25010 REVISION DETAIL https://phabricator.kde.org/D25010 To: meven, #frameworks, dfaure, kossebau Cc: kde-frameworks-devel, LeGast00n, GB_2, michaelh, ngraham, bruns
