> -----Original Message----- > From: Bert Huijben [mailto:[email protected]] > Sent: vrijdag 5 oktober 2012 15:30 > To: Michael Pilato; 'Ben Reser' > Cc: [email protected] > Subject: RE: svn commit: r1394332 - > /subversion/trunk/subversion/include/svn_client.h > > > > > -----Original Message----- > > From: C. Michael Pilato [mailto:[email protected]] > > Sent: vrijdag 5 oktober 2012 14:49 > > To: Ben Reser > > Cc: [email protected] > > Subject: Re: svn commit: r1394332 - > > /subversion/trunk/subversion/include/svn_client.h > > > > On 10/04/2012 10:06 PM, Ben Reser wrote: > > > On Thu, Oct 4, 2012 at 6:48 PM, C. Michael Pilato <[email protected]> > > wrote: > > >> On 10/04/2012 09:46 PM, C. Michael Pilato wrote: > > >>> Perhaps you meant something like: > > >>> > > >>> "... it will enter versioned directories, scheduling any unversioned > > >>> children thereof for addition." > > >> > > >> Sorry -- I just saw that you fixed the *unversioned* bit. My additional > > >> questions remain: > > >> > > >>> But why only #svn_depth_infinity? Will it not do the same (to > different > > >>> depths, of course) for #svn_depth_files and > #svn_depth_immediates? > > > > > > How about: > > > > > > [[[ > > > When used with @a depth it will enter versioned directories (per the > > > rules of the argument), and schedule unversioned children. > > > ]]] > > > > > > > Honestly, the original phrasing of the docstring remains a better starting > > point, in my opinion. Your changes lose the context that all this > > discussion about depth and unversioned items in a versioned tree are still > > tried primarily to the use of the force flag. So if it were up to me, I > > would restore that paragraph to the state it was in and make only minor > > changes: > > > > * If @a force is not set and @a path is already under version > > * control, return the error #SVN_ERR_ENTRY_EXISTS. If @a force is > > * set, do not error on already-versioned items. When used on a > > * directory in conjunction with a @a depth value greater than > > * #svn_depth_empty, this has the effect of scheduling for addition > > * any unversioned files and directories scattered within even a > > * versioned tree (up to @a depth). > > Currently this function fails with an error when called on the working copy > root, with and without force set to TRUE. > (Probably because we originally handled force to suppress some light errors, > while we can't add a working copy root to its parent) > > Maybe we should fix this as well if we are touching this code anyway?
http://stackoverflow.com/questions/11659867/is-there-a-directory-bug-in-svnclient-add made me remember this problem (Reported against SharpSvn, but the error originates in libsvn_client/libsvn_wc) > > Bert
