On Tue, Jan 24, 2012 at 5:00 PM, Bart Trojanowski <bart [ at ] jukie [ dot ] ca> wrote: > On Tue, Jan 24, 2012 at 13:32, Aidan Van Dyk <aidan [ at ] highrise [ dot ] ca> wrote: >> >> > does that sound about right? and is there a single command that >> > would show all of that new content? thanks. >> >> I think the easiest way would be to pack everything before you start, >> and then after your commit, just look at all the loose objects made. > > > That's probably the best way. > >> >> Or if you want to work backwards from 1st principles, you can work >> with cat-file to show them from the commits through the trees/files >> how things (objects and their SHA1s) have changed... > > > The cat-file approach would be painful, since git doesn't track diffs, but > rather the state. There is no way to know which of the 37k files of the > linux kernel changed... at least not with cat-file alone. > > You probably want something like: git diff-tree -t HEAD > (which is semantically equivalent to git diff-tree -t HEAD~1 HEAD) > > You can then use the hashes from diff-tree output and send those to > cat-file. So with cat-file on theommit, you can show explicitly that the commit tracks different trees, and with diff-tree and ls-tree, you can show exactly what objects are different in the trees... And of course, since it's git, you don't actually need to give exact sha1's to diff-tree and ls-tree, anything "tree-ish" is good enough... a. -- Aidan Van Dyk Create like a god, aidan [ at ] highrise [ dot ] ca command like a king, http://www.highrise.ca/ work like a slave.