Messages in this thread |  | | | Date | Mon, 1 Mar 2010 08:48:00 -0800 | | From | "Paul E. McKenney" <> | | Subject | Re: [PATCH RFC] kconfig: place git SHA1 in .config output if in git tree |
| |
On Mon, Mar 01, 2010 at 08:22:15AM -0800, Linus Torvalds wrote: > > > On Sun, 28 Feb 2010, Paul E. McKenney wrote: > > > > Uses "git log" to print the hash. > > Please don't use "git log" for something like this. > > Sure, it works, but it's kind of silly to say "I want a log, except I want > only a single entry, and btw, I don't actually even want the log entry for > that single entry at all, just the hash". > > It boils down to "I want a log, except with none of the log part". It > should make you go "Do I really want a log"? > > If you really want the hash, maybe just using "git rev-parse HEAD" would > do it. > > However, in this case I think _any_ of those would be wrong. Wouldn't it > make sense to use the same thing that we already compute for 'uname' > (scripts/setlocalversion)? Especially as that one already knows how to > handle other SCM's too (ie the whole hg/svn parts). > > That script also ends up using a nicer format, ie it uses "git describe" > to give a better idea of where it all is.
Agreed! The "-dirty" modifier for the case of changes not yet checked into git looks especially helpful. Geert Uytterhoeven already straightened me out on this one, and I posted an updated patch that uses scripts/setlocalversion (as a reply to his email).
Still learning about git, and I suspect that I always will be in that state. ;-)
Thanx, Paul
|  |