Johannes Schindelin wrote Monday, January 12, 2009 12:45 PM
On Mon, 12 Jan 2009, Trevor Daniels wrote:
Valentin Villenave wrote Sunday, January 11, 2009 6:30 PM
> By the way, one sensible addition to the CG might be to invite Windows
> users to do
>
> git config global.autocrlf = false
>
> or something like that (by default under Windows, git converts all
> line endings to crlf)
I'm not sure this is advisable, or that the assertion is correct.
Git requires only LF endings in the repository, but Windows uses CRLF
natively.
If autocrlf = true, CRLF endings are converted to LF during commit and
LF endings are converted to CRLF during checkout. This seems eminently
sensible. This is the option I use, seemingly without problem.
The problem starts when at least one file was committed _with_ CR/LF.
Then Git fscks up.
Yes, that's why I think using Windows with autocrlf
false can cause problems. It's then impossible to
commit with CRLF endings.
ps BTW, Macs use just CR natively, which really screw up git for Windows
users if present in the repository, although easy to fix once
understood. Does autocrlf affect this in any way?
P.S.: as this applies only to MacOS < 10 (AKA X), we just don't care.
Well, I think I've had to correct at least 3 occurrences of
Mac CR line endings which were committed to Git, 2 of them
quite recently.
Trevor
_______________________________________________
lilypond-devel mailing list
lilypond-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/lilypond-devel
