Matthieu Moy venit, vidit, dixit 26.04.2010 15:48: > Štěpán Němec <stepnem@xxxxxxxxx> writes: > >>> Hello, >>> >>> it seems there is no global .gitattribute equivalent to .gitignore or >>> .gitconfig. Wouldn't it make much sense to have one? Failing that, there >>> really should be a way to specify some global diff options at least. > > I'd like this feature too. My use-case is to have antiword and odt2txt > as textconv filters for .doc and .odt files, which I always want to > do. But probably the feature won't get implemented until one of us > writes a patch ;-) (I may have time later, but not before a few > days/weeks). I leave it up to you to do the (re)search, but there's some preexisting patches and discussion around this topic. The upshot is that currently: SYSTEM | GLOBAL | VERSIONED | LOCAL $(prefix)/etc/gitconfig | $(HOME)/.gitconfig | N/A | $(GIT_DIR)/config N/A | N/A | .gitattributes (rec.) | $GIT_DIR/info/attributes N/A | N/A | .gitignore (rec.) | $GIT_DIR/info/exclude There's also GIT_CONFIG overriding the defaults, and core.exludesfiles (yes, with s!); and of course command line parameters. There's been requests for versioned config, for multiple sources of system config (think config.d/*), for global and system versions of attributes and exclude/ignore, for saner naming of exclude(s) vs. ignore, and, finally, for stopping all this madness, or rather: putting all of it in a subdir, aptly named gitconfig... no-no, no-go... or following the XDG conventions, or... So, you see: You may suggest something simple like core.attributesfile. I have a prediction how that goes. But now, right after 1.7.1, it might be a good time to really settle this in a future-proof yet backwards-compatible way. Michael -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html