These patches might be slightly controversial. Since there is no perfect solution, we may want to try something that works to some extent and gives what a user might expect, i.e. presenting to a viewer the same glyphs that the user who entered them saw, to the extent it is possible. We already handle commit messages like this for the old style objects (sort of). This patch set also affects other data like refs. Currenly no sane solution exists in git so nothing really works well outside the non-ascii range for refs anyway so we can discuss what should happen with refs that contain non-ascii characters. The best thing is to avoid them, but some of us live in countries with funny dots in what we do and other have even stranger ways of expressing what they do, and hence things like branch names etc. Legacy SCM to GIT conversion programs seem to do every variation of transcoding/ not transcoding commit messages and file names to UTF-8 so there is an issue here. The nice thing about transcoding filenames to UTF-8 is that they work on all platforms. A non-UTF-8 filename in a UTF-8 environement doesn't. In particular such filenames are more or less inaccessible to a Java programs. For the reverse case it looks really bad. C Git currently does not transform file names. Missing from this patch set is test cases. As it is quite undefined in git what happens that is sort of ok so far, but I'd like to define it too in the same way. -- robin -- 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
