German umlauts in file names

Matt Mackall mpm at selenic.com
Thu Jun 19 12:02:42 CDT 2008


On Thu, 2008-06-19 at 14:13 +0200, Jens.Wulf at sew-eurodrive.de wrote:
> Though I can't help directly: to me it is unclear what your setup is.
> You work on both linux and windows, but it is not clear on which
> system your example command lines are issued and what the history of
> the repo in use is (cloned from other system? which one? etc...)
> 
> Furthermore I wonder how hg should fix a certain problem at all: if
> you create utf8 filenames containing umlauts on your linux box and
> then clone to a windows box (which uses some 8-bit encoding), the repo
> will still be encoded as utf8 (filenames AND content). hg cannot
> convert filenames and content; the mapping between utf8 and some 8-bit
> character set is not...well, don't know the word, it is 'eineindeutig'
> in german, but means you can't easily convert one to the other and
> back without probably losing or changing data.

The word is 'one-to-one'. And indeed, if we write out files to the local
working directory during conversion, then our local encoding must allow
storing all the relevant characters (on Windows). On most UNIX
filesystems, the filesystem is 8-bit clean and will accept names in any
encoding, so there's no such problem.

Patrick has lately been working on making convert bypass the working
directory, at least for the hg sink.

-- 
Mathematics is the supreme nostalgia of our time.



More information about the Mercurial mailing list