UTF-8 and filenames

Simo Sorce ssorce at redhat.com
Wed Mar 14 22:55:58 UTC 2007


Replying to myself :(

On Wed, 2007-03-14 at 18:51 -0400, Simo Sorce wrote:
> On Wed, 2007-03-14 at 17:44 -0500, Callum Lerwick wrote:
> > Okay, so people missed my point.
> > 
> > The point is, if its a tool that doesn't *have* to care about encodings,
> > then it should not care about encodings. If its not something that
> > performs stuff like alphabetical sorting or case insensitive comparison,
> > then it should just accept and pass on the filename as a bag of bits,
> > not caring what they mean.
> > 
> > For example, for a long time rsync wouldn't allow you to specify files
> > and directories with spaces in the name on the command line, no matter
> > how hard you tried to escape or quote it. It would interpret the space
> > as a filename seperator. This finally got fixed at some point though. :)

Except that rsync may run on windows or mac OS X as well, and this
reintroduce the need for rsync to be able to understand what are the
bits so that it can translate names to UTF-16 ...

> > This also applies to people not properly quoting variables in shell
> > scripts... :(
> 
> This is much more clearer and I agree 101%

I still agree, it is just that the number of apps that can really avoid
caring are less than what most people may think without careful
consideration.

Simo.




More information about the Fedora-maintainers mailing list