[Unison-hackers] Unicode names on Windows

[Gregg Tavares]

----- Original Message ----
> From: Gregg Tavares <unison at greggman.com>
> To: Unison hackers <unison-hackers at lists.seas.upenn.edu>
> Sent: Monday, June 16, 2008 10:26:26 PM
> Subject: Re: [Unison-hackers] Unicode names on Windows
> 
> ----- Original Message ----
> > From: Ralph Lehmann 
> > To: Unison hackers 
> > Sent: Monday, June 16, 2008 2:53:09 PM
> > Subject: Re: [Unison-hackers] Unicode names on Windows
> > 
> > Gregg Tavares schrieb:
> > >> From: Benjamin Pierce 
> > >> To: Unison hackers 
> > >> Sent: Sunday, June 15, 2008 8:48:53 AM
> > >> Subject: Re: [Unison-hackers] Unicode names on Windows
> > >> 
> > >> > I know that this discussion about Unicode characters in file name  
> > >> > is an old one, but I can't figure the problem is still alive ...
> > >> > So please, just tell me: Am I wrong with my usage of Unison  
> > >> > (2.27.57) on my computers here ? (France / China, Ubuntu / Windows  
> > >> > XP, and of course files Chinese characters in files ...)
> > > No, unison does not work with unicode file names. It's not unison's fault, 
> > > it's that it only runs on Windows, AFAIK, through the unix emulation 
> library,
> > > cygwin, and cygwin does not yet support unicode.
> > 
> > Please take a look at this site:
> > http://www.okisoft.co.jp/esc/utf8-cygwin/
> > 
> 
> I did take a look at that. Unfortunately the filename buffers in cygwin
> are too small for typical chinese/japanese/korean filenames. It may 
> work for short filenames but for a typical .mp3 filename like 
> 
> mp3folder/Band Name/Album Name/Track Name
> 
> it will fail
> 
> I tried recompiling that version of cygwin with larger filename
> buffers but it was beyond the time I had. The maximum filename
> size is not defined in one place. It could be that someone more
> familar with the code could get that patch to work easily but I
> couldn't. If anyone does I'd love to hear about it.
> 

I should add that if you manage to get the filename buffer size large
enough to work with Japanese/Chinese/Korean you'll also have to
re-compile ocaml so that it starts using larger buffers for filenames
and then recompile unison to use that new ocaml and that will
probably require recompiling every tool that ocaml needs to get
built.