Re: [Monotone-devel] monotone win32 crlf questions

[Magnus Therning]

On Thu, Jul 01, 2004 at 09:34:00PM -0600, Derek Scherger wrote:
>Jon Bright wrote:
>>The differences in normalisation are actually the reason for about half 
>>the self-tests failing on Windows.  This is something I've been meaning 
>>to look into, but haven't got to yet...

>I guess my initial expectation on this was that it would do something
>like cvs (or is it wincvs?) does... if you're checking in from a
>windows client to a unix server I think it converts the line endings to
>\n and when checking out does the reverse translation. However I'm
>wondering if this is even possible given that hashes presumably need to
>agree when/if a file has changed and introducing line ending conversion
>would appear to mess this up. I'm probably just missing something here
>though, and hopefully this is all accounted for. It's certainly a nice
>thing to have when people are working on various different platforms.

Wouldn't it be possible if hashes always are calculated on "normalised"
files? I.e. on some platforms the files will be processed before hashes
are calculated.

And of course I don't _really_ know what I am talking about :-)

/M

-- 
Magnus Therning                    (OpenPGP: 0xAB4DFBA4)
address@hidden
http://magnus.therning.org/

It doesn't matter who you are. Most of the smartest people work for
somebody else.
     -- Bill Joy

signature.asc
Description: Digital signature