monotone-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Monotone-devel] Re: line endings with 0.31


From: Bruce Stephens
Subject: [Monotone-devel] Re: line endings with 0.31
Date: Tue, 21 Nov 2006 13:54:05 +0000
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.0.90 (gnu/linux)

Richard Levitte - VMS Whacker <address@hidden> writes:

[...]

> Oh, I dunno, but I'm imagining how amusing it might be to redo a
> large revision, with additions, drops and whatnot, just because one
> jpeg got modified.  Thing is, you don't always discover that
> immediately, so at the time it's discovered, that revision may have
> children and may have been synced to other servers.  I can imagine
> the joy of the CEO of the company that has a garbled logo on their
> web because of something like this.  I sure would be majorly pissed!

Sure, that may happen.  That's just a bug, though, isn't it, so you
commit a revision containing the changes to that one file (including
attribute change) as a child of the original, and merge it wherever
necessary?

> And the way things are currently, such things can't be corrected
> with a simple "fix up the setting".  And quite honestly, that didn't
> work with CVS either on other platforms than Unix, because CVS also
> used LF as internal line ending.  Forgetting -kb Windows when
> checking in a binary that happened to have embedded CRLF was never a
> good thing, and doing a 'cvs admin -kb' on the file afterwards
> didn't fix anything at all, just sets the flag, but the file itself
> is still screwed up.

Well no: we fix the setting, and then check the file, usually
committing a fixed version.

The whole thing's annoying, and I can see the attraction of what
you're suggesting.

I doubt it's too bad, though: if we can do binary detection well
enough, and can catch the common cases where people want odd EOL
settings (.dsp and .dsw, I guess?), and do that all by default (while
allowing people to override it if necessary), I suspect there'll be
few enough problems.

We've not found many problems when using CVS: sure there's the
occasional file that's binary but not detected as such (and .dsp files
that are text but must be stored as binary if you do weird things like
check out on Unix and build on Windows), but that's rare enough that
it really doesn't matter that much.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]