[Unison-hackers] patches for 2.10.2 and 2.12.0

Jerome Vouillon Jerome.Vouillon at pps.jussieu.fr
Fri May 27 15:55:11 EDT 2005


On Mon, May 23, 2005 at 05:40:29PM -0400, Zvezdan Petkovic wrote:
> On Mon, May 23, 2005 at 03:44:11PM -0400, Zvezdan Petkovic wrote:
> > Also, what about the two patches for 2.10.2 that Andrew uses for Cygwin?
> > 
> > dot-unison.patch - make Unison use the Unix-like behavior in looking for 
> > the .unison directory (look first in $UNISON, then $HOME/.unison).
> > 
> > This one seems to be applied to 2.12.0.
> > I assume it is safe to patch 2.10.2 with it too, isn't it?
> 
> I lied. :-)
> After comparing _carefully_ the code in 2.12.0 and Andrews patch I see
> that his patch still doesn't give the preference to $UNISON environment
> variable although the README claims so.

Actually, Unison always gives the preference to $UNISON.  It searches
in the user's home directory only when $UNISON is not set.
So, Andrew's patch for 2.10.2 is correct, but the fix in 2.12.0 is
wrong (using $UNISON as a home dir should be a last resort option)...

> I think the patch should be as attached below (just picked from 2.12.0).
> 
> > text-mode.patch - fix three places where Unison opened files in text mode
> > instead of binary (http://groups.yahoo.com/group/unison-users/message/3184, 
> > http://article.gmane.org/gmane.network.unison.general/2406).
> > 
> > I don't know about this one.
> > Is this something Cygwin specific, or does it apply to all platforms?
> > Should I patch OpenBSD 2.10.2 stable with this?
> 
> Well, copy.ml part of this patch is quite different (simpler) in 2.12.0
> than in Andrew's patch.  The part in fingerprint.ml is identical, so I
> assume it is safe to apply.  The part in update.ml is again simpler in
> 2.12.0 than in Andrew's patch.  Should we roll a new patch using source
> from 2.12.0?

This is a Cygwin specific issue.

I'm going to make a better patch.

-- Jerome


More information about the Unison-hackers mailing list