[pkg-horde] patch system for pkg-horde team

[Lionel Elie Mamane]

On Wed, Jul 11, 2007 at 03:48:13AM +0200, Gregory Colpart wrote:

> I think I'm right because I discovered patches in horde package
> with this method. Then "tla star-merge" doesn't fail but
> generates *.rej files. And commit is not possible while there are
> this *.rej files => patches must be adapted/removed.

Yes, and here are instructions on a nice way to treat them:

http://www.gnuarch.org/gnuarchwiki/Process_%2A.rej_files

Depending on the case, I either "apply the .rej file" by hand in the
source file, or I actually edit the .rej file itself so that it
applies and then apply the hunk with emacs.

For example, if our version of the source file says:

/* Here we do magic */
perform_magic()
/* Done! */

but the .rej file from a star-merge says:

 /* Here we do magick */
-perform_magic()
+perform_deeper_magic()
 /* Done! */

Then the reason the .rej fails to apply is that we corrected the
spelling mistake in the comment in our version with respect to the
upstream version, and then it can't find the same unchanged lines in
the context to find back the line it has to change. So what I do is I
change the .rej file to be:

 /* Here we do magic */
-perform_magic()
+perform_deeper_magic()
 /* Done! */

and I apply that hunk with emacs. Instead, I could have added the
"_deeper_" by hand directly in the source, but then I can make a
typing error in doing that (think of a more complex example than that,
and the risk becomes real), or I may miss subtle changes in the
source, like the string "cvs" becoming "csv" if the line also has more
obvious changes.

-- 
Lionel