10:56 a.m.
Just in case someone else is bitten by this: I've just discovered an
unfortunate effect of the new Mercurial-based layout for the packages:
doing an "hg diff" from the packages root directory doesn't work
anymore[1]. You need to cd to the package directory before doing the
diff.
As a consequence, it's not possible anymore to define a single Patcher
project for all the packages and use the sub-project facility for
restricting the diff/commit process.
If you try it, you get an empty diff and Patcher then complains that the
sources don't differ from the archive...
Footnotes:
[1] in fact, it's worse: it just stays silent and returns nothing.
--
Resistance is futile. You will be jazzimilated.
Scientific site: http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
3:24 p.m.
Didier Verna writes:
Just in case someone else is bitten by this: I've just
discovered an
unfortunate effect of the new Mercurial-based layout for the packages:
doing an "hg diff" from the packages root directory doesn't work
anymore[1].
How about "hg diff -S"?
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4:53 p.m.
Stephen J. Turnbull wrote:
How about "hg diff -S"?
Oh, cool. I didn't know that one.
--
Resistance is futile. You will be jazzimilated.
Scientific site: http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4:57 p.m.
I wrote:
Stephen J. Turnbull wrote:
> How about "hg diff -S"?
Oh, cool. I didn't know that one.
Hmmm, but now, hg diff -S shows me some differences in the mmm-mode
subrepo. These are changes that I pushed directly from there, and in
this subdirectory, the repo is up to date.
So the question is: when I push something from a subrepo, do I also need
to do something at the root level ?
--
Resistance is futile. You will be jazzimilated.
Scientific site: http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
2:07 a.m.
Didier Verna writes:
So the question is: when I push something from a subrepo, do I also
need
to do something at the root level ?
The answer is basically "it depends".
Yes, if you want the root level updated to a different version of the
subrepo. Eg, this would always be true if the change to the subrepo
is a pure optimization ready for publication, and no APIs used by the
root level are changed.
No, if not (and you need to be careful to avoid committing the updated
subrepo's current revision id to the root repo). Eg, this might be
true if you've changed the subrepo's API, and are satisfied that it's
ready to go for *new* projects, but you haven't updated the root's
code base to use the new APIs. In this case if you commit and push
from the root, you'll break the build (or worse) in new clones of the
public repo.
BTW, we haven't yet figured out *who* is supposed to update the root.
Obviously Norbert is allowed to do so. I haven't thought carefully
about it, but I doubt there's harm to allowing the developer who
commits in the subrepo to do it, as long as no other changes are
committed in root and pushed to the public repo.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4:11 p.m.
I wrote:
Just in case someone else is bitten by this: I've just
discovered an
unfortunate effect of the new Mercurial-based layout for the packages:
doing an "hg diff" from the packages root directory doesn't work
anymore[1]. You need to cd to the package directory before doing the
diff.
Stephen pointed me to diff -S for that, but in fact it doesn't
completely solve the real issue, which is that
it's not possible anymore to define a single Patcher project for
all
the packages and use the sub-project facility for restricting the
diff/commit process.
This still holds unless you define a patcher-subproject explicitely
for every package in the repo (with a :subdirectory option; otherwise,
Patcher will work from the root directory, including for commit
commands). So I still need to do something about that.
--
Resistance is futile. You will be jazzimilated.
Scientific site: http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
6:09 p.m.
Didier Verna writes:
This still holds unless you define a patcher-subproject
explicitely
for every package in the repo (with a :subdirectory option; otherwise,
Patcher will work from the root directory, including for commit
commands). So I still need to do something about that.
The user-specified list of subrepos lives in .hgsub. The list of
current commits lives in .hgsubstate. (These are siblings of .hg in
the parent repository.)
Could it hurt to automatically define Patcher subprojects for all
subrepos?
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
9:17 p.m.
Stephen J. Turnbull wrote:
The user-specified list of subrepos lives in .hgsub. The list of
current commits lives in .hgsubstate. (These are siblings of .hg in
the parent repository.)
Could it hurt to automatically define Patcher subprojects for all
subrepos?
I assume that you mean to do that dynamically (Patcher itself parsing
.hgsub). Otherwise, that wouldn't be really satisfactory (would require
regular updating etc).
But in fact, it might even be simpler to do. Patcher already knows how
to deal with a /permanent/ subproject based on a single project's
subdirectory (in such a case, the commands are executed in that
subdirectory which is what we want). I need to clean up the code so that
it is also possible for temporary subprojects (C-u M-x patcher-mail for
instance). This way, you won't ever need to maintain a list of subrepos.
--
Resistance is futile. You will be jazzimilated.
Scientific site: http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
3:27 a.m.
Didier Verna writes:
I assume that you mean to do that dynamically (Patcher itself
parsing
.hgsub).
Yes.
But in fact, it might even be simpler to do.
You're the man, Didier!
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4:12 p.m.
Stephen J. Turnbull wrote:
> But in fact, it might even be simpler to do.
You're the man, Didier!
A slightly OT question: is there a way to tell hg diff to output paths
relative to the current directory (i.e. like git diff --relative) ?
I can't find anything about that (apart from an old item on a wishlist).
There is a trick for hg status, which is to pass it the output of 'hg
root', but it doesn't work for hg diff.
Thanks.
--
Resistance is futile. You will be jazzimilated.
Scientific site: http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4734
days inactive
4740
days old
9 comments
2 participants
participants (2)
-
Didier Verna -
Stephen J. Turnbull