Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

bug#1158: 23.0.60; vc mode not working under newest Emacs.app

80 views
Skip to first unread message

Neo Lee

unread,
Oct 13, 2008, 7:01:16 AM10/13/08
to emacs-pr...@gnu.org
Dear all,



I'm using both 22.2(in terminal) and 23.0 in Emacs.app on my MacBook Pro, 

both installed by MacPorts. 


There is no critical problem, almost everything are smoothly running, except the 

vc mode. It runs perfectly in 22.2 in terminal, but never really work in Emacs.app.

When I open some file which IS under source control (svn or git), it says 

'Load vc-git...done' or something like that (no errors), but the mode actually 

not activated. It's really annoying.


Any suggestions? Or just some similar situation that have been mentioned here before?


Thanks.



Some system info below. 


In GNU Emacs 23.0.60.1 (i386-apple-darwin9.5.0, *Step 9.0rc3)

 of 2008-10-05 on oasis

configured using `configure '--with-ns' '--without-x' '--prefix=/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_aqua_emacs-app/work/emacs-23.0.0_NS-9.0rc3/nextstep/build/Emacs.app/Contents/Resources' '--exec_prefix=/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_aqua_emacs-app/work/emacs-23.0.0_NS-9.0rc3/nextstep/build/Emacs.app/Contents/MacOS' '--libexecdir=/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_aqua_emacs-app/work/emacs-23.0.0_NS-9.0rc3/nextstep/build/Emacs.app/Contents/MacOS/libexec' '--with-pop' '--enable-font-backend' '--without-freetype' 'CC=gcc-4.0' 'CFLAGS=-g -O3 -arch ppc -arch i386' 'CPPFLAGS=' 'CPP=' 'LDFLAGS=''


Important settings:

  value of $LC_ALL: nil

  value of $LC_COLLATE: nil

  value of $LC_CTYPE: nil

  value of $LC_MESSAGES: nil

  value of $LC_MONETARY: nil

  value of $LC_NUMERIC: nil

  value of $LC_TIME: nil

  value of $LANG: nil

  value of $XMODIFIERS: nil

  locale-coding-system: nil

  default-enable-multibyte-characters: t


Major mode: Fundamental


Minor modes in effect:

  ecb-minor-mode: t

  tooltip-mode: t

  show-paren-mode: t

  mouse-wheel-mode: t

  menu-bar-mode: t

  file-name-shadow-mode: t

  global-font-lock-mode: t

  blink-cursor-mode: t

  global-auto-composition-mode: t

  unify-8859-on-decoding-mode: t

  auto-compression-mode: t

  line-number-mode: t

  transient-mark-mode: t


Recent input:

<help-echo> <menu-bar> <help-menu> <report-emacs-bug>


Recent messages:

Loading /Users/neo/.emacs.d/cedet/common/cedet.el (source)...done

ECB 2.33beta2 uses loaded semantic 2.0pre4, eieio 1.0 and speedbar 1.0.1.

Loading /Users/neo/.emacs.d/yaml-mode.el (source)...done

Loading /Users/neo/.emacs.d/nxml-mode/rng-auto.el (source)...done

Loading /Users/neo/.emacs.d/find-recursive.el (source)...done

Loading /Users/neo/.emacs.d/snippet.el (source)...done

Loading /Users/neo/.ecb-user-layouts.el (source)...done

The ECB is now activated.

There are no NEWS to display.

For information about GNU Emacs and the GNU system, type C-h C-a.



--
No, I will not fix your computer!

neo...@gmail.com

Nick Roberts

unread,
Oct 14, 2008, 3:48:40 AM10/14/08
to Neo Lee, 11...@emacsbugs.donarmstrong.com
> When I open some file which IS under source control (svn or git), it says
>
> 'Load vc-git...done' or something like that (no errors), but the mode
> actually
>
> not activated. It's really annoying.
>
>
> Any suggestions? Or just some similar situation that have been mentioned
> here before?

>...


> Recent messages:
>
> Loading /Users/neo/.emacs.d/cedet/common/cedet.el (source)...done
>
> ECB 2.33beta2 uses loaded semantic 2.0pre4, eieio 1.0 and speedbar 1.0.1.
>
> Loading /Users/neo/.emacs.d/yaml-mode.el (source)...done
>
> Loading /Users/neo/.emacs.d/nxml-mode/rng-auto.el (source)...done
>
> Loading /Users/neo/.emacs.d/find-recursive.el (source)...done
>
> Loading /Users/neo/.emacs.d/snippet.el (source)...done
>
> Loading /Users/neo/.ecb-user-layouts.el (source)...done
>
> The ECB is now activated.

Have you read the node in the Emacs manual about bug reporting?

In particular what happens if you don't load your .emacs, i.e., start Emacs
with "-q"?

--
Nick http://www.inet.net.nz/~nickrob


Nick Roberts

unread,
Oct 14, 2008, 6:35:40 AM10/14/08
to Neo Lee, 11...@emacsbugs.donarmstrong.com
[Please keep 11...@emacsbugs.donarmstrong.com as a recipient as others might
resolve this issue and so that it can be closed when appropriate]

Neo Lee writes:
> Hi Nick,
>
> Thanks for your response.


>
> On Tue, Oct 14, 2008 at 3:48 PM, Nick Roberts <nic...@snap.net.nz> wrote:
>
> >
> > Have you read the node in the Emacs manual about bug reporting?
> >
> > In particular what happens if you don't load your .emacs, i.e., start Emacs
> > with "-q"?
> >
>

> Sure I have. At very first time I noticed that issue I tried to remove my
> .emacs, started a clean Emacs.app instance, and meet exactly the same
> situation that I described in my last email. If any possibility of other
> mods causing this issue, I would not ask here :) For your convenience I
> attached the standard debug info on this fresh environment at the end of
> this email.

We use this information to analyse the bug. If it tells us that CEDET and
ECB have been loaded then we presume that may be a factor.

> Since everything is OK in my emacs 22.2, I guess that it has something to do
> with either emacs 23 or Emacs.app. If somebody has similar issue on emacs
> 23, it will be some bug in the beta code. If not (people using Emacs.app are
> not as many as the ones using emacs 23), maybe there's some trick in
> Emacs.app (the Mac OS X Cocoa port), e.g. some path configuration hacking?

vc-mode works for me on Mac OS X for SVN. However, I've not used MacPorts
but checked the code out from the repository at Savannah and just built with
"configure --with-ns".

--
Nick http://www.inet.net.nz/~nickrob


Glenn Morris

unread,
Oct 14, 2008, 9:06:54 PM10/14/08
to 11...@emacsbugs.donarmstrong.com, neo...@gmail.com
reassign 1158 emacs,ns
tags 1158 unreproducible,moreinfo

"Neo Lee" wrote:

> In GNU Emacs 23.0.60.1 (i386-apple-darwin9.5.0, *Step 9.0rc3)

This seems to be the last stand-alone release of Emacs.app, rather
than the newest version, which would come from the Emacs CVS trunk.

The fact that Nick cannot reproduce this indicates it may not be
present in the CVS trunk.


Neo

unread,
Oct 14, 2008, 9:37:30 PM10/14/08
to

Just done such steps:

1. Remove all correlated ports from MacPorts (emacs, emacs-app, etc.)
2. Fetch newest Emacs source from cvs.savannah.gnu.org. ("cvs -z3 -
d:pserver:anon...@cvs.savannah.gnu.org:/sources/emacs co" Is this
command checking out the trunk branch of newest source?)
3. Build it with default configuration and install the X edition of
Emacs with no errors, ECB brings some issue on customization, but
others runs like a charm, esp. the vc-mode.
4. After 'make distclean', build with --with-ns and install a
Emacs.app, still no errors. But it's a pity that it has the same issue
we talked about (even using -q parameter to start a clean
environment).

I'll let it be for a while :S

Glenn Morris

unread,
Oct 14, 2008, 9:45:23 PM10/14/08
to Neo, 11...@emacsbugs.donarmstrong.com
Neo wrote:

> 1. Remove all correlated ports from MacPorts (emacs, emacs-app, etc.)
> 2. Fetch newest Emacs source from cvs.savannah.gnu.org. ("cvs -z3 -
> d:pserver:anon...@cvs.savannah.gnu.org:/sources/emacs co" Is this
> command checking out the trunk branch of newest source?)
> 3. Build it with default configuration and install the X edition of
> Emacs with no errors, ECB brings some issue on customization, but
> others runs like a charm, esp. the vc-mode.
> 4. After 'make distclean', build with --with-ns and install a
> Emacs.app, still no errors. But it's a pity that it has the same issue
> we talked about (even using -q parameter to start a clean
> environment).

What is the actual bug?

it says

'Load vc-git...done' or something like that (no errors), but the mode
actually not activated.

I don't understand what this means.


Neo Lee

unread,
Oct 14, 2008, 10:04:24 PM10/14/08
to Glenn Morris
On Wed, Oct 15, 2008 at 9:45 AM, Glenn Morris <r...@gnu.org> wrote:
What is the actual bug?

    it says

    'Load vc-git...done' or something like that (no errors), but the mode
    actually not activated.

I don't understand what this means.

It means, when I open any file under source control(say, git), the vc-mode will be auto-loaded and display some message in mini-buffer like (without any error message):

Load vc-git... done.

But the mode actually not working. If I try to use any of the vc-* commands after that, the message in mini-buffer will be:

No fileset is available here.

Other vc-mode functions are also missing, e.g. the revision displayed on the bar.

Glenn Morris

unread,
Oct 16, 2008, 1:58:57 AM10/16/08
to neo lee, 11...@emacsbugs.donarmstrong.com

neo lee wrote (on Wed, 15 Oct 2008 at 10:04 +0800):

> But the mode actually not working. If I try to use any of the vc-* commands
> after that, the message in mini-buffer will be:
>
> No fileset is available here.
>
> Other vc-mode functions are also missing, e.g. the revision displayed on the
> bar.

It is surprising that a change in configure options could affect such
a thing.


What happens if you start with `emacs -Q', and open the README file
from the directory where you checked out the CVS trunk?

If you repeat the experiment, but do M-x toggle-debug-on-error before
opening README, do you get any backtrace?

What does

M-: (executable-find "cvs")

return?

What does

M-: (vc-cvs-registered "/path/to/emacs/cvs/README")

return?


What does

M-x list-load-path-shadows

say?


Neo Lee

unread,
Oct 16, 2008, 4:05:21 AM10/16/08
to Glenn Morris
On Thu, Oct 16, 2008 at 1:58 PM, Glenn Morris <r...@gnu.org> wrote:

It is surprising that a change in configure options could affect such
a thing.


What happens if you start with `emacs -Q', and open the README file
from the directory where you checked out the CVS trunk?

If you repeat the experiment, but do M-x toggle-debug-on-error before
opening README, do you get any backtrace?

What does

M-: (executable-find "cvs")

return?

What does

M-: (vc-cvs-registered "/path/to/emacs/cvs/README")

return?


What does

M-x list-load-path-shadows

say?

I've got some new discovery about this problem. After installing Emacs.app in my Application folder, I start Emacs.app from terminal by:

% /Applications/Emacs.app/Contents/MacOS/Emacs

It runs like a charm, even with my full customization (includes slime, ecb, ruby & rails mode, as well as lots of other modes). But if I launch it by click the Emacs icon on the Application folder, vc-mode will be in the weird situation as I described before, no matter I load my .emacs or not. I guess this could direct us to some correct way to solve this problem. 

To make thing clearer, I removed my .emacs, launched Emacs.app from Application folder, and did the following tests:

1. C-x C-f ~/Code/Emacs/emacs/README
It opened, with the cvs mode correctly activated. I play the vc-* commands for a while, seems everything is OK.

2. C-x C-f ~/Ruby19/src/load.c
It opened, with "Loading vc-svn...done" message in the mini buffer, but all vc-* commands not working (feedback "No fileset is available here." message). 

3. Running diagnosis evaluation as you instructed:

M-: (executable-find "svn") says:
"/usr/bin/svn"

Which is NOT the correct path of my svn. In my working MacBook Pro, the SVN from Leopard installation is in /usr/bin, but the working SVN installation is in /opt/local/bin/ (from MacPorts). Here is some information about my SVNs:

$ whereis svn
/usr/bin/svn
$ which svn
/opt/local/bin/svn
$ /usr/bin/svn --version
svn, version 1.4.4 (r25188) compiled Sep 23 2007, 22:32:34
$ svn --version
svn, version 1.5.3 (r33570) compiled Oct 11 2008, 10:20:01

Furthermore, 

M-: (vc-svn-registered "~/Ruby19/src/load.c") says:
nil

M-x list-load-path-shadows says (in mini buffer):
Checking 1 files in /Applications/Emacs.app/Contents/Resources/site-lisp...
Checking 587 files in /Applications/Emacs.app/Contents/Resources/lisp...
Checking 58 files in /Applications/Emacs.app/Contents/Resources/lisp/url...
Checking 90 files in /Applications/Emacs.app/Contents/Resources/lisp/textmodes...
Checking 166 files in /Applications/Emacs.app/Contents/Resources/lisp/progmodes...
Checking 57 files in /Applications/Emacs.app/Contents/Resources/lisp/play...
Checking 62 files in /Applications/Emacs.app/Contents/Resources/lisp/org...
Checking 29 files in /Applications/Emacs.app/Contents/Resources/lisp/obsolete...
Checking 49 files in /Applications/Emacs.app/Contents/Resources/lisp/nxml...
Checking 1 files in /Applications/Emacs.app/Contents/Resources/lisp/nxml/char-name...
Checking 108 files in /Applications/Emacs.app/Contents/Resources/lisp/net...
Checking 48 files in /Applications/Emacs.app/Contents/Resources/lisp/mh-e...
Checking 88 files in /Applications/Emacs.app/Contents/Resources/lisp/mail...
Checking 58 files in /Applications/Emacs.app/Contents/Resources/lisp/language...
Checking 64 files in /Applications/Emacs.app/Contents/Resources/lisp/international...
Checking 264 files in /Applications/Emacs.app/Contents/Resources/lisp/gnus...
Checking 58 files in /Applications/Emacs.app/Contents/Resources/lisp/eshell...
Checking 70 files in /Applications/Emacs.app/Contents/Resources/lisp/erc...
Checking 52 files in /Applications/Emacs.app/Contents/Resources/lisp/emulation...
Checking 128 files in /Applications/Emacs.app/Contents/Resources/lisp/emacs-lisp...
Checking 57 files in /Applications/Emacs.app/Contents/Resources/lisp/calendar...
Checking 87 files in /Applications/Emacs.app/Contents/Resources/lisp/calc...
Checking 1 files in /Applications/Emacs.app/Contents/Resources/leim...

and finally in *Shadow* buffer:
No Emacs Lisp load-path shadowings were found

I did all these tests on git, and gave similar results, which has no reason to be repeated here.
NOTE: My git is built from cloned source and installed in /usr/local/bin. 

$ whereis git
$ which git
/usr/local/bin/git

And when I start Emacs.app from terminal, it gives correct path.

Actually I do not know what the list-load-path-shadows exactly is, so I just list all I saw. But from all the information listed above, I guess the whole problem is something about the search path, and the key is the way Mac OS X launches GUI application from Finder (my PATH setting in .zshrc bypassed, which I consider as a common mistake, but I've never noticed). 

I now know how to workaround my problem, but if any suggestion or other helpful information about this issue, please let me know. Thanks a lot for your response.

Emacs bug Tracking System

unread,
Oct 16, 2008, 2:10:13 PM10/16/08
to Glenn Morris
Dear all,



I'm using both 22.2(in terminal) and 23.0 in Emacs.app on my MacBook Pro, 

both installed by MacPorts. 


There is no critical problem, almost everything are smoothly running, except the 

vc mode. It runs perfectly in 22.2 in terminal, but never really work in Emacs.app.

When I open some file which IS under source control (svn or git), it says 

'Load vc-git...done' or something like that (no errors), but the mode actually 

not activated. It's really annoying.


Any suggestions? Or just some similar situation that have been mentioned here before?


Thanks.



Some system info below. 


In GNU Emacs 23.0.60.1 (i386-apple-darwin9.5.0, *Step 9.0rc3)

 of 2008-10-05 on oasis

Recent messages:

Loading /Users/neo/.emacs.d/cedet/common/cedet.el (source)...done

ECB 2.33beta2 uses loaded semantic 2.0pre4, eieio 1.0 and speedbar 1.0.1.

Loading /Users/neo/.emacs.d/yaml-mode.el (source)...done

Loading /Users/neo/.emacs.d/nxml-mode/rng-auto.el (source)...done

Loading /Users/neo/.emacs.d/find-recursive.el (source)...done

Loading /Users/neo/.emacs.d/snippet.el (source)...done

Loading /Users/neo/.ecb-user-layouts.el (source)...done

The ECB is now activated.

There are no NEWS to display.

For information about GNU Emacs and the GNU system, type C-h C-a.


Neo Lee

unread,
Oct 16, 2008, 4:11:01 AM10/16/08
to Glenn Morris
One more word about this, I think that when such things happen Emacs would better give out some clearer message. :)

Stefan Monnier

unread,
Oct 17, 2008, 11:24:59 AM10/17/08
to Neo Lee, 11...@emacsbugs.donarmstrong.com
> 3. Running diagnosis evaluation as you instructed:
> M-: (executable-find "svn") says:
> "/usr/bin/svn"

> Which is NOT the correct path of my svn. In my working MacBook Pro, the SVN
> from Leopard installation is in /usr/bin, but the working SVN installation
> is in /opt/local/bin/ (from MacPorts). Here is some information about my
> SVNs:

That's not Emacs's fault: it's because /opt/local/bin is not in
your PATH. Most likely, you added /opt/local/bin to your PATH in your
~/.bashrc or somesuch, but this only affects things you do inside
a terminal since Mac OS X doesn't load this file when you login.
IIRC there's a ~/.macosx/environment.plist file you can use to set
envvars globally independently from any shell.


Stefan


0 new messages