svn and smerge-mode

svn and smerge-mode

[Nick Roberts]

With Emacs 23.0.60, visiting a file with conflicts under SVN control puts it
into smerge-mode.  I think this is too prescriptive.  On the few occasions that
I need to resolve conflicts I do it by hand and I've lost information trying to
use smerge-mode.  If someone wants smerge-mode they can easily turn it on.

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

Re: svn and smerge-mode

[Stefan Monnier]

> With Emacs 23.0.60, visiting a file with conflicts under SVN control
> puts it into smerge-mode.  I think this is too prescriptive.  On the
> few occasions that I need to resolve conflicts I do it by hand and
> I've lost information trying to use smerge-mode.  If someone wants
> smerge-mode they can easily turn it on.

The intention of smerge-mode is to be "out of the way" as much
as possible.  Now admittedly, it's not perfectly so, but rather than
leave it off by default or provide a config var for it, I'd rather fix
the problem you encountered.

Basically, the only effect of turning on smerge-mode should be to
highlight the conflicts with font-lock (as long as you don't use the C-x
^ prefix, obviously).  So which part of smerge-mode annoys you?
And more importantly, which part caused you to lose information?


        Stefan

Re: svn and smerge-mode

[Nick Roberts]

 > Basically, the only effect of turning on smerge-mode should be to
 > highlight the conflicts with font-lock (as long as you don't use the C-x
 > ^ prefix, obviously).  So which part of smerge-mode annoys you?
 > And more importantly, which part caused you to lose information?

I just use it too infrequently to remember how it works.

ISTR I lost my original working (.#) file trying to exit smerge-ediff.  It
wasn't clear to me where the changes I had selected were going to end up (I
thought I would need to save the *ediff-merge* buffer).  Even though I quit
smerge-ediff without selecting any changes, I was told:

Conflict resolution finished; you may save the buffer

Evidently this actually only differed by the conflict markers,
e.g.,"<<<<<<< filename" vs "<<<<<<< variant A").

I thought that when I saved this buffer the original working was removed
although I can't reproduce this now.

In summary, smerge-mode seems to start off quite innocently but may lead you
into places that you don't want to be.


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

Re: svn and smerge-mode

[Stefan Monnier]

>> Basically, the only effect of turning on smerge-mode should be to
>> highlight the conflicts with font-lock (as long as you don't use the C-x
>> ^ prefix, obviously).  So which part of smerge-mode annoys you?
>> And more importantly, which part caused you to lose information?

> I just use it too infrequently to remember how it works.

> ISTR I lost my original working (.#) file trying to exit smerge-ediff.  It
> wasn't clear to me where the changes I had selected were going to end up (I
> thought I would need to save the *ediff-merge* buffer).  Even though I quit
> smerge-ediff without selecting any changes, I was told:

> Conflict resolution finished; you may save the buffer

> Evidently this actually only differed by the conflict markers,
> e.g.,"<<<<<<< filename" vs "<<<<<<< variant A").

> I thought that when I saved this buffer the original working was removed
> although I can't reproduce this now.

> In summary, smerge-mode seems to start off quite innocently but may lead you
> into places that you don't want to be.

So it sounds like a problem in the smerge-ediff code.  I never use(d)
that code, so it's not really a surprise.  If you can provide
a reproducible testcase, that would be very helpful, but otherwise just
give us as detailed a description as possible (e.g. what .#* file are
you talking about?).


        Stefan

Re: svn and smerge-mode

[Miles Bader]

Nick Roberts <nickrob@snap.net.nz> writes:
> With Emacs 23.0.60, visiting a file with conflicts under SVN control puts it
> into smerge-mode.  I think this is too prescriptive.  On the few occasions that
> I need to resolve conflicts I do it by hand and I've lost information trying to
> use smerge-mode.  If someone wants smerge-mode they can easily turn it on.

No, if smerge-mode is causing information loss without user action that
should be fixed.

-Miles

-- 
Run away!  Run away!

Re: svn and smerge-mode

[Nick Roberts]

 > > With Emacs 23.0.60, visiting a file with conflicts under SVN control puts
 > > it into smerge-mode.  I think this is too prescriptive.  On the few
 > > occasions that I need to resolve conflicts I do it by hand and I've lost
 > > information trying to use smerge-mode.  If someone wants smerge-mode they
 > > can easily turn it on.
 > 
 > No, if smerge-mode is causing information loss without user action that
 > should be fixed.

Clearly it wasn't without user action.  I think it should only enabled by
default if it is the only natural choice.  If that's what you mean by "fixed",
then that may not always be possible.

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

Re: svn and smerge-mode

[Miles Bader]

Nick Roberts <nickrob@snap.net.nz> writes:
>  > No, if smerge-mode is causing information loss without user action that
>  > should be fixed.
>
> Clearly it wasn't without user action.  I think it should only enabled by
> default if it is the only natural choice.  If that's what you mean by "fixed",
> then that may not always be possible.

It seems a natural choice when there are conflict markers..

If you're having problems because you explicitly used smerge-mode
commands, then why not just stop using those commands?  Why does smerge
have to be disabled (after all, even if you don't use any smerge
commands at all, the syntax highlighting behavior is quite useful)?

[Of course if your problems stemmed from _bugs_ in the smerge commands,
then those bugs should be fixed too.]

-Miles

-- 
Logic, n. The art of thinking and reasoning in strict accordance with the
limitations and incapacities of the human misunderstanding.