vc commit clears undo history

vc commit clears undo history

[Sam Steingold]

GNU Emacs 23.0.50.29 (x86_64-unknown-linux-gnu, GTK+ Version 2.8.20)
 of 2007-12-03 on nyc-qws-005
 '-with-gtk' '-with-gif=no'

a long-standing bug (bad feature?) is that when a file is committed
using vc (C-x v v) or pcvs (c in *cvs* buffer), it buffer's undo history
is discarder.

-- 
Sam Steingold (http://sds.podval.org/) on Fedora Core release 5 (Bordeaux)
http://palestinefacts.org http://thereligionofpeace.com http://truepeace.org
http://ffii.org http://iris.org.il http://honestreporting.com
I don't have an attitude problem. You have a perception problem.

Re: vc commit clears undo history

[David Kastrup]

"Sam Steingold" <ssteingold@janestcapital.com> writes:

> GNU Emacs 23.0.50.29 (x86_64-unknown-linux-gnu, GTK+ Version 2.8.20)
>  of 2007-12-03 on nyc-qws-005
>  '-with-gtk' '-with-gif=no'
>
> a long-standing bug (bad feature?) is that when a file is committed
> using vc (C-x v v) or pcvs (c in *cvs* buffer), it buffer's undo history
> is discarder.

I think the rationale for that might be that in order to have the files
identifiable as unmodified, some version control systems might do a
recheckout on them, and replace strings like $Id: whatever$ with
different information.  In any case, a revert-buffer is done on the file
after checking it in, I believe.

It might not be easy to get consistent undo information in this case.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: vc commit clears undo history

[Sam Steingold]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David Kastrup wrote:
> "Sam Steingold" <ssteingold@janestcapital.com> writes:
> 
>> GNU Emacs 23.0.50.29 (x86_64-unknown-linux-gnu, GTK+ Version 2.8.20)
>>  of 2007-12-03 on nyc-qws-005
>>  '-with-gtk' '-with-gif=no'
>>
>> a long-standing bug (bad feature?) is that when a file is committed
>> using vc (C-x v v) or pcvs (c in *cvs* buffer), it buffer's undo history
>> is discarder.
> 
> I think the rationale for that might be that in order to have the files
> identifiable as unmodified, some version control systems might do a
> recheckout on them, and replace strings like $Id: whatever$ with
> different information.  In any case, a revert-buffer is done on the file
> after checking it in, I believe.
> 
> It might not be easy to get consistent undo information in this case.

fwiw, "other vc interfaces", such as dvc (which interfaces to
distributed VCs (haz/hg/git &c), do not discard the undo information
after a commit.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHVGu7Pp1Qsf2qnMcRAvg3AJ9NhgEuV/Hd57OoiuBV6Wd1J4Ch9QCeMiBS
pWPh5yPidKRjkWQUoOUr1YM=
=Zsp3
-----END PGP SIGNATURE-----

Re: vc commit clears undo history

[Dan Nicolaescu]

Sam Steingold <sds@gnu.org> writes:

  > David Kastrup wrote:
  > > "Sam Steingold" <ssteingold@janestcapital.com> writes:
  > > 
  > >> GNU Emacs 23.0.50.29 (x86_64-unknown-linux-gnu, GTK+ Version 2.8.20)
  > >>  of 2007-12-03 on nyc-qws-005
  > >>  '-with-gtk' '-with-gif=no'
  > >>
  > >> a long-standing bug (bad feature?) is that when a file is committed
  > >> using vc (C-x v v) or pcvs (c in *cvs* buffer), it buffer's undo history
  > >> is discarder.

I have a patch that tries to avoid discarding the undo by looking at the
md5 checksum of the buffer contents and the file on disk. But I am not
100% confident that it is entirely correct....

  > > I think the rationale for that might be that in order to have the files
  > > identifiable as unmodified, some version control systems might do a
  > > recheckout on them, and replace strings like $Id: whatever$ with
  > > different information.  In any case, a revert-buffer is done on the file
  > > after checking it in, I believe.
  > > 
  > > It might not be easy to get consistent undo information in this case.
  > 
  > fwiw, "other vc interfaces", such as dvc (which interfaces to
  > distributed VCs (haz/hg/git &c), do not discard the undo information
  > after a commit.

How does it deal with the case when the file contains keywords that are
expanded on commit?
BTW VC interfaces with distributed VCs just fine.