State of Emacs repository backup

State of Emacs repository backup

[Andreas Schwab]

I have compared the 20090529 backup of the Emacs CVS repository with the
state of the git mirror on repo.or.cz (which was last mirrored on
2009-05-29 23:00 +0200).  There are two incomplete changesets with a
total of two missing revisions:

commitid Fvex0CkUAhHjSDPt:
revision 1.286 of doc/lispref/ChangeLog (about 2009-05-28 13:30:46 +0000)

commitid lPtsAMeJsIIBRHPt:
revision 1.15662 of lisp/ChangeLog (2009-05-29 01:31:40 +0000)

There would be the possibility to reconstruct the two missing revisions
and ask the Savannah admins to overwrite the two affected RCS files in
the repository, bringing it back to the state as of 2009-05-29 23:00
+0200.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: State of Emacs repository backup

[Chong Yidong]

Andreas Schwab <schwab@linux-m68k.org> writes:

> I have compared the 20090529 backup of the Emacs CVS repository with the
> state of the git mirror on repo.or.cz (which was last mirrored on
> 2009-05-29 23:00 +0200).  There are two incomplete changesets with a
> total of two missing revisions:
>
> commitid Fvex0CkUAhHjSDPt:
> revision 1.286 of doc/lispref/ChangeLog (about 2009-05-28 13:30:46 +0000)
>
> commitid lPtsAMeJsIIBRHPt:
> revision 1.15662 of lisp/ChangeLog (2009-05-29 01:31:40 +0000)
>
> There would be the possibility to reconstruct the two missing revisions
> and ask the Savannah admins to overwrite the two affected RCS files in
> the repository, bringing it back to the state as of 2009-05-29 23:00
> +0200.

That sounds good.  Are the ChangeLog files the only ones affected?  I
believe these changes are

2009-05-28  Chong Yidong  <cyd@stupidchicken.com>

  * frames.texi (Text Terminal Colors): Multi-tty is already
  implemented, but tty-local colors are not.

and

2009-05-29  Stefan Monnier  <monnier@iro.umontreal.ca>

  * startup.el (normal-no-mouse-startup-screen): Use F1 rather than C-h
  if C-h is remapped to something else like DEL.

Should we contact the Savannah sysadmins directly, or wait for another
announcement from them regarding CVS recovery?

Re: State of Emacs repository backup

[Jim Meyering]

Chong Yidong wrote:
> Andreas Schwab <schwab@linux-m68k.org> writes:
>
>> I have compared the 20090529 backup of the Emacs CVS repository with the
>> state of the git mirror on repo.or.cz (which was last mirrored on
>> 2009-05-29 23:00 +0200).  There are two incomplete changesets with a
>> total of two missing revisions:
>>
>> commitid Fvex0CkUAhHjSDPt:
>> revision 1.286 of doc/lispref/ChangeLog (about 2009-05-28 13:30:46 +0000)
>>
>> commitid lPtsAMeJsIIBRHPt:
>> revision 1.15662 of lisp/ChangeLog (2009-05-29 01:31:40 +0000)
>>
>> There would be the possibility to reconstruct the two missing revisions
>> and ask the Savannah admins to overwrite the two affected RCS files in
>> the repository, bringing it back to the state as of 2009-05-29 23:00
>> +0200.
>
> That sounds good.  Are the ChangeLog files the only ones affected?  I
> believe these changes are
>
> 2009-05-28  Chong Yidong  <cyd@stupidchicken.com>
>
>   * frames.texi (Text Terminal Colors): Multi-tty is already
>   implemented, but tty-local colors are not.
>
> and
>
> 2009-05-29  Stefan Monnier  <monnier@iro.umontreal.ca>
>
>   * startup.el (normal-no-mouse-startup-screen): Use F1 rather than C-h
>   if C-h is remapped to something else like DEL.
>
> Should we contact the Savannah sysadmins directly, or wait for another
> announcement from them regarding CVS recovery?

Thanks for investigating that.
If you provide correct ,v files, I can put them in place.

Once the CVS repository is restored,
I will reenable the mirror-to-git process.

Re: State of Emacs repository backup

[Andreas Schwab]

Chong Yidong <cyd@stupidchicken.com> writes:

> That sounds good.  Are the ChangeLog files the only ones affected?

Yes.  The other members of these changesets are complete, and the file
contents of all other files match as well.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: State of Emacs repository backup

[Chong Yidong]

Andreas Schwab <schwab@linux-m68k.org> writes:

> Yes.  The other members of these changesets are complete, and the file
> contents of all other files match as well.

Huh, we lucked out.  I guess we only have to wait for the savannah
sysadmins to restore the CVS service.

Re: State of Emacs repository backup

[Miles Bader]

Chong Yidong <cyd@stupidchicken.com> writes:
>> I have compared the 20090529 backup of the Emacs CVS repository with the
>> state of the git mirror on repo.or.cz (which was last mirrored on
>> 2009-05-29 23:00 +0200).  There are two incomplete changesets with a
>> total of two missing revisions:
>>
>> commitid Fvex0CkUAhHjSDPt:
>> revision 1.286 of doc/lispref/ChangeLog (about 2009-05-28 13:30:46 +0000)
>>
>> commitid lPtsAMeJsIIBRHPt:
>> revision 1.15662 of lisp/ChangeLog (2009-05-29 01:31:40 +0000)
>>
>> There would be the possibility to reconstruct the two missing revisions
>> and ask the Savannah admins to overwrite the two affected RCS files in
>> the repository, bringing it back to the state as of 2009-05-29 23:00
>> +0200.
...
> Should we contact the Savannah sysadmins directly, or wait for another
> announcement from them regarding CVS recovery?

I have a snapshot of the savannah emacs git mirror which might be a bit
later; what's the your git log look like?  [Also I suppose somebody else
might have done a git pull after me and gotten something even newer.]

Here's the last 6 commits in my repo:

   commit c2f9b5857cc50663229ad3ec28608fe615ac9e34
   Author: Chong Yidong <cyd@stupidchicken.com>
   Date:   Sat May 30 00:31:59 2009 +0000

       * coding.c (get_translation_table): Check Venable_character_translation.

   commit 60b9cf6a256bd72427f41c510af9c5d933a163a8
   Author: Stefan Monnier <monnier@cs.yale.edu>
   Date:   Fri May 29 01:31:40 2009 +0000

       (normal-no-mouse-startup-screen): Use F1 rather than C-h
       if C-h is remapped to something else like DEL.

   commit 4bdfa5650c33b0873d7cd69777b8d706052b9d6b
   Author: Chong Yidong <cyd@stupidchicken.com>
   Date:   Thu May 28 20:36:03 2009 +0000

       * mark.texi (Mark): Further clarifications.
       (Setting Mark): Emphasize that C-SPC activates the mark.

   commit 5a189b651281f380aad67809c70b95e6d3908cca
   Author: Chong Yidong <cyd@stupidchicken.com>
   Date:   Thu May 28 16:53:05 2009 +0000

       * mark.texi (Mark): Mention disabling Transient Mark mode.

   commit 0344485f3fea5ad82b170c267da4fa39b92fecb5
   Author: Chong Yidong <cyd@stupidchicken.com>
   Date:   Thu May 28 16:48:05 2009 +0000

       * mark.texi (Mark): Clarify introduction.
       (Using Region, Persistent Mark): Use "active mark" instead of "active
       region".

   commit 3c65b5f6d3858f32915f47006cad2c2907ba3c7f
   Author: Chong Yidong <cyd@stupidchicken.com>
   Date:   Thu May 28 13:30:46 2009 +0000

       * frames.texi (Text Terminal Colors): Multi-tty is already
       implemented, but tty-local colors are not.

-Miles

-- 
Patience, n. A minor form of despair, disguised as a virtue.

Re: State of Emacs repository backup

[Andreas Schwab]

Miles Bader <miles@gnu.org> writes:

> I have a snapshot of the savannah emacs git mirror which might be a bit
> later; what's the your git log look like?  [Also I suppose somebody else
> might have done a git pull after me and gotten something even newer.]
>
> Here's the last 6 commits in my repo:
>
>    commit c2f9b5857cc50663229ad3ec28608fe615ac9e34
>    Author: Chong Yidong <cyd@stupidchicken.com>
>    Date:   Sat May 30 00:31:59 2009 +0000
>
>        * coding.c (get_translation_table): Check Venable_character_translation.
>

This is the only commit ahead of the repo.or.cz tree.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: State of Emacs repository backup

[Glenn Morris]

[-- Attachment #1: Type: text/plain, Size: 807 bytes --]


Should it prove useful, I have an rsync of the CVS repository (updated
twice a day) that seems to include everything from emacs-diffs except
the May 29 changes to src/coding.c and associated ChangeLog.

Diffing this against the backup-20090529-incomplete provided by
Savannah:

http://savannah.gnu.org/forum/forum.php?forum_id=5828

reveals some other differences - there seem to be a few missing
top-level CVSROOT files in the Savannah 20090529 backup, as well as a
missing emacs/nt/inc/sys/CVS/fileattr.

The diff is small and is attached.

The context diff against the Savannah 20090429 backup is ~ 80000
lines, I haven't examined it. I'm not sure if the Savannah RAID
problems mean there might be some corrupted files in the 20090529
backup and in my copy that would make this a worthwhile exercise?



[-- Attachment #2: backup-20090529-incomplete.diff.gz --]
[-- Type: application/octet-stream, Size: 1666 bytes --]

Re: State of Emacs repository backup

[Andreas Schwab]

Glenn Morris <rgm@gnu.org> writes:

> reveals some other differences - there seem to be a few missing
> top-level CVSROOT files in the Savannah 20090529 backup, as well as a
> missing emacs/nt/inc/sys/CVS/fileattr.

I think these files are maintained outside of CVS and will be
automatically recreated.

> The diff is small and is attached.

This matches exactly what I was able to reconstruct from the git
cvsimport.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: State of Emacs repository backup

[Miles Bader]

Jim Meyering <jim@meyering.net> writes:
> Once the CVS repository is restored,
> I will reenable the mirror-to-git process.

So what should be done with the current state of the emacs git
mirror...?

Right now my local git pull is ahead of what's on savannah, and I
suppose could be pushed; is there any reason to do that, or will things
automatically get restored by re-importing from CVS?

-miles

-- 
"Whatever you do will be insignificant, but it is very important that
 you do it."  Mahatma Gandhi

Re: State of Emacs repository backup

[Chong Yidong]

Glenn Morris <rgm@gnu.org> writes:

> Should it prove useful, I have an rsync of the CVS repository (updated
> twice a day) that seems to include everything from emacs-diffs except
> the May 29 changes to src/coding.c and associated ChangeLog.

Let's go with this one, as it seems easiest.  I'll check the coding.c
change back in; hopefully, this won't cause any problems.

Glenn, could you post a support request with the Savannah admins?  The
instructions are at

https://savannah.gnu.org/forum/forum.php?forum_id=5828

Re: State of Emacs repository backup

[Glenn Morris]

Chong Yidong wrote:

> Glenn, could you post a support request with the Savannah admins?

Done (item #106848).

BTW, I also have a copy of the web repository; not sure if that will
be needed too.

Re: State of Emacs repository backup

[Jim Meyering]

Miles Bader wrote:
> Jim Meyering <jim@meyering.net> writes:
>> Once the CVS repository is restored,
>> I will reenable the mirror-to-git process.
>
> So what should be done with the current state of the emacs git
> mirror...?
>
> Right now my local git pull is ahead of what's on savannah, and I
> suppose could be pushed; is there any reason to do that, or will things
> automatically get restored by re-importing from CVS?

There should be no need to push manually.

As soon as the CVS situation is restored, I'll restart
the mirroring process, and that should restore everything
to working order.

Re: State of Emacs repository backup

[Jim Meyering]

Jim Meyering wrote:
> Miles Bader wrote:
>> Jim Meyering <jim@meyering.net> writes:
>>> Once the CVS repository is restored,
>>> I will reenable the mirror-to-git process.
>>
>> So what should be done with the current state of the emacs git
>> mirror...?
>>
>> Right now my local git pull is ahead of what's on savannah, and I
>> suppose could be pushed; is there any reason to do that, or will things
>> automatically get restored by re-importing from CVS?
>
> There should be no need to push manually.
>
> As soon as the CVS situation is restored, I'll restart
> the mirroring process, and that should restore everything
> to working order.

In case you're not following, storage for savannah's cvs
repositories is not yet stable,

    http://identi.ca/group/fsfstatus

so don't count on having a usable CVS repository today.

Re: State of Emacs repository backup

[Jim Meyering]

Jim Meyering wrote:

> Jim Meyering wrote:
>> Miles Bader wrote:
>>> Jim Meyering <jim@meyering.net> writes:
>>>> Once the CVS repository is restored,
>>>> I will reenable the mirror-to-git process.
>>>
>>> So what should be done with the current state of the emacs git
>>> mirror...?
>>>
>>> Right now my local git pull is ahead of what's on savannah, and I
>>> suppose could be pushed; is there any reason to do that, or will things
>>> automatically get restored by re-importing from CVS?
>>
>> There should be no need to push manually.
>>
>> As soon as the CVS situation is restored, I'll restart
>> the mirroring process, and that should restore everything
>> to working order.
>
> In case you're not following, storage for savannah's cvs
> repositories is not yet stable,
>
>     http://identi.ca/group/fsfstatus
>
> so don't count on having a usable CVS repository today.

Looks like you get it anyway:

The CVS repository is restored, and the git repo
is temporarily in sync.  Details here:

  https://savannah.gnu.org/support/index.php?106848

Re: State of Emacs repository backup

[Chong Yidong]

Glenn Morris <rgm@gnu.org> writes:

> Chong Yidong wrote:
>
>> Glenn, could you post a support request with the Savannah admins?
>
> Done (item #106848).

Seems to be working fine now.  Thanks.

> BTW, I also have a copy of the web repository; not sure if that will
> be needed too.

I don't think there were any updates in the last month, were there?

Re: State of Emacs repository backup

[Glenn Morris]

Chong Yidong wrote:

>> BTW, I also have a copy of the web repository; not sure if that will
>> be needed too.
>
> I don't think there were any updates in the last month, were there?

Don't know. Actually, I think my copy is corrupt - it has no CVS
directories. Anyway, since Savannah found a May 27th backup, it will
probably be OK.

Re: State of Emacs repository backup

[Jim Meyering]

I wrote:
> The CVS repository is restored, and the git repo
> is temporarily in sync.  Details here:
>
>   https://savannah.gnu.org/support/index.php?106848

I've done one more manual cvs-to-git sync and have reenabled
the automatic cvs-to-git mirroring that runs every 30 minutes.

Re: State of Emacs repository backup

[Jim Meyering]

David Reitter wrote:
> On Jun 4, 2009, at 7:03 AM, Jim Meyering wrote:
>
>> I wrote:
>>> The CVS repository is restored, and the git repo
>>> is temporarily in sync.  Details here:
>>>
>>>  https://savannah.gnu.org/support/index.php?106848
>
> Is there a permissions problem with the history file?
>
> /sources/emacs/emacs/src/ChangeLog,v  <--  ChangeLog
> new revision: 1.7558; previous revision: 1.7557
> cvs commit: warning: cannot open history file `/sources/emacs/CVSROOT/
> history' for write: Permission denied

It was missing, and the containing directory was writable
only by root.  I've just created the file and made it writable
by group "emacs", so you should no longer see that warning.

Re: State of Emacs repository backup

[Andreas Schwab]

Glenn Morris <rgm@gnu.org> writes:

> Don't know. Actually, I think my copy is corrupt - it has no CVS
> directories.

A CVS repository normally doesn't have CVS directories.  They are only
used for the optional fileattr server optimisation.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."