You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Ben Collins-Sussman <su...@collab.net> on 2003/05/09 22:23:30 UTC

Re: svn commit: rev 5872 - branches/release-0.22.1

mprice@tigris.org writes:

> + * update cvs2svn.py script

This is actually a really cool new feature going out in 0.22.1.  I
feel like Karl hasn't bragged enough.  :-)

Karl modified cvs2svn.py so that it now produces a dumpfile, rather
than committing directly into an svn repository.  This is great for
two reasons:

  1. a dumpfile is super-flexible; you can carry it around, filter it,
     and load it whereever you want (such as appending it to an
     existing repository)

  2. no more swig/python dependencies.  This means the bar for cvs2svn
     conversion has been drastically lowered for many users.  :-)

Yay Karl!


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by Garrett Rooney <ro...@electricjellyfish.net>.
On Saturday, May 10, 2003, at 11:51 PM, kfogel@collab.net wrote:

>    $ cvsup -s supfile
>
> which gave me this error in the CVSup window (after I hit the green,
> uh, "play" button):
>
>    Nonexistent prefix "/pub/FreeBSD/development/FreeBSD-CVS" for 
> cvs-all/cvs
>    No collections selected
>
> Hmmm.  Anything wrong with this supfile?
>
>    #
>    # FreeBSD CVS supfile from master server
>    #
>    *default host=cvsup-master.FreeBSD.org
>    *default base=/usr
>    *default prefix=/pub/FreeBSD/development/FreeBSD-CVS
>    *default release=cvs
>    *default delete use-rel-suffix
>    *default umask=002
>
>    # If your network link is a T1 or faster, comment out the following 
> line.
>    #*default compress
>
>    cvs-all

I believe that cvsup-master is only supposed to be used by other cvsup 
mirrors.  You might want to try another server.

-garrett


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
Garrett Rooney <ro...@electricjellyfish.net> writes:
> I believe Robert said he saw this while converting a copy of the
> FreeBSD CVS repository.  You can get a copy via CVSup, see the
> instructions here:
> 
> http://www.freebsd.org/doc/en_US.ISO8859-1/articles/hubs/mirror-
> howto.html
> 
> (Look for the section on Mirroring the CVS Repository)

Thanks.  If Robert drops off the face of the earth, I'll try that --
but if he's around, it'll be a lot more efficient for him to tell me
exactly how to get the same data he has (rather than me flailing
around).

I tried the CVSup instructions for mirroring the CVS repository, as
given on that page, and using their example supfile:

   $ cvsup -s supfile

which gave me this error in the CVSup window (after I hit the green,
uh, "play" button):

   Nonexistent prefix "/pub/FreeBSD/development/FreeBSD-CVS" for cvs-all/cvs
   No collections selected

Hmmm.  Anything wrong with this supfile?

   #
   # FreeBSD CVS supfile from master server
   #
   *default host=cvsup-master.FreeBSD.org
   *default base=/usr
   *default prefix=/pub/FreeBSD/development/FreeBSD-CVS
   *default release=cvs
   *default delete use-rel-suffix
   *default umask=002
   
   # If your network link is a T1 or faster, comment out the following line.
   #*default compress
   
   cvs-all

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by Garrett Rooney <ro...@electricjellyfish.net>.
On Saturday, May 10, 2003, at 11:17 PM, kfogel@collab.net wrote:

> Robert Pluim <rp...@bigfoot.com> writes:
>> I have very little python-fu, but this appears to mean that
>> self.db[parent_dir_key] doesn't exist.  What information can I get you
>> to help debug this?
>
> This is probably the same error that David just reported.  Can you
> make your CVS repository available to me?
>

I believe Robert said he saw this while converting a copy of the  
FreeBSD CVS repository.  You can get a copy via CVSup, see the  
instructions here:

http://www.freebsd.org/doc/en_US.ISO8859-1/articles/hubs/mirror- 
howto.html

(Look for the section on Mirroring the CVS Repository)

-garrett


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
Robert Pluim <rp...@bigfoot.com> writes:
> I have very little python-fu, but this appears to mean that
> self.db[parent_dir_key] doesn't exist.  What information can I get you
> to help debug this?

This is probably the same error that David just reported.  Can you
make your CVS repository available to me?

-K


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by Robert Pluim <rp...@bigfoot.com>.
> Please post if you encounter errors or misconversions, folks.

I just tried converting over a copy of the FreeBSD CVS repository, and
got the following in pass 4 :

    adding or changing 1.1 : trunk/src/lib/libc/locale/Makefile.inc
Traceback (most recent call last):
  File "./cvs2svn/cvs2svn.py", line 1161, in ?
    main()
  File "./cvs2svn/cvs2svn.py", line 1158, in main
    convert(ctx, start_pass=start_pass)
  File "./cvs2svn/cvs2svn.py", line 1077, in convert
    _passes[i](ctx)
  File "./cvs2svn/cvs2svn.py", line 1026, in pass4
    c.commit(dump, ctx)
  File "./cvs2svn/cvs2svn.py", line 707, in commit
    dump.add_or_change_path(cvs_path, svn_path, cvs_rev, rcs_file)
  File "./cvs2svn/cvs2svn.py", line 531, in add_or_change_path
    if self.head_mirror.ensure_path(svn_path, self.add_dir):
  File "./cvs2svn/cvs2svn.py", line 315, in ensure_path
    parent_dir = marshal.loads(self.db[parent_dir_key])
KeyError: aa9

I have very little python-fu, but this appears to mean that
self.db[parent_dir_key] doesn't exist.  What information can I get you
to help debug this?

Robert

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
David Summers <da...@summersoft.fay.ar.us> writes:
> committing: Sun Feb 24 20:39:03 2002, over 0 seconds
>     deleting 1.1 : trunk/src/student_station/ssvega/app/makefile_other

Hmmm!

It's *deleting* revision 1.1 of the file?  That's odd.

Yes, maybe that one RCS file would help.  Can you send it to me
privately, David?

-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
David Summers <da...@summersoft.fay.ar.us> writes:
> Ack, sorry, I can't.  Proprietary.  Also very big (Over 1.5 GB).  Someone 
> else just posted a similar situation.   If their repository conversion can be
> debugged then maybe that will help on mine also.  

Probably it's the same error, so let's hope Robert's repository can be
made available.

> Otherwise is there any other way to debug rather than give you access?
> 
> Does the error message point to one file?  If so, I could probably send 
> you that file or look at it for wierdness if you specified what was 
> "weird" and what to look for.

Well, I guess it's talking about `makefile_other,v', but it might be
hard to debug without the rest of the history.  Let's see if Robert
Pluim can let me use his repository; if not, then we'll try just that
one file from you.

(If there turns out to be no other way to reproduce this, I'm willing
to sign an NDA to see the data, too.  Would that fly?)

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by David Summers <da...@summersoft.fay.ar.us>.

On 10 May 2003 kfogel@collab.net wrote:

> David Summers <da...@summersoft.fay.ar.us> writes:
> > I'm test converting a CVS repository where the conversion has previously 
> > worked using ealier versions of cvs2svn (SWIG).  Today it errors out with:
> 
> Thank you for the report!
> 
> >   File "/usr/bin/cvs2svn", line 588, in delete_path
> >     deleted_path = self.head_mirror.delete_path(svn_path, 1) # 1 means 
> > prune
> >   File "/usr/bin/cvs2svn", line 392, in delete_path
> >     self._delete_tree(parent_dir[basename])
> > KeyError: makefile_other
> > 
> > 
> > What do I need to do to help debug this?
> 
> Send me your CVS repository (or post it somewhere), and I'll run the
> conversion and see what's causing the error.  I know what code that
> error is in, but it's supposed to be a "can't happen" situation.
> Obviously, it can happen, though... :-)
> 

Ack, sorry, I can't.  Proprietary.  Also very big (Over 1.5 GB).  Someone 
else just posted a similar situation.   If their repository conversion can be
debugged then maybe that will help on mine also.  

Otherwise is there any other way to debug rather than give you access?

Does the error message point to one file?  If so, I could probably send 
you that file or look at it for wierdness if you specified what was 
"weird" and what to look for.

-- 
David Wayne Summers          "Linux: Because reboots are for hardware upgrades!"
david@summersoft.fay.ar.us   PGP Key: http://summersoft.fay.ar.us/~david/pgp.txt
PGP Key fingerprint =  C0 E0 4F 50 DD A9 B6 2B  60 A1 31 7E D2 28 6D A8 



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
David Summers <da...@summersoft.fay.ar.us> writes:
> I'm test converting a CVS repository where the conversion has previously 
> worked using ealier versions of cvs2svn (SWIG).  Today it errors out with:

Thank you for the report!

>   File "/usr/bin/cvs2svn", line 588, in delete_path
>     deleted_path = self.head_mirror.delete_path(svn_path, 1) # 1 means 
> prune
>   File "/usr/bin/cvs2svn", line 392, in delete_path
>     self._delete_tree(parent_dir[basename])
> KeyError: makefile_other
> 
> 
> What do I need to do to help debug this?

Send me your CVS repository (or post it somewhere), and I'll run the
conversion and see what's causing the error.  I know what code that
error is in, but it's supposed to be a "can't happen" situation.
Obviously, it can happen, though... :-)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by David Summers <da...@summersoft.fay.ar.us>.
On 10 May 2003 kfogel@collab.net wrote:

> Please post if you encounter errors or misconversions, folks.
> 
> -K
> 

OK.

I'm test converting a CVS repository where the conversion has previously 
worked using ealier versions of cvs2svn (SWIG).  Today it errors out with:

committing: Sun Feb 24 20:09:16 2002, over 0 seconds
    deleting 1.2 : trunk/src/aclocal.m4
    (deleted trunk/src/aclocal.m4)
    deleting 1.2 : trunk/src/install-sh
    (deleted trunk/src/install-sh)
    deleting 1.2 : trunk/src/missing
    (deleted trunk/src/missing)
    deleting 1.2 : trunk/src/mkinstalldirs
    (deleted trunk/src/mkinstalldirs)
    new revision: 179
committing: Sun Feb 24 20:39:03 2002, over 0 seconds
    deleting 1.1 : trunk/src/student_station/ssvega/app/makefile_other
Traceback (most recent call last):
  File "/usr/bin/cvs2svn", line 1160, in ?
    main()
  File "/usr/bin/cvs2svn", line 1157, in main
    convert(ctx, start_pass=start_pass)
  File "/usr/bin/cvs2svn", line 1076, in convert
    _passes[i](ctx)
  File "/usr/bin/cvs2svn", line 1025, in pass4
    c.commit(dump, ctx)
  File "/usr/bin/cvs2svn", line 713, in commit
    dump.delete_path(svn_path)
  File "/usr/bin/cvs2svn", line 588, in delete_path
    deleted_path = self.head_mirror.delete_path(svn_path, 1) # 1 means 
prune
  File "/usr/bin/cvs2svn", line 392, in delete_path
    self._delete_tree(parent_dir[basename])
KeyError: makefile_other


What do I need to do to help debug this?

   Thanks!

-- 
David Wayne Summers          "Linux: Because reboots are for hardware upgrades!"
david@summersoft.fay.ar.us   PGP Key: http://summersoft.fay.ar.us/~david/pgp.txt
PGP Key fingerprint =  C0 E0 4F 50 DD A9 B6 2B  60 A1 31 7E D2 28 6D A8 



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
Garrett Rooney <ro...@electricjellyfish.net> writes:
> One thing that we should probably work on is the fact that doing an
> svnadmin load on a large dumpfile produces a LOT of logfiles from
> berkeley db, and it's quite easy to run out of disk if you're not
> careful to do a db_archive -l | xargs rm every now and then while it's
> going.  Perhaps we should install a temporary post-commit hook to do
> that during the conversion or something?

Yes, that happened to me too.

I was sort of hoping Berkeley would come through with their
automatic-logfile-removing release before we release 1.0, but maybe it
wouldn't be wise to depend on that...

> (Note that I imagine this is not a new problem, the old cvs2svn would
> have had the same issue AFAICT.)

I think so, yeah.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by Garrett Rooney <ro...@electricjellyfish.net>.
On Saturday, May 10, 2003, at 03:19 PM, kfogel@collab.net wrote:

> Ben Collins-Sussman <su...@collab.net> writes:
>> This is actually a really cool new feature going out in 0.22.1.  I
>> feel like Karl hasn't bragged enough.  :-)
>
> Heh!  Thanks, Ben.  But I'm not going to brag until the branch/tag
> conversion is working too.  Also, people should be aware that cvs2svn
> is slower now, since it has that intermediate dumpfile stage.  I think
> the tradeoff is well worth it, but in fairness the slowdown ought to
> be mentioned, anyway.

I just tried out cvs2svn for the first time last night, and I must say 
I'm impressed.  It converted the linux kernel 2.4 cvs repository 
(there's an exported cvs repository based on the bitkeeper repository 
that you can rsync) to svn in about 4 and a half hours.

One thing that we should probably work on is the fact that doing an 
svnadmin load on a large dumpfile produces a LOT of logfiles from 
berkeley db, and it's quite easy to run out of disk if you're not 
careful to do a db_archive -l | xargs rm every now and then while it's 
going.  Perhaps we should install a temporary post-commit hook to do 
that during the conversion or something?

(Note that I imagine this is not a new problem, the old cvs2svn would 
have had the same issue AFAICT.)

-garrett


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5872 - branches/release-0.22.1

Posted by kf...@collab.net.
Ben Collins-Sussman <su...@collab.net> writes:
> This is actually a really cool new feature going out in 0.22.1.  I
> feel like Karl hasn't bragged enough.  :-)

Heh!  Thanks, Ben.  But I'm not going to brag until the branch/tag
conversion is working too.  Also, people should be aware that cvs2svn
is slower now, since it has that intermediate dumpfile stage.  I think
the tradeoff is well worth it, but in fairness the slowdown ought to
be mentioned, anyway.

Please post if you encounter errors or misconversions, folks.

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org