[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: disturbing, weird, reproducible repository error

From: John Szakmeister <john_at_szakmeister.net>
Date: 2003-09-09 13:24:14 CEST

On Monday 08 September 2003 22:35, Ben Collins-Sussman wrote:
> On IRC, housel managed to create a reproducible repository error.
> I can't figure out if it's a dumper/loader bug, or a bug in our new
> 'log' code in the fs, or what.
[snip]
> svn: file not found: revision `78', path
> `/branches/GD-2-1-callbacks/src/tools/shared-misc/make-dylan-app.in'
>
> svn: Filesystem path
> `branches/GD-2-1-callbacks/src/d2c/runtime/gc/cord/private/cord_pos.h' is
> neither a file nor a directory
>
> When I 'svn ls' r78 of branches/GD-2-1-callbacks/src/tools/, I see no
> 'shared-misc' directory at all. And yet the output of 'svnadmin load'
> says it was added. And r78 in the dumpfile looks correct to me too.

Here is something I find a bit disturbing. On line 357233, we can see
branches/GD-2-1-callbacks/src/tools/shared-misc being added. At line 357312,
I see that very same node being deleted. The catch? It's all occuring in
rev 78. I can also see in the output of 'svnadmin load' that indeed the path
is being created and destroyed in the same rev. My question is how does this
happen?

As for the other error, I see the node for
branches/GD-2-1-callbacks/src/d2c/runtime/gc/cord/private/cord_pos.h created
at line 357288, but the directory
branches/GD-2-1-callbacks/src/d2c/runtime/gc/cord/private is removed at line
357299. This strikes me as odd for two reasons:
 1) This happens in one transaction.
 2) Because however this occured, it didn't go through an schedule deletion
for everything in the directory. I was under the impression that if a
directory was deleted then all of it's contents would be scheduled for
deletion too.

> Can anyone else reproduce, or have any insight? Apparently this
> dumpfile was created by cvs2svn. housel truncated it at r79 for me,
> because r78 was the 'troublesome' revision. The original dumpfile was
> much, much larger, but r78 was discovered as 'bad' long after it had
> been loaded.

Hopefully, the above might bring some insight into the problem. :-)

> I anxiously await cmpilato's return. :-)

Me too!

-John

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Sep 9 13:23:24 2003

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.