Setting to need-eg until we have a way to reproduce it

As I said:
there is only one repo. I served it using hg serve

Hudson: https://hudson.dev.java.net/
I think it does: 
1. Find out the changesets that have changed since hudson last checked out.
2. Request those changesets
3. Build the program

I just found out: I sent the wrong repo. The repo with a problem is the copy
hudson uses. I misinterpreted hudsons output...
Upon hg verify that repo says:
abort: index data/tcpfile/eventbus/.classpath.i unknown format 19823!

I attached that file. It looks nothing like the one from the original repo,
which has binary data in it...

Sorry for this bug report, it's been a bit messy and it could be the file got
messed up outside of mercurials influence (I am using windows after all, and it
may have crashed).

What is hudson?

If you do hg in by hand, does it work?
Can you make hg in work locally on the server (not going over the network)?

Can you save the incoming bundle ?
So that we would be able to compare to a working bundle.

Well whenever hudson tries to get the changeset from the server the same thing
happens, so locally, I can reproduce it 100% of the time. Not by hand, but by
starting hudson.

If hg verify doesn't have any problems, I don't think we can do anything. Unless
it's reproducible.

I just noticed it is a snapshot version
Mercurial Distributed SCM (version 1.0.1+20080525)

I usually dont install snapshots. Could that explain it?
There is no version field in this bugtracker, otherwise I might have noticed...

It just happened with this one repo, I do not know what steps exactly I did...

But I just noticed, I dont have to zip everything, I can zip just the .hg folder
and then revert the repo. That is only about 2mb :)
Attached it

Then I can't see anything else than an external corruption (network, or
something else) :/

Can you reproduce it ? (if yes a bundle might help and would be smaller than a
full repo)

there is only one repo. I served it using hg serve

both remote and local ?

F:\dev\tcpfile.bak>hg verify
checking changesets
checking manifests
crosschecking files in changesets and manifests
checking files
791 files, 19 changesets, 870 total revisions

seems so

does it pass verify ? (both repos)

I used mercurial to track my progress on a little project. 
But a few commits ago, when trying to build my repository I got this error from
hudson:

started
[workspace] $ hg incoming --quiet --bundle hg.bundle --debug --template
"<changeset node='{node}' author='{author|xmlescape}' rev='{rev}'
date='{date}'><msg>{desc|xmlescape}</msg><added>{file_adds|stringify|xmlescape}</added><deleted>{file_dels|stringify|xmlescape}</deleted><files>{files|stringify|xmlescape}</files><parents>{parents}</parents></changeset>\n"
using http://localhost:45821/
comparing with http://localhost:45821
sending heads command
searching for changes
sending branches command
examining 539ae20afe5a:10367cbcfc3d
found incomplete branch 539ae20afe5a:10367cbcfc3d
sending between command
narrowing 1:6 39bd2dde678c
found new branch changeset 539ae20afe5a
found new changesets starting at 539ae20afe5a
1 total queries
sending changegroup command
** unknown exception encountered, details follow
** report bug details to http://www.selenic.com/mercurial/bts
** or mercurial@selenic.com
** Mercurial Distributed SCM (version 1.0.1+20080525)
Traceback (most recent call last):
  File "hg", line 20, in <module>
  File "mercurial\dispatch.pyc", line 20, in run
  File "mercurial\dispatch.pyc", line 29, in dispatch
  File "mercurial\dispatch.pyc", line 45, in _runcatch
  File "mercurial\dispatch.pyc", line 364, in _dispatch
  File "mercurial\dispatch.pyc", line 417, in _runcommand
  File "mercurial\dispatch.pyc", line 373, in checkargs
  File "mercurial\dispatch.pyc", line 356, in <lambda>
  File "mercurial\commands.pyc", line 1636, in incoming
  File "mercurial\cmdutil.pyc", line 588, in show
  File "mercurial\cmdutil.pyc", line 864, in _show
  File "mercurial\templater.pyc", line 148, in stringify
  File "mercurial\templater.pyc", line 114, in __call__
  File "mercurial\cmdutil.pyc", line 810, in showadds
  File "mercurial\cmdutil.pyc", line 803, in getfiles
  File "mercurial\localrepo.pyc", line 1004, in status
  File "mercurial\localrepo.pyc", line 987, in mfmatches
  File "mercurial\manifest.pyc", line 59, in read
  File "mercurial\bundlerepo.pyc", line 114, in revision
  File "mercurial\revlog.pyc", line 951, in revision
  File "mercurial\bundlerepo.pyc", line 80, in chunk
  File "mercurial\revlog.pyc", line 904, in chunk
  File "mercurial\revlog.pyc", line 102, in decompress
zlib.error: Error -3 while decompressing data: invalid distance too far back
ERROR: Failed to determine incoming changes
finished: FAILURE

I am not the only person who ever had this problem:
http://article.gmane.org/gmane.comp.version-control.mercurial.general/1497

Maybe very rare...
The repo seems functional, I can commit, rollback etc but when hudson is trying
to build from a certain changeset on it fails. I could send you the repo, but it
being 30mbyte (7zipped) I would rather not :)