Re: [Monotone-devel] Is this the right mailing list? [Was: upgrading 0.36-0.37: "mtn: fatal: std::logic_error: roster.cc:186: invariant 'fetching nonexistent entry from children' violated"]

[Arthur A. Gleckler]

On 12/6/07, Zbigniew Zagórski <address@hidden> wrote:

> Yes this is correct mailing list, your problem is noted.
> However nobody currently has time/idea how to solve your problem.
> Sorry. Someone surely is investigating your problem silently ...

Sorry, I just wanted to make sure that someone was going to see it
eventually, and that I wasn't supposed to do something like file a bug
report.  Thanks.

> Only one thing I can propose to you is to go back to 0.36. There was
> no database schema changes so this step back should be simple.

Alas, I tried that, but I still get the same error.  I'm afraid that
something has now corrupted my database for that branch.

> You can also check if previous revisions can be checked out/updated
> to; it would surely tell if it's only one rev broken or there is
> something more.

That's a great idea.  Here's what I've now tried, again with 0.37:

* Created a new database with <mtn db init>.

* Sync-ed to the database containing the problematic branch.

* Ran <mtn checkout>, specifying a revision before the problems occurred.

Here is the result:

  mtn: misuse: rename target
'lisp/personal-site-lisp/jde/java/src/jde/debugger/Jdebug.java'
already exists

After that, I started over in that directory and tried the same steps
with 0.36.  I got the same results.  I'm confused, because I hadn't
done anything to the file that it complains about for many, many
revisions before that, but 0.36 had never complained before.  That
seems to indicate that the history has been corrupted even before the
version where I upgraded.

Thanks very much!