test failure on import for xrms/trunk

ddaa has added more details here:

https://help.launchpad.net/VcsImportRequests
(search on xrms)

and here
https://macquarie.warthogs.hbd.com/roomba/status/xrms-trunk/events/6/log
(protected link may not be accessible)

discussion on #launchpad on 2006-10-03 regarding this:

(08:08:11) ddaa: braverock: look at https://help.launchpad.net/VcsImportRequests
(08:08:20) ddaa: I have written down the status of xrms
(08:09:09) ddaa: it looks like the default branch setting of of xrms/include/adobd/pear/readme.Auth.txt confuses our import tool
(08:09:17) ddaa: maybe other files have the same problem
(08:09:56) ddaa: braverock: I do not think that bug is going to have any attention soon, we are focusing of svn support now
(08:10:24) braverock: ok, well did I do something wrong, or is this a bazaar bug?
(08:10:41) braverock: I believe that the repository root is correct
(08:11:03) ddaa: neither
(08:11:12) ddaa: it's a bug in cscvs
(08:11:34) ddaa: the tool we use to magically create changesets out of CVS repository, and apply them
(08:12:08) braverock: ok. can I change anything on the " the default branch setting of of xrms/include/adobd/pear/readme.Auth.txt" to unconfuse your tool?
(08:12:34) ddaa: braverock: if you have write access to the cvs repo, you could fix it so no file in has a default branch set
(08:12:40) braverock: and could you consider maybe adding a link to the full import report to the "import Failed" error?
(08:12:54) ddaa: we do plan to do that
(08:13:03) ddaa: the wiki page is a stopgap measure
(08:13:11) ddaa: I maintain manually.
(08:13:23) matsubara: kiko: just added to the list
(08:13:37) braverock: I don't think we've set branches on anything, but I'll check
(08:14:59) braverock: someone else registered xrms in launchpad, and then abandoned it. I picked it up, and hoped to get it in, because broader exposure on ubuntu would be good for xrms (even though xrms is already a very popular project)
(08:15:19) ddaa: braverock: when you do "cvs -d :pserver:<email address hidden>:/cvsroot/xrms/ rlog xrms/include/adodb/pear/readme.Auth.txt you should see a "branch" line in the heading (before individual revisions), it's that setting that is causing us trouble
(08:15:53) ddaa: I do not know cvs very much, so I cannot help you much more, though.
(08:16:12) braverock: unfortunately, this is an Attic file
(08:16:23) ddaa: ?
(08:16:26) braverock: so it's already been deleted from the repository
(08:16:40) braverock: perhaps you could ignore all Attic files
(08:16:51) ddaa: what makes you say it's an attic file?
(08:17:06) braverock: xrms/include/adodb/pear/Attic/readme.Auth.txt
(08:17:17) braverock: from your error
(08:17:19) ddaa: the error message reads Attic because cscvs tries to look up file there if there are not found at the expected location
(08:17:34) braverock: oh...
(08:17:37) ddaa: it's not an attic file, it's present when you do a checkout
(08:17:51) lifeless: ddaa: cvs import creates a default branch
(08:17:54) lifeless: fwiw
(08:18:03) ddaa: lifeless: look at the error
(08:18:21) ddaa: I looks like the default branch was reset after the initial import
(08:18:29) braverock: yes, adodb is a database abstraction library, and was probably created using cvs import
(08:18:30) ddaa: though I'm not too sure about it
(08:18:37) lifeless: ddaa: uel to the error ?
(08:19:07) ddaa: https://macquarie.warthogs.hbd.com/roomba/status/xrms-trunk/events/6/log
(08:19:24) ddaa: look at https://help.launchpad.net/VcsImportRequests for some diagnostic info
(08:20:28) ddaa: braverock: lifeless is the guy who wrote the default branch support code. I leave you in his capable hands
(08:20:34) ***ddaa -> back to lunch etc.
(08:20:57) braverock: thanks
(08:21:49) braverock: lifeless: I'd be happy to remove the default branch tag there too, if you've got an idea on the cvs command to do that with
(08:22:54) lifeless: ddaa: Attic should not be in the request line
(08:23:12) ddaa: lifeless: it does fix some imports
(08:23:20) lifeless: ddaa: for this specific one
(08:23:30) ddaa: right, but it's a fallback
(08:23:46) ddaa: first it try in the normal location, then if it does not find the file, it looks into the attic
(08:23:47) lifeless: so heres the glitch: there is no revision 1.1.1.2
(08:23:54) lifeless: the branch 1.1.1 is fine
(08:24:09) ddaa: and if the attic lookup fail, the error bubble up
(08:24:21) ddaa: so it's entirely an error reporting issue, not a logic issue
(08:24:28) lifeless: ddaa: sure. but look closer : rlog of xrms/include/adodb/pear/readme.Auth.txt
(08:24:36) lifeless: there is /no/ 1.1.1.2
(08:24:37) ddaa: lifeless: I looked
(08:24:46) ddaa: I know the revision is not there
(08:25:04) ddaa: but I have no clue why cscvs tries to make a filler cs 1.1->1.1.1.2
(08:25:20) lifeless: is that what its trying to do ?
(08:25:38) lifeless: whats in changeset MAIN.4167
(08:25:51) ddaa: yes, It's explained in https://help.launchpad.net/VcsImportRequests
(08:26:17) ddaa: MAIN.4167 is a filler cs 1.1->1.1.1.2
(08:27:37) lifeless: so fillers appear when the branch changes
(08:28:25) lifeless: I'd do a search in the file revision table for that file, all revisions, see what cscvs is thinking happened to it
(08:28:43) ddaa: that's more than I know about cscvs
(08:28:55) ddaa: but thanks for the advice
(08:29:45) ddaa: you suggest it might be a problem in the log parser and checking the catalog contents would help narrowing the cause of the problem?
(08:33:11) lifeless: something is inventing the .2
(08:33:23) lifeless: if its in the catalog then the catalog generator can be checked
(08:33:42) lifeless: if its not in the catalog then the application-of-changesets can be checked
(08:33:46) lifeless: it will narrow it down
(08:34:03) lifeless: As its a very small rlog I'd make a test case from it

This support request was expired because it remained in the 'Open' state without activity for the last 15 days.

This is still an open issue.

Followed up with an email giving directions to debug cscvs.

http://thread.gmane.org/gmane.comp.version-control.bazaar-ng.general/20338/focus=20338

Please ask on the mailing if you need more help fixing the bug.

Brian,

If you want to do a one-off conversion to evaluate or to switch to Bazaar, it might be worth trying this

  https://launchpad.net/products/bzr-cvsps-import

You can ask here in the support tracker or on the list if you have any problems. cc me.

This is a plugin you can install into Bazaar to read in a cvs branch. Unlike the launchpad service, this does a one-off conversion, not an ongoing mirror.

Just for the record:

At the time Brian filed this support requet, I had looked at the problem, and told him I would try to work on it soon, but was not sure I would be able to.

A couple of weeks later, he came back and asked if there was any progress. Unfortunately I had not been able to fix the problem, but I informed him the code had been released and that he could try fixing the problem if he wanted to.

I also asked him if he preferred a one-off conversion, but since xrms is not going to switch to bzr in the forseeable future, an mirror branch is needed.