One extra piece of data: Jane's 1.4 content appears to be identical to the recreated 1.3 content. This may or may not be a red herring, but there is a new log message, though I have no current proof that Jane's 1.4 was not originally a force-commit without any changes.
Unfortunately, none of this gets me much closer to an answer. I want to suspect locking issues, but if it was a locking issue, it seems extremely unlikely that the data would line up so exactly. Regards, Derek Derek R. Price wrote: > Has anyone ever seen RCS archive corruption in the form of two head > revisions or have any idea how it may have come about? > > It appears to have happened as part of a commit. The head revision was > initially 1.4. CVS overwrote the metadata entry for 1.4 (commit time, > author, state, branches, next), but put a second, undiffed entry for the > head revision after the first. So, before the commit, the file probably > looked like: > > RCSHEADER > > 1.4 > date YYYY.MM.DD.hh.mm.ss; author janedoe; state Exp; > branches; > next 1.3; > > [REVISION METADATA 1.3-1.1) > > desc > @@ > > 1.4 > log > @what jane did > @ > text > @jane's content > @ > > 1.3 > log > @what's gone before > @ > text > @a diff from jane's content > @ > > ... > > > > Then, after the (otherwise successful) commit: > > RCSHEADER > > 1.4 > date YYYY.MM.DD.hh.mm.ss; author joeschmoe; state Exp; > branches; > next 1.3; > > [REVISION METADATA 1.3-1.1) > > desc > @@ > > 1.4 > log > @what jane did > @ > text > @jane's content > @ > > 1.4 > log > @what joe did > @ > text > @joe's content (based on jane's content, but not the diff from it) > @ > > 1.3 > log > @what's gone before > @ > text > @a diff from joe's content > @ > > ... > > > > I know for a fact that the final 1.3 text is a diff from joe's content, > because if I reconstruct the file minus joe's head revision, I get an > "invalid change text" error checking out 1.3, but if I reconstruct it > minus jane's head revision, revision 1.3 will check out. > > Using the two head revisions, I cannot reproduce this problem via a > simple commit. > > Regards, > > Derek -- Derek R. Price CVS Solutions Architect Ximbiot <http://ximbiot.com> v: +1 248.835.1260 f: +1 248.835.1263 <[EMAIL PROTECTED]> _______________________________________________ Bug-cvs mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/bug-cvs
