Julian Pullen <Julian.Pullen at Sun.COM> writes: > I have had an intermittent hg recommit error. > The problem has gone away without any intervention > I don't know what can cause it but here is a log. > There seamed to be a mismatch between the Mercurial > and Cadimum view of the changes. > > Can you please reply directly to me as I am not in the > tools-discuss alias. >
I'm going to need more detail than this, I'm afraid. It may actually require me seeing a copy of the workspace, if that is possible (I am not an SMI employee, I am, however, the person who wrote the recommit code in cdm, so if the code is closed or not yet public that won't be possible) I've seen one report of this before, but that user (like you, seemingly) had the problem vanish of its own accord. There is one case where something like this can happen due to a lack of verbosity in saying what we mean. If you have files in state '!' from hg status (deleted, not 'hg removed'), cdm will view the workspace as modified for the purposes of these sanity checks, but Hg will not for the purposes of 'commit' (as the change made to them is not versionable). It's unclear from your output whether that happened in your case, as you say, the modified files seem to cease being modified of their own accord, in the output you presented, which seems unlikely. -- Rich
