corrupt local repository

[Pierre-Yves David]

That's very odd, it seems like a fresh corruption of a very new 
repository I have head of that in a very long time.

I see you are using a rc version, can you move the latest official version ?

Is there anything else strange that happened like a power failure on the 
machine or something similar ? Any process being abrutly killed ? (not 
that this point should matters ?)

On 6/10/20 5:55 PM, Uwe Brauer wrote:
>>>> "PD" == Pierre-Yves David <pierre-yves.david at ens-lyon.org> writes:
> 
>     > Is this as new repostiory that you never pushed anywhere ?
> 
> Never ever.
> 
>     > What is the underlying file system ?
> 
> Ext4 (Ubuntu 16.04)
> 
>     > Do you have any build/cleanup script running as part of your
>     > development ? possibly deleting file automatically ?
> 
> Nope,
>     > On 6/8/20 8:42 PM, Uwe Brauer wrote:
>     >>>>> "PD" == Pierre-Yves David <pierre-yves.david at ens-lyon.org> writes:
>     >> > Sounds like you have at least two different files whose data
>     >> are
>     >> > missing. How have you been using your repository ? is the repository
>     >> > server side corrupted ?
>     >> I create this repository this morning and just committed around 10
>     >> times.
>     >> However I did a lot of renaming  of directories and sometimes I
>     >> committed without having run hg addremove before, may that is the
>     >> reason.
> 
>     > That should not matter. Do you use symlink in the mix ?
> 
> No
> 
> 
>     >> I also found out that
>     >> hg convert --config convert.hg.ignoreerrors=True Corr/ Repair
>     >> Is basically worseless, the resulting repo contains only half of my
>     >> commits.
> 
>     > Well, it cannot "invent" the missing content, so I guess it simply
>     > dropped the associated commit
> 
> So it seems that this repository cannot be recovered.
> 
> I am running
> hg version 5.2rc0+20200125
> 
> And the last time I suffered a repository corruption is I think with
> 3.8.
> 
> But it seems difficult to track down and to know how to avoid this in
> the feature.  I still think it has to do with the amount of renaming I
> did, but I have no proof and no recipe.
> 
> 
> _______________________________________________
> Mercurial mailing list
> Mercurial at mercurial-scm.org
> https://www.mercurial-scm.org/mailman/listinfo/mercurial
> 

-- 
Pierre-Yves David