.hg folder contains no store folder.

[Steve Gardner]

On 2012-08-31 13:03, Adrian Buehlmann wrote:
> On 2012-08-31 11:58, Steve Gardner wrote:
>> We have a strange situation with one of our Mercurial repositories, 
>> and
>> I'd like to confirm if its likely to cause us any problems in 
>> future.
>>
>> The repository is being updated using push over hgweb.cgi and in the
>> past there were a number of problems with permissions errors, so I 
>> think
>> at some point all of the files under the hg folder were deleted and 
>> then
>> recreated via a push from a copy of the repository.
>>
>>
>> Most of our repositories contain a cache folder and a store folder
>> within the .hg folder, but now in this one there is a data folder
>> instead of the store folder.
>> Everything else seems to work as it should (the history is still
>> correct etc), but I would like to be certain that nothing will be 
>> broken
>> by this.
>
> This story seems to match with a repository having Mercurial's very
> first repository format.
>
> Just creating an empty ".hg" directory (instead of using 'hg init' or
> 'hg clone') leaves a valid Mercurial repository.
>
> I just tried this on my Windows 7 box:
>
>   $ hg version -q
>   Mercurial Distributed SCM (version 2.2.2+12-d1a0ac9b6b8a)
>   $ mkdir b
>   $ cd b
>   $ mkdir .hg
>   $ hg verify
>   repository uses revlog format 0
>   checking changesets
>   checking manifests
>   crosschecking files in changesets and manifests
>   checking files
>   0 files, 0 changesets, 0 total revisions
>
> As you can see, an empty .hg directory is a valid repository. But 
> it's
> in an ancient format. Continuing the experiment:
>
>   $ echo foo > bla.txt
>   $ hg add
>   adding bla.txt
>   $ hg ci -m1
>   $ cd .hg
>   $ dir
>    Volume in drive C has no label.
>    Volume Serial Number is F80E-0A52
>
>    Directory of C:\Users\adi\hgrepos\tests\b\.hg
>
>   31.08.2012  13:42    <DIR>          .
>   31.08.2012  13:42    <DIR>          ..
>   31.08.2012  13:42               104 00changelog.d
>   31.08.2012  13:42                76 00changelog.i
>   31.08.2012  13:42                50 00manifest.d
>   31.08.2012  13:42                76 00manifest.i
>   31.08.2012  13:42    <DIR>          cache
>   31.08.2012  13:42    <DIR>          data
>   31.08.2012  13:42                64 dirstate
>   31.08.2012  13:42                 1 last-message.txt
>   31.08.2012  13:42                43 phaseroots
>   31.08.2012  13:42                 2 thgstatus
>   31.08.2012  13:42               102 undo
>   31.08.2012  13:42                 0 undo.bookmarks
>   31.08.2012  13:42                 7 undo.branch
>   31.08.2012  13:42                 9 undo.desc
>   31.08.2012  13:42                64 undo.dirstate
>   31.08.2012  13:42                 0 undo.phaseroots
>                 14 File(s)            598 bytes
>                  4 Dir(s)  374'871'592'960 bytes free
>
> You will notice that there is no file named "requires" in the .hg
> directory. And thus no "store" subdirectory. Because that format [aka
> "layout"] didn't use a "store" subdirectory.
>
> See also http://mercurial.selenic.com/wiki/RequiresFile
>
> The "requires" file was introduced later. It is used to denote in 
> what
> format a repository is.
>
> Current Mercurial versions can read and write all repository formats
> ever released (up to that version you are using).
>
> If your repo passes "hg verify", then your data should be fine.
>
> If you want to profit from a newer repo format (which is 
> recommended),
> then you can clone it using the pull protocol.
>
>   $ hg clone --pull repoA repoB
>
> repoB will then be in the format that is "most recent" (for the 
> version
> of "hg" you are using). repoA can be in any old format (which 
> includes
> the very first one).

Adrian,

Thanks very much for a clear and detailed epxlanation.  That helps a 
lot.

Steve