subversion-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andy Levy <>
Subject Re: Vedr: revision number offset
Date Thu, 20 Oct 2011 23:59:48 GMT
On Thu, Oct 20, 2011 at 11:14, morten bjoernsvik
<> wrote:
> Thanks Andy
> Most of the servers I work on do not have internet access, only citrix/cisco
> vpn via
> putty and openscp. So I can only scp a file back and forth, but no ssh+svn.

I'm puzzled by this. SCP uses the same port(s) as SSH - since svn+ssh
tunnels svn *through* SSH, I would expect that to work in your setup.
What am I missing?

> Git is fine, but the support for it in komodo(our IDE) is poor and we have
> beeen using svn for years.

Have you looked at Komodo's Mercurial support? Just because it doesn't
work well w/ git doesn't mean hg will be similar.

> I also like to clean up repositores from time to time, aka
> a svnadmin -rXX:YY dump then delete and load to only save the latest 10
> revisions and drop the rest.
> But if others see the revison-number being lower they believe it to be older
> code.

That's a user education issue combined with a very non-standard
process you're using. And I can definitely see why users would get
confused - it goes against most of the conventions of using

> If only the revision number was kept this would have worked.
> Thanks
> --
> MortenB
> ________________________________
> Fra: Andy Levy <>
> Til: morten bjoernsvik <>
> Kopi: "" <>
> Sendt: Torsdag, 20. oktober 2011 16.26
> Emne: Re: revision number offset
> On Thu, Oct 20, 2011 at 10:08, morten bjoernsvik
> <> wrote:
>> Hi All
>> I'm a very pleased svn user, but I have this little issue.
>> In many cases with small projects I move the repository around
>> (like we have been used to with git)
>> svnadmin dump <repository> | bzip2 > repository_revXX.svndump.bz2
>> and then:
>> svnadmin create <parth to repository>
>> bzip2 -dc repository_revXX.svndump.bz2 | svnadmin load <path to
>> repository>
>> This works fine up to around 100-150 revisions, then it becomes slow and I
>> delete the old repository and start a completely new with the latest
>> revisions.
>> But I'm unable to keep the revision number, I like my new commit to start
>> with
>> rev XX+1 not 1. Why cant I start on any revision number I like?
> Because the revision number is a running count of the number of
> changes committed to the repository. When you create a new, empty
> repository you start all over again.
>> Even if I use the -rXX for dump it is converted to start on 1 when I load.
>> The only solution I've seen is to add XX empty revisions. But this does
>> not
>> look good
>> when revisison becomes very high.
> That is the only way to force Subversion to "start" your revision
> numbers at something other than 0.
>> Is there a solution or workaround?
> The solution is to not keep moving your repository around they way you
> are. Why do you feel you need to do this? Are svnsync mirrors not
> sufficient for your usage (if they'd even be applicable to your use
> case)? Perhaps a DVCS like hg or git *is* more appropriate for you
> than Subversion?

View raw message