SVN repo

Stephen Montgomery-Smith stephen at missouri.edu
Tue Aug 9 19:22:13 UTC 2011


On 08/09/2011 02:07 PM, Chris Rees wrote:
> On 9 August 2011 19:17, Stephen Montgomery-Smith<stephen at missouri.edu>  wrote:
>> On 08/09/2011 01:03 PM, Chris Rees wrote:
>>>
>>> On 9 August 2011 17:17, Stephen Montgomery-Smith<stephen at missouri.edu>
>>>   wrote:
>>>>
>>>> Chris Rees wrote:
>>>>>
>>>>> Hi Stephen,
>>>>>
>>>>> Have you any plans to start distributing the subversion repo? I keep
>>>>> hearing little hints about the Other Two Trees being moved too... If
>>>>> you need any more hands on it I'll be happy to help.
>>>>>
>>>>> Chris
>>>>
>>>>
>>>> I hadn't thought about it, but it is a good idea.
>>>>
>>>> A good start would be for you to tell me some basic commands by which I
>>>> can
>>>> fetch the whole subversion tree, and how I can regularly update it.  I
>>>> know
>>>> almost nothing about svn.  (Also on cvs I have access to
>>>> cvsup-master.freebsd.org, and it would be nice to have the equivalent for
>>>> svn.)
>>>>
>>>> Setting it up so that CTM creates regular updates of the svn repository
>>>> will
>>>> then be very straightforward for me, as the infrastructure is essentially
>>>> all there.
>>>>
>>>> Also, if the svn repository is big (as big as the cvs repository), I
>>>> probably need to warn the mirrors of the big files that will arrive.
>>>
>>> Hahaha, good one. The svn repo (just for src) is enormous -- the seed
>>> file is 1GB alone.
>>
>> The cvs repos are about 1.2G (compressed).  There are about three or four
>> *xEmpty* files in in cvs-cur at any one time.  So adding another three or
>> four of these might only double the size of CTM.  I think it is doable.
>>
>> Tell me a command sequence for downloading the svn repo, and I can do a few
>> preliminary experiments to see how big it would all be.
>
> Although I brought this up, I'm not technically a svn user -- just
> interested in future compatibility.
>
> I'm signing off for today now I'm afraid, but the instructions are at
> [1], I'll get a command sequence tomorrow if it's complicated :)
>
> Chris
>
> [1] http://wiki.freebsd.org/SubversionPrimer#head-a9bf2ef502d986092afa9606a2619735b87ded60

Those instructions will do nicely.  I will do some preliminary trials, 
and see how well it fits with CTM.

One thing I see is that svn stores a lot of stuff in databases.  I 
wonder how well this will work with "diff" which is how CTM creates its 
deltas.  I presume that two databases with exactly the same contents may 
be represented by completely different files, because the data won't be 
stored in a linear fashion, but according to index files whose exact 
structure might be random (depending upon the order in which items are 
placed into the database).

As you say, it may well be incompatible with CTM.

Stephen


More information about the ctm-users mailing list