D. Richard Hipp wrote:
> On Aug 11, 2009, at 11:48 AM, Andreas Kupries wrote:
> 
>> D. Richard Hipp wrote:
>>> An interesting data point:  Fossil can be about ten times more space
>>> efficient than CVS.
>>>
>>> We are in the process of converting a 10-year-old project from CVS to
>>> fossil.
>> And some more curiosity ...
>>
>>      What tool chain was/is used to perform this conversion ?
>>
>>      Something direct CVS -> fossil, or are you going through
>>      intermediaries, like SVN, Hg, etc. ?
>>
> 
> 
> I have a 193-line TCL script that uses both CVSTrac and CVS data to do  
> the conversion.  The script is not general-purpose, but is sufficient  
> for SQLite.

Ok. CVSTrac ... Let me guess, that provides enough data to identify the 
changesets within the CVs logs or some such ? Instead of dealing just with 
individual files ...

(Much of the complexity of cvs2svn (and my spiritual descendant cvs2fossil) 
comes from the fact that no cvs log information can be used to identify the 
changesets directly, and so they go through the RCS archives and then puzzle 
things, i.e. file revisions, together into changesets. (sourceforge, and maybe 
others, actually seems to truncate the log))

 > Only version information is converted at this time.  Wiki
> and ticket conversion is TBD.

Does the sqlite CVS have branches ? And they are handled as well ?

> The script execs cvs to extract each check-in then execs fossil to  
> check it into the fossil repository under construction.  The fossil  
> check-in is relatively fast.  It is the 6655 separate cvs check-outs  
> that take time.

Heh. Right, cvs is not the fastest. And it becomes slower the farther back you 
go in the history due to its reverse delta storage (head without delta, and 
historic revisions as delta to newer until we reach head). This is essentially 
quadratic in the number of revisions.

 >  Several hours are needed to perform the transfer.

Andreas.
_______________________________________________
fossil-users mailing list
[email protected]
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Reply via email to