[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Proposed svndiff format

From: ibej <brane_at_xbc.nu>
Date: 2000-10-08 23:08:21 CEST

Karl Fogel wrote:

> Greg Hudson <ghudson@MIT.EDU> writes:
>
>> Well, perhaps we should be cautious. But I think polishing the vcdiff
>> generator and writing a vcdiff parser would be a lot harder than
>> writing a generator and parser for the simpler format.
>
> Yah. I think we should hear from Branko before deciding, but you
> might be right...
>
> Branko, thoughts?

In no particular order:

    * What my envoder is (supposed to) do is generate vcdiff windows
      using a subset of coding table 0, which is about half of what
      vcdiff is (supposed to be) capable of. It's also far from being
      fully functional.
    * The vcdiff format can in theory support very compact output (since
      you can generate optimized instruction tables), but actually
      writing code that implements all of the features is IMO not
      trivial. I'd compare it to a simple optimizing compiler -- not
      rocket science, granted, but probably more than SVN needs right now.
    * Vcdiff isn't used anywhere, as far as I know; what are we trying
      to interoperate with?. What's more, the Internet Draft describing
      it has expired and I haven't found any follow-up. This may or may
      not be significant.
    * In fact, I'm not aware of *any* industry-standard binary delta
      format. Hmmm. Probably a deficiency on my part. Does xdelta count
      as industry-standard?
    * Greg's proposal would give us a usable binary delta format and
      should be very straight-forward to implement (he could also re-use
      some of my vcdiff code).
    * It's entirely possible that it'll take him less time to implement
      his format than to finish up even the rudimentary vcdiff encoder.
    * The vdelta paper, besides giving a short (incomplete) description
      of the vdelta algorithm, also describes an encoding format for
      vdelta. It might make sense to take a look at that. see if it's
      useable.
    * As you (Karl) say, we'll have support for other delta formats
      later on, anyway.
      
/Summa summarum/: Greg should go ahead and implement his simpler proposal.

In fact ... call me an idiot, but I feel that with a bit of work, it
could be a pretty good alternative to vcdiff. We might even want to
write up an Internet Draft one we've had some experience with the format.

That's my EUR0.02.

-- 
Brane C(ibej
home: <brane_at_xbc.nu> http://www.xbc.nu/brane/
ACM: <brane_at_acm.org> http://www.acm.org/
Received on Sat Oct 21 14:36:10 2006

This is an archived mail posted to the Subversion Dev mailing list.