gnu-arch-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnu-arch-users] Re: RFC: arch protocol, smart server, and tla imple


From: Jason McCarty
Subject: Re: [Gnu-arch-users] Re: RFC: arch protocol, smart server, and tla implementation prototypes
Date: Thu, 5 Feb 2004 23:16:20 -0500
User-agent: Mutt/1.5.4i

Chris Gray wrote:
> Currently, when you commit a patch in arch, it is like you are adding
> a new element to a linked list.  I'm saying that we could change the
> data structure to a skiplist and keep the dumb server.  All of the
> summary deltas and data structure maintenance would happen client
> side.  

I was wondering what you meant by the suggestion of skiplists, but that
explains it. Effectively I think it would amount to just another
summary-delta placement strategy, because you still must keep
single-revision deltas, which make the other deltas just cached data.

Traversing a graph of summary-deltas already looks like traversing a
skiplist, so it's really just the proposed placement of deltas that
differs. The most-often proposed binary (though it could be ternary or
higher) hierarchical decomposition is one method; skiplists are another.

Because revisions are never inserted or deleted in the middle of a
version though, I don't think you get an advantage from skiplists, and
possibly a disadvantage (you must traverse a logarithmic subset of the
graph to decide which deltas to add, no?).

And any placement scheme ought to support manual adding of deltas
between popular versions.

> Something more deterministic would probably be a better approach for a
> smart server.

On a smart server you can use whatever indexing scheme you like too, so
weird placement of deltas might not be so large a concern.

-- 
Jason McCarty <address@hidden>




reply via email to

[Prev in Thread] Current Thread [Next in Thread]