[Top][All Lists]

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

Re: [Gnu-arch-users] Re: situations where cached revisions are not so go

From: Andrew Suffield
Subject: Re: [Gnu-arch-users] Re: situations where cached revisions are not so good
Date: Sun, 28 Sep 2003 16:11:23 +0100
User-agent: Mutt/1.5.4i

On Sun, Sep 28, 2003 at 07:44:37AM -0700, Tom Lord wrote:
> c) the costs of examining the graph are in some places high-enough
>    that if we do too much of it, the optimal SPF solution is useless
>    because we've spent too much time computing it
> >From those reasons, it's not trivially SPF, it's a heuristic search
> whose goal is to approximate SPF in the face of incomplete knowledge
> and with the catch that asking the wrong questions during the
> heuristic makes the heuristic worse than useless.  (Perhaps there is a
> meta-problem that reduces to a different instance of SPF, sure...)

Here's a trivial variation that fixes it: if the cost of examining a
branch of the graph is too high, prune that branch (assume its cost
would be infinite).

We can give a trivial definition of "too high" as "neither a local
archive nor an archive which was specified as part of the original

Both of these criteria can be improved upon to make the search produce
better results, but this will _work_ so long as you always cache
tagged revisions - so it's probably a good starting point.

  .''`.  ** Debian GNU/Linux ** | Andrew Suffield
 : :' : |
 `. `'                          |
   `-             -><-          |

Attachment: signature.asc
Description: Digital signature

reply via email to

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