Re: [gnugo-devel] fuseki db again

[Pierce T . Wetter III]

On Apr 13, 2005, at 10:13 AM, Douglas Ridgway wrote:

> On Sat, 9 Apr 2005, Arend Bayer wrote:
>
> > Maybe we should just put it in and wait for someone being annoyed 
> > enough
> > about the size to add a gzip solution.
>
> If someone wants to do this, here's what should go in:
>
>   1. http://dridgway.com/Go/fuseki.tgz contains replacement fuseki*.db
> files. This is unchanged from November.
>   2. The code patches from the Nov 18 email.
>   3. The patches from Dec 14 reinserting a couple of patterns.

 Re-reading the thread, it seems to me that the real issue is actually 
CVS.

 That is, its not so much that anyone really has an objection to the 
fuseki.db as it stands (they believe it would help, even without 
testing).

 The real objection was carrying 20 M for fuseki*.db around every time 
you checked out GnuGo.

 I'm wondering if rather then compression, if fuseki.db could be loaded 
at runtime like some chess programs load their opening books. Then 
fuseki-large.db need not be stored in the CVS tree, or if there could 
be a GNUGO-Data CVS tree instead.

 In fact, having a gnugo-data cvs tree that was a peer of gnugo would 
probably be a good idea in general, and has the virtue of requiring no 
code be written...

 So I guess my final suggestion would be to integrate the code changes, 
while moving data to a parallel tree.

 Pierce