Re: [Bug-gnubg] New fuctionallity for batch operations, opinions wanted!

[Joern Thyssen]

On Fri, Jul 11, 2003 at 08:10:53AM +1200, Joseph Heled wrote
> 
> Adding more and more specialized features is not good. While Joern 
> generalization is in the right direction, still you are limited in the 
> number of atomic operations, and have to add new ones to the code.
> Such a wizard might be a good idea, as long as it sits on top of a 
> reasonable scripting language. A better alternative is to provide 
> non-programmers with ready made snippets which they can Cut'N'Paste.

I think my wizard is the right way to go. It just helps a non-programmer
to construct a script and execute it. Since the Scheme part of gnubg is
not finished, it'll run the script in the native commands. Of course,
this is limited, but will do what most people want to do.

> GNUbg is in danger of getting bloated. I am not sure what the
> situation is right now. For example, the new "extending rollouts"
> feature has a non trivial cost in memory consumption. 

This has been fixed. The cost is now 8 bytes (Jim correct me if I'm
wrong) per move stored. This "bloating" is minimal as a move already
is 1024 bytes.

> Can one turn it off in compilation? 

I don't think it's necessary, since it's only bloated by 1%.

> Same question for other features which might not be applicable to
> everybody.  While playing with xml might be fun, personally I feel it
> is an overkill for something as simple as a MET table. And if you do
> already link the library, why not save matches in xml instead of sgf.

I've suggested that a long time ago, and Jim has also suggested that
recently. Some of the advantages is that it is extensible, which is good
when we add new stuff to the internal of gnubg that must be saved.
Using XSL it's possible to generate html, pdf, and text from this XML
file.

Jørn