RE: computer-go: Sharing Go modules - ready for download

[Jean-Pierre Vesinet <jpvesinet@xxxxxxxxxx>]

I think JDK1.3 needs a recompile to optimize the byte code for HotSpot VM.
But aren't some source files missing?

.\tesuji\games\go\computer\evaluation\ModuleTest.java:3: cannot resolve
symbol
symbol  : class MoveEvent
location: package general
import tesuji.games.general.MoveEvent;
                            ^
.\tesuji\games\go\computer\evaluation\ModuleTest.java:6: cannot resolve
symbol
symbol  : class GoMove
location: package go
import tesuji.games.go.GoMove;
                       ^
.\tesuji\games\go\computer\evaluation\TestGames.java:3: cannot resolve
symbol
symbol  : class MoveEvent
location: package general
import tesuji.games.general.MoveEvent;
                            ^
.\tesuji\games\go\computer\evaluation\TestGames.java:6: cannot resolve
symbol
symbol  : class GoMove
location: package go
import tesuji.games.go.GoMove;
                       ^

* * *

BTW the download page has lots of JavaScript error messages in Internet
Explorer 5.
They open pop-up windows, which makes it impossible to click the download
button.

* * *

I am writing a tsumego solver in Java.
My main problem is finding the most efficient way to update incrementally
lots of structures (chains, liberties, regions, links between regions and
chains, potential eye spaces, etc...). I made my own hashtable and iterator
classes to improve their reusability. For example the standard Enumeration
needs to be reinstantiated for each loop, whereas mine has a rewind()
method. I will try your reusable object scheme "FlyWeight". Having a pool of
recycled object instances may be a big improvement over reallocation - don't
know how big.
So, thanks for sharing!


> -----Message d'origine-----
> De : Mark Boon [mailto:tesuji@xxxxxxxxxxxxxxxxx]
> Envoyé : vendredi 15 juin 2001 07:20
> À : computer-go@xxxxxxxxxxxxxxxxx
> Objet : RE: computer-go: Sharing Go modules - ready for download
> 
> 
> Hmmm, did you try it on the test-program that I included? 
> Which version of
> the IBM VM did you use?
> 
> When I tried IBM's 1.2.2 VM it took almost 30% longer to 
> complete the tests
> than with the Symantec VM. Sun's 1.3 VM (don't know if it's 
> HotSpot or not)
> takes 60% longer.
> 
> I think a lot depends on being lucky that a particular VM has 
> optimized
> interpreting the kind of statements a program predominantly uses.
>