colossus-developers Mailing List for Colossus (Page 230)

I did a bunch more cleanup for the client/server split over the past few 
days.  Several more of the dialog classes now just pass strings in and 
out rather than using real game state objects.  Next I need to fix 
negotiation.

I expect to be ready to repackage the game into net.sf.colossus.client, 
.server, and probably .util (for generic classes like Utils, Probs, 
Perms, Minimax) very soon.  I want to leave the base net.sf.colossus 
package empty (except maybe for startup methods), so I'll need to move 
Tom's stuff that's in that package down a level.  There's not much there 
and only the caretaker view is currently used, so that shouldn't be a 
big deal.

JDK 1.4 can't call from a named package back into the default package. 
This may cause some compilation problems under 1.4 while things are 
halfway moved.  Once the default package is empty, the game should 
resume working fine under 1.4.  Use 1.3.1 until then.

I'm leaning toward doing cvs adds and removes to move files to new 
directories, rather than making a local copy of the repository and 
moving the ,v files around then asking the SF admins to copy my local 
copy over their copy.  That will cause a discontinuity in CVS history. 
(The old files will be in the Attic so history will not truly be lost, 
just painful to see.)  But it will mean that we won't lose SF CVS access 
for a few days, won't have to worry about changes during that time 
possibly being destroyed, etc.  If anyone objects to this strategy let 
me know.

-- 
David Ripton   dr...@wi...

> OK, I'll do that. I'll tag HEAD before merging, just in case...

done ; tag on trunk are before_merging_creature_loader and
after_merging_creature_loader.

One problematic thing : battlelands are reloaded each time they are
displayed, maybe that should be changed (it's a drop-in replacement to
the old set-hex-data-by-hand in HexMap.java).

-- 
DOLBEAU Romain               | The Gods made Heavy Metal
ENS Cachan / Ker Lann        |     and it's never gonna die
Thesard IRISA / CAPS         |           -- Manowar,
dol...@cl...    |    'The Gods made Heavy Metal'

> So if you think your code pretty much works and you'd like other people to
> look at it sooner, you might want to go ahead and merge it into the head
> branch.

OK, I'll do that. I'll tag HEAD before merging, just in case...

> Control-click is the standard way on the Mac?  Oops, I thought 
> command-click was.  I'll fix MasterBoard.isPopupButton() to also check
> for control.  (You'd think that Java would handle this kind of thing 
> more transparently, but whoever designed MouseEvent wasn't thinking very
> hard.)

left button are usually set to do control-click on MacOS9, same on OSX
5I just tried :-)

-- 
DOLBEAU Romain               | We'll know for the first time
ENS Cachan / Ker Lann        |     If we're evil or divine
Thesard IRISA / CAPS         |          -- Dio,
dol...@cl...    |             'The Last In Line'

Romain Dolbeau wrote:


> "cvs update -r creature_loader" from inside a checked-out
> tree _should_ work... if it doesn't there's a permission
> problem somewhere.


Looks like a well-known CVS bug.  People who can't write to the val-tags 
file in the repository can't check out code outside the default branch. 
  I found a small patch to CVS that's supposed to fix the bug, but it 
didn't work.  I suspect that the val-tags file is owned by you (because 
you're the first programmer on this project to use tags or branches) and 
has 644 permission, so only you can check out non-default branches until 
the SourceForge admins chmod it.  I'll submit a support request asking 
them to do that, but it sometimes takes them a few days.  So if you 
think your code pretty much works and you'd like other people to look at 
it sooner, you might want to go ahead and merge it into the head branch.


>>Great.  Did you have to fix anything in particular to make it Mac-compatible?
>>Does right-button functionality work with option-click on a one-button mouse?
>>
> 
> It's on MacOSX, not MacOS9, so it's mostly Unix underneath.
> I only tried with the external 3-buttons mouse of my laptop,
> and left-click works fine (no particular driver). I guess
> control-click should work too, as it is the standard way of
> right-clicking.


Control-click is the standard way on the Mac?  Oops, I thought 
command-click was.  I'll fix MasterBoard.isPopupButton() to also check 
for control.  (You'd think that Java would handle this kind of thing 
more transparently, but whoever designed MouseEvent wasn't thinking very 
hard.)

-- 
David Ripton   dr...@wi...

David Ripton wrote:

> I can't check out your branch to test it.
> 
> Not sure if we have a permissions problem in the repository at SF, or if
> I'm doing something wrong.  (The equivalent cvs commands without the -r
> work fine.)

"cvs update -r creature_loader" from inside a checked-out
tree _should_ work... if it doesn't there's a permission
problem somewhere.

> Great.  Did you have to fix anything in particular to make it Mac-compatible?
> Does right-button functionality work with option-click on a one-button mouse?

It's on MacOSX, not MacOS9, so it's mostly Unix underneath.
I only tried with the external 3-buttons mouse of my laptop,
and left-click works fine (no particular driver). I guess
control-click should work too, as it is the standard way of
right-clicking.

-- 
DOLBEAU Romain           | I witness a time and a place that never dies,
ENS Cachan / Ker Lann    | still frozen in time, this darkness the only 
Thesard IRISA / CAPS     | place that I can hide. I witness, a dream.   
dol...@cl...|  -- Black Sabbath, 'I Witness'

Romain Dolbeau wrote:


> I just merged your last change in creature_loader, it should
> be able to replace HEAD whenever you want. It only needs
> some testing & reviews.


I can't check out your branch to test it.

~/src/Colossus]$ cvs up -r creature_loader_update_05_done
cvs [server aborted]: cannot write /cvsroot/colossus/CVSROOT/val-tags: 
Permission denied

~/src/Colossus]$ cvs -z3 
-d:ext:dr...@cv...:/cvsroot/colossus co -r 
creature_loader_update_05_done Colossus
cvs [server aborted]: cannot write /cvsroot/colossus/CVSROOT/val-tags: 
Permission denied

Not sure if we have a permissions problem in the repository at SF, or if 
I'm doing something wrong.  (The equivalent cvs commands without the -r 
work fine.)

 
> BTW, both branches are compiling and workinf fine on the
> latest MacOS X (only a few spurious refresh once in a while,
> guess it's a problem with the OSX JVM).

Great.  Did you have to fix anything in particular to make it Mac-compatible?  Does right-button functionality work with option-click on a one-button mouse?


-- 
David Ripton   dr...@wi...

David Ripton wrote:

> So I'll wait until Romain's branch is merged in before doing it.

I just merged your last change in creature_loader, it should
be able to replace HEAD whenever you want. It only needs
some testing & reviews.

BTW, both branches are compiling and workinf fine on the
latest MacOS X (only a few spurious refresh once in a while,
guess it's a problem with the OSX JVM).

-- 
DOLBEAU Romain               | We'll know for the first time            
ENS Cachan / Ker Lann        |     If we're evil or divine              
Thesard IRISA / CAPS         |          -- Dio,                         
dol...@cl...    |             'The Last In Line'

I took all Battle, Legion, and Critter references out of BattleMap and 
BattleChit, and then got the game working again.  The only new bug I 
know about is that the clicked-on battle chit no longer moves to the top 
of the z-order, which is not a really big deal.  (In any case, z-order 
should be maintained purely on the client side, rather than rearranging 
the order of critters in the server-side legions, so it needs to be 
reimplemented, for both battle chits and legion markers.)  It's possible 
that I missed something.  If anyone sees any new NullPointerExceptions, 
please forward me the stack trace.

I plan to take the Game and Legion references out of MasterBoard next.

Once that is done, it might be time to start taking code out of the 
default package and putting it into net.sf.colossus.client and .server 
so that we can use package private to enforce the future network 
boundary.  CVS doesn't track files being moved to new directories very 
well, so this will create a bit of a mess.  So I'll wait until Romain's 
branch is merged in before doing it.

-- 
David Ripton   dr...@wi...

> I used a white border for masterboard hexes, and a red border for 
> battlemap hexes.  

On a related note, I think a good thing would be to add a visual clue of
the "nativeness" of an Hex on battlemap hexes, i.e. use a different
border color for non-native that would suffer penalty for staying in the
hex (non-drift native in drift, non-bramble native in bramble, ...). Of
course assuming it's doable :-)

-- 
DOLBEAU Romain               |
ENS Cachan / Ker Lann        | l'histoire est entierement vraie, puisque
Thesard IRISA / CAPS         |     je l'ai imaginee d'un bout a l'autre
dol...@cl...    |           -- Boris Vian

I changed hex selection to affect only the outer part of the hex, rather 
than the whole hex, so the underlying terrain color is still visible.

I used a white border for masterboard hexes, and a red border for 
battlemap hexes.  I don't think white is the best color, but all of the 
better ones I could think of blend in with one of the terrain colors, 
and I thought that using different selection colors for different 
terrain types would be distracting.

I played with the idea of using a wider stroke for the hex border, but 
the strokes went outside the selected hex, which left garbage behind 
when the hex was unselected and repainted, and also had issues with 
adjacent non-selected hexes painting over part of the selection line.  I 
could have probably fixed these by clipping, but instead I created an 
inscribed hex with an AffineTransform, and filled the outer hex with the 
selection color and the inner hex with the terrain color, with a black 
line between to help with contrast.

I think this is pretty good, but not perfect.  If anyone has a better 
idea for hex selection, please let me know.

-- 
David Ripton   dr...@wi...

Hello,

Well, it seems it works ; CVS branch creature_loader can
now load a VARiant file that describre the variant (i.e.
which 3 files to use) ; another important changes
is that the 3 starting creatures are now the 3 tower
creatures from the 'recruit' files. SimpleAI has
been changed to reflect this, but only to the extent
of allowing initial split, so it's a little dumb if
big changes are made to the original game.

Any comments welcome.

-- 
DOLBEAU Romain               | For a hill, men would kill,              
ENS Cachan / Ker Lann        |   Why ? They do not know.                
Thesard IRISA / CAPS         |      -- Metallica,                       
dol...@cl...    |       'For Whom The Bell Tolls'

Hello,

I just finished commiting a BattlelandLoader on the
branch 'creature_loader'. it dynamically loads the
battlelands from files under the Battlelands directory.

So it should now be possible to change everything
except the BattleHex (only the basis stuff is available
to create new Battlelands).

Any comments welcome.

-- 
DOLBEAU Romain               | We'll know for the first time            
ENS Cachan / Ker Lann        |     If we're evil or divine              
Thesard IRISA / CAPS         |          -- Dio,                         
dol...@cl...    |             'The Last In Line'

I've commited a small change to HEAD, that add the possiblity
of randomly choosing the AI for AI players. Of course
the actual AI used is displayed in the log, so don't look :-)

As soon as the MinimaxAI works properly (according to a friend
that tried it, there's still problems, I didn't check), I propose
'A Random AI' to be the default choice of AI in place of
'SimpleAI'.

-- 
DOLBEAU Romain               | Well if you don't like rock 'n' roll       
ENS Cachan / Ker Lann        |        Then you're too late now            
Thesard IRISA / CAPS         |   -- Rainbow,                              
dol...@cl...    |         'If You Don't Like Rock 'N' Roll'

Romain Dolbeau wrote:
> 
> I've added a branch 'creature_loader' (branched
> from tag 'creature_loader_root')

I've update the branch to track HEAD, it's synced
to tag creature_loader_update_01 (the branch
after sync is tagged creature_loader_update_01_done)

> any comments/advices/criticisms welcome

-- 
DOLBEAU Romain               | Brothers of Metal will always be there     
ENS Cachan / Ker Lann        | Standing together with hands in the air    
Thesard IRISA / CAPS         |           -- Manowar,                      
dol...@cl...    |                     'Brothers of Metal'

Romain Dolbeau wrote:

> Dunno what, but very recent CVS doesn't have proper
> damage carry. It only proposes the already killed
> opponents.


Swapped variable names.  A problem with overly-long methods like 
Critter.strike().  Checking in the fix now.

Dunno what, but very recent CVS doesn't have proper
damage carry. It only proposes the already killed
opponents.

Might be related to:

#####
RCS file: /cvsroot/colossus/Colossus/BattleMap.java,v
 [SNIP]
revision 1.152
date: 2001/10/27 02:57:39;  author: dripton;  state: Exp;  lines: +2 -2
branches:  1.152.2;
Move Critter.carryFlag to Battle.carryTargets.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Fix countBrambleHexes to use correct toLeft() logic, and
   bail out if terrain is not brush or jungle.
Color-code numbers in CreatureCollectionView.
Add Colossus.jnlp file for Java Web Start.
#####

-- 
DOLBEAU Romain            | The Gods made Heavy Metal                  
ENS Cachan / Ker Lann     | and it's never gonna die.                  
Thesard IRISA / CAPS      | -- Manowar in 'The Gods made Heavy Metal'  
dol...@cl... |

I've added a branch 'creature_loader' (branched
from tag 'creature_loader_root'), that add
creature loader support. This one required a lot
of changes, as it removes all static Creature
(one must use getCreatureByName() for explicit
access). I haven't updated the various main()
that were used for testing, only commented
them out for proper compilation.

i've splitted initServerAndClients for savegame,
one must load Creatures before parsing the savefile.

Recruit.ter onw include the color to use for the terrain.

syntax of MAP file changed

removed access to CharacterArchetype.getMaxCount to
use file-defined MaxCount.

apparently it works (including using additional
terrain and creatures), but it needs testing before
going to HEAD.

any comments/advices/criticisms welcome

-- 
DOLBEAU Romain            | For a hill, men would kill, Why ?          
ENS Cachan / Ker Lann     | They do not know.                          
Thesard IRISA / CAPS      | -- Metallica in 'For Whom The Bell Tolls'  
dol...@cl... |

Jerome Reiger wrote:

  > For the webpage and curiosity ....what changes were made from the OCT 19
  > to OCT 24  ...in noticed it seems to play much faster

I limited the number of battle move order permutations to try to 100.
Doesn't seem to hurt the AI much, and it speeds it up a lot.

Romain added the code to support multiple AIs at once and also allow
variable maps and recruiting trees.  To test his multiple AI support I
split Bruce's Minimax masterboard AI code out into a separate class and 
got it sort of working (though it's slow and only moves one legion per 
turn.)  I suspect that even if we get MimimaxAI working well, it'll 
still be slow.

I also fixed a few bugs, added the helper recruit chits, and moved the 
movement die and battle dice inside the board and map windows.

After the last zip file drop, I fixed a few more bugs, color-coded the 
caretaker display, and moved the carry flag from Critter to Battle.  I 
will check all this in soon.

I'm probably going to finish Java Web Start support tonight.  I have the 
.jnlp file correct, but I need to change the Makefile to autosign
the jar file.

Next I want to fix the teleport bug and clean up the battle dice display.

-- 
David Ripton   dr...@wi...

I've commited changes to Game.java to allow the Recruit.ter
file to change Tower Creatures ; first 3 are recruited
with anything, fourth recruited with 3 of same type
(but not immortal) or itself, fifth recruited with titan
or itself. I've also added a file in docs/ to explain the
file format or .map and .ter

The changes shouldn't have broke anything...

-- 
DOLBEAU Romain               | Nothing is real but the way that I feel    
ENS Cachan / Ker Lann        | And I feel like going down, down, down     
Thesard IRISA / CAPS         |   -- Rainbow,                              
dol...@cl...    |                 'Self Portrait'

David Ripton wrote:

> I think the mapLoader patch is a good idea.  I need to learn JavaCC.  I'm
> not quite sure of the right way to check in this code.  I think that the
> JavaCC-generated .java files should be checked in, not just the .jj and
> .map files, so that people without JavaCC can still build the game.  It
> probably makes sense to move the mapLoader stuff into a separate package so
> it doesn't add too much clutter and distract people trying to absorb the
> main code.  I'll take a shot at doing that before checking it in, since CVS
> doesn't handle moving files around after they're checked in very
> gracefully.

Adding all the .java to the CVS is possible ; they only differ
with differing version of JavaCC. The StrategicMapLoader.jj
works with 1.1 and 2.1, but I suggest 2.1 which is the latest
release.

The main necessary change is IMHO that the code shouldn't
always ask for the map, it should be a new button in the
first dialog or something like that.

BTW I haven't found a way to compute the LabelSide from
the Hex position, the big list of SetLabelSide is ugly.
Could simply be added to the file, but I'd rather compute
them on-the-fly...

-- 
DOLBEAU Romain               |        Long live rock and roll             
ENS Cachan / Ker Lann        |        Long live rock 'n' roll             
Thesard IRISA / CAPS         |        Long live rock and roll             
dol...@cl...    | -- Rainbow, 'Long Live Rock 'N' Roll'

I have partially integrated my look and feel of the masterboard into the game. A version is downloadable from http://chris.howeville.com/Colossus.jar .

Mainly I have replaced the paint() method of GUIMasterHex with a version that paints pre-rendered images. There are a few other changes to support the details of this, but there is not a separate swing component for every hex.

There are still some issues with threading as the hex images get loaded, as the way it works now is not terribly sophisticated.  I seem to have made a rounding error somewhere, as there are soume double horizontal lines between some of the hexes.  And of course I still have not put the hex labels on yet.

--Chirs