JPortal current status

[Malban]

Hello,

to you *all* - I mean the JPortal Fan-Club !

1)
Below an "unfinshed" picture of new Shop feature.
You can chose your card from a list of all cards, sortable (amongst others) by count of owned cards. As you see the background picture is not as yet updated.

2)
Work on StarterSet 2000 begun, 18 cards to go
Doesn´t sound like much left - but these are the ones where actually work must be done .

3) Mana - Pool
Started new Mana Management, which will result in more configureable Mana.
With that it will be easier to establish
a) new Mana-Sources
b) multicolored and colorless mana
c) restrictions on mana sources (only usable for creatures... etc)

Next release might be in March or so...

Malban

[justme]

Hello,

to you *all* - I mean the JPortal Fan-Club !

That's me if anyone was in doubt

YES!
Just the shop feature I was looking for. Selling excessive commons by the bucketload.

... and a fanclub current status.
Just got a Lenovo T43 dirt cheap. Threw Ubuntu on it right away, and will be adding Sun JRE 1.6 shortly in order to play JPortal.

Should be interesting to try this on a different OS

[Malban]

Update:

- working on a new AI, which will handle the game differently
- meaning:
* right now AI decides in each turn what to do, chosing attackers is independend of casting sorceries, using abilities and so on
* for implemtentation of more "non starter" (enchantments, more abilities, artefacts) the actions of each turns are more dependend on each other
* the new AI will at the beginning of its round "plan" the round, what cards it will play out / use for each turn to come (thus being able to damage opponents creatures in combat turn and finish them off in the second main turn - or things like that)
* naturally the "plan" must be adjusted on the fly during each turn, but if a plan is there as a beacon, a guideline - it is possible e.g. to save some mana for later use...
- that said, you will see it is not really easy to implement that, and trivial tasks (like implementing the 15 or so cards of starter 2000) will have to wait
- possibly the next 1.3 release will for that reason also be delayed.
- so in good old blizzard tradition. It is ready when it is ready


Malban

[justme]

- so in good old blizzard tradition. It is ready when it is ready

... and we're ready for it, when it's ready

[Malban]

Hiho,

Status:

- Conversion of old Hintsystem to new Hintsystem is ongonig
- I built a converter, which more or less migrates the hints
- nonetheless all cards must be double checked whether hints are ok.
- so far only the original portal cards are completely converted
- to differentiate my programming a bit I will probably now start with the AI itself befor turning to convert the other cards

- Task List todo befor next release:
* convert hints for all supported sets
* do the new AI
- finish new battle simulator
- implement new sorcery simulator
- phase based AI computation
* finnish of the Starter 2000 cards (only a handfull missing)

MayBe´s:
- semi automatic hint building from card text
- semi automatic code generation from advanced hints
-> if implemented "semi-automatic" card building from card text
-> will surely not work out of the box - but will certainly fasten the
development of new cards...
- cards with "ok" flag set can be played with / other not
- cards with AI flag - to show which AI is supported

Malban

[Malban]

Hi,

Implemented a feature, that lets you import all sets and examines the sets whether a card is already supported by "JPortal" known sets.

I don´t think any MTG-Like game should be measured in numbers of cards supported, since much of the fun derives from gameplay and not from some obscure card numbers. And for that it is essential that it is well playable and AI-wise a challenge.

Nonetheless, for once I would like to state, that JPortal Dev is supporting right now over 2000 official cards.

Malban

[Huggybaby]

Hi,

Implemented a feature, that lets you import all sets and examines the sets whether a card is already supported by "JPortal" known sets.

That will be very helpful to those who wish to add cards!
Also, I'm glad to see ongoing work on the hint system. I think the hint system is unique to Jportal and it's very cool.

2,000 cards is quite enough to have many hours of enjoyment!

[Malban]

Update...
All of "Starter 2000" - set is implemented now - at least game wise.

I have not implemented AI Code to actually handle the cards on the AI side - but that is the next step - as told many times befor .

For other MTG this might not be special, but since JPortal (as the name suggests) was originally only supposed to support "Portal" cards, which means many features of MTG were not supposed to be supported at all - this is quite a stop forward (or sideways?).

But since than all was implemented that I wanted - and interest in going on was still there. For that the "Engine" (if you might call it so) had to be changed to support new Card types.

JPortal now supports:

Enchantments, Artifacts, Instants (Portal instants were not really instantanous most of the time) and Abilities (Portal abilities were also very restricted) as new card types.

Plan after the new AI is still to implement 5th edition, since it is as far as I know the last befor the Planeswalker.

After that I might do (if still JPortal is on my mind) a last "current" edition - whatever that might be 2011 or 2012 by than .

If all that may come to pass - I will most probably quit. Than I would have realized
most of what can be done - and the challange will most certainly be gone.
I won´t singlehandedly support 12000 cards. Perhaps at some stage someone may join in and lend a hand. But if not thats also Ok with me


Cheers

Malban

[Malban]

Hiho,

just a small post about the ongoing status.

Here some screenies...

Here you see what the current "Stack" action is about.
In the new version game there will be a lot more ongoing on the stack. In order to have a little visual feedback, I keep track of the stack cards and what they "refer" to....

Probably only interesting for me !
I made an interface that I can call the game while ingame!
This can be seen as an ingame live debugging feature...
(Can also be used to cheat, and change cards on the fly, as you see)

Here you see another "Debugging" feature. Here you can select the current schedule, and plan of the Enhanced AI and look thru the cards it wants to play and what cards are needed to play...

As last Visual feedback you see here a information screen, that can be switched on. It shows the current weighting of the game situation and the comparrison to the oppnent.

That was all stuff I did more or less today. All this was neccessary to come to better terms with the new AI. But I keep deliberately piling stuff in front of me to avoid actually programming the darn thing.

But right now I can´t think of much more to do for preperation.

All I can do now to avoid is - do something completely different.
Like playing WoW - or doing a campaign mode or sealed deck... or or or...

Malban

[Malban]

Hip Hip Hurray,
the new AI played out a land.

"Small are the steps to success!"



Malban

[Malban]

Still working on new AI.

Now I know why I kept pushing this away again and again.
It is fiddlesticks .

I guess development is about 40% done.

AI can decide on playing cards, attacking.
Todo is stack activity and blocking. Changed structure again a bit. Now working (how surprising...) with trees.
I will do one go on brute force go and than build clues on how to reduce the multitude of possible leafs. Scoring/Weighting will probably be configurable...

Building a good AI takes time. Next release will have to wait a bit. Probably not befor June. A pitty that users wont see what huge changes the programm went thru in the last couple of months. It nearly looks the same as befor - but the internals are quite different...

Malban

[Malban]

Haven´t done anything for a week now... start possibly next week after easter... much todo private an dwork related...

Regards...

Malban

[Malban]

Hiho,

I still am working on the new AI -> EAI meaning enhanced AI .
Progress is a bit slow, since many many things have to be taken care of.

I will just present a visual impression on how things stand now.

Assume following situation:

On the bottom are the current cards in Hand, than the played out lands above, and above that the current played out creatures....

Next:

Here you see the current EAI - Debug Window.
The nodes of the tree are labeled by comvention:
NUMBER_PHASE_[CARDS...CARDS]

NUMBER here is the relative score of the decision (the relativ score is the difference of the player score to the opponent score), negative is bad - positiv is good. PHASE - is the name of the current phase. CARD - are the cards that will be part of the action (playing out, activation, attacking...)

Computer is now in the Main1 Phase and has layed out some possible paths to follow. Look at the tree.
The tree is build by "phases". Each level of the tree represents one phase. Each level can have as many sibblings as needed.

The example shows the decision the EAI is about to compute (are rather - has already computed). The open leaf shows one possible "path".

You can see the branching out on the phases. Computer now thinks about playing out the "Bear cub" and a "Plains" in Main one.
In the "Plan/Stack" part of the display, you can see the corresponding cards - and action. (Right-Clicking on the card gives the normal card description). Left clicking on one such card "fills" the Tab below.

In case of the "Bear Cub" it shows, what the EAI plans to pay the card with (TAB - Mana sources).

Clicking on a branch further "below" shows:

Here we see, that the Computer planned to attack with both already played out cards, namely "Norwood Ranger" and "Goblin Mountaineer". I selected the Node below that, which shows what the computer assumes the other player will use as blocking manouver. Since only one card is untapped it assumes that very card (shown again in the Plan/Stack Tab).
Clicking that card shows (in the Target Sources tab) what it presumely will block.

At the end of the combat section of the branch:

You can see/examine (Virtual Match tab) What the result of the combat is. Namely we only have one attacker left - and player 1 (opponent) lost one life point.
The Scoring result of that branch alltogether is -94.

(Not attacking at all - at the moment results in a score of -76 - so the EAI will decide against the above shown Path )

Many things are left todo... but it is still very challanging to do an AI. But the relase of 1.3 will have to wait, since many things I planned for 2.0 are now scheduled for 1.3 .
(2.0 will "only" be fifth edition - I guess, everything else will be done befor.)

Regards...

Malban

[Huggybaby]

Wow, that's impressive. I don't know if any other game exposes the AI reasoning process like that.

[Malban]

Ha!
Now my Computer went all to pieces - doesn´t even boot anymore.

This will cause quite some slowdown...
Well - time will tell...

Regards

Malban