Forge Beta: 09-12-2014 ver 1.5.26
Moderators: timmermac, Blacksmith, KrazyTheFox, Agetian, friarsol, CCGHQ Admins
Re: Forge Beta: 09-12-2014 ver 1.5.26
by elcnesh » 16 Sep 2014, 15:18
One of the things I've fixed with the GUI refactoring! (It's not just bugs xD)lujo wrote:If you morph Luminthread Field, the screen starts flickering. I think it's turning it into a creature momentarily and then into an enchantment all the time, or during some of the phases (it flashes is among creatures and then back). I think I've noticed a simmilar occurence in Theros block limited with things which can change status between creature and enchantment.
Try a recent snapshot to see it in effect. (But beware that you may encounter other problems there, so don't use it for important quests or anything! And please report anything you find.)
- elcnesh
- Posts: 290
- Joined: 16 May 2014, 15:11
- Location: Netherlands
- Has thanked: 34 times
- Been thanked: 92 times
Re: Forge Beta: 09-12-2014 ver 1.5.26
by lujo » 16 Sep 2014, 15:44
I know, but while that sort of thing can slide most of the time, some of the time it's disasterously stupid. An AI which doesn't even consider targeting Lin Sivvi even though it has coupios removal is an AI that needs tweaking, even if that case would just be an exception. Really. For the sake of the format, please make Lin Sivvi a top priority target if there is any way to do it, if you're personaly not aware of how ridiculous it is find someone who played during the masques block and tell him the story, he'll tell you the same. Very few formats have ever been so warped by one card as that one was.friarsol wrote:This isn't really a bug, it's just the AI being dumb (not evaluating Lin Sivvi's ability when comparing it to other creatures).
Also, this level of AI dumbmess is causing problems elsewhere and is worth looking into. It's not just undervaluing Lin Sivvi, it holds removal and doesn't use it on Lin Sivvi sitting solo on the board, which means it would probably never target him. This + wither is causing the Plant pet to be way better and it was supposed to be as the AI won't attack into it and won't remove it (same deal low power high toughness - AI ignores).
Bigger issue than it looks.
Without even getting into what it means for constructed. A Lin Sivvi which the AI will ignore while holding a hand full of removal is beyond silly, just read the card.
IMPORTANT TLDR: I doesn't matter what's causing it - Lin Sivvi, the Plant and several other things are not supposed to have HEXPROOF. Lin Sivvi with effective hexproof is a fantasy card, which also happens to be a buffed fantasy version of one of the most broken cards in various environments ever. So you can't possibly justify claiming that it's even been functionally ported into the game, because it hasn't.
---
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
Re: Forge Beta: 09-12-2014 ver 1.5.26
by friarsol » 16 Sep 2014, 16:15
Thanks for your patronizing attitude. Please post AI Improvement requests in the appropriate thread as the link in my signature requests.lujo wrote:I know, but while that sort of thing can slide most of the time, some of the time it's disasterously stupid. [.. snip..]
- friarsol
- Global Moderator
- Posts: 7593
- Joined: 15 May 2010, 04:20
- Has thanked: 243 times
- Been thanked: 965 times
Re: Forge Beta: 09-12-2014 ver 1.5.26
by lujo » 16 Sep 2014, 16:51
Verdant Field ability description box in the drop down menu on the land it enchants is blank.
And f**k my patronizing attitude, there's no way to state that problem, it's huge if it affects more cards, and any card which gets the effects of being hexproof simply because it has 1 power or an unmarked ability isn't ported properly and it shows in game. Just because someone can handwave the issue, or stick it on the ported list without realizing fully what he's done doesn't mean it isn't there, and it's indistinguishable from a bug. I will report it in the AI thread, though.
And f**k my patronizing attitude, there's no way to state that problem, it's huge if it affects more cards, and any card which gets the effects of being hexproof simply because it has 1 power or an unmarked ability isn't ported properly and it shows in game. Just because someone can handwave the issue, or stick it on the ported list without realizing fully what he's done doesn't mean it isn't there, and it's indistinguishable from a bug. I will report it in the AI thread, though.
---
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
Re: Forge Beta: 09-12-2014 ver 1.5.26
by squee1968 » 16 Sep 2014, 18:10
Here's the script for Lin Sivvi, Defiant Hero:lujo wrote:Verdant Field ability description box in the drop down menu on the land it enchants is blank.
And f**k my patronizing attitude, there's no way to state that problem, it's huge if it affects more cards, and any card which gets the effects of being hexproof simply because it has 1 power or an unmarked ability isn't ported properly and it shows in game. Just because someone can handwave the issue, or stick it on the ported list without realizing fully what he's done doesn't mean it isn't there, and it's indistinguishable from a bug. I will report it in the AI thread, though.
: Search your library for a Rebel permanent card with converted mana cost Where do you see hexproof anywhere? You do realize that Forge is in a BETA state? This means there are always bugs or things that don't work exactly as they should. There are also things that need to be improved upon. You also need to realize that the developers are volunteers, no one's being paid anything for their time and effort. They fix things and improve things on their schedule, not yours. Your attitude wouldn't motivate me to do anything for you, and I imagine most of the developers would agree with me. The developers are doing a fantastic job. If you want things to improve on your schedule, then apply to become a developer yourself, if you're a java programmer.
Re: Forge Beta: 09-12-2014 ver 1.5.26
by lujo » 16 Sep 2014, 20:57
I'm currently hugely annoyed at myself for not being a java programmer. And anyway, the developers don't need to do anything for me, but fixing or not fixing or even paying attention to what I'm talking about here has major concequences.
Hexproof - a creature can't be target of spells or abilities your opponents control. There is none on that card, and there is supposed to BE none, it's powerful enough as it is. But the way the game treats it is as if it effectively has hexproof, since the AI won't target it even if it's alone on the board and even if the AI has Seal of Doom. Multiple ones. I've won two drafts simply due to this. So the implemented card, in fact, has hexproof for you, and if the AI would be using it then it doesn't as you'd target it.
Now if the AI can ignore Lin Sivvi, who is top removal priority in any match she's in, it means that something about her stats makes a creature have "effective hexproof", meaning it will be passed over for removal. Since the vast majority of creatures, agressive ones especially, have 2 power, getting rid of something with 3-4-5 toughness requires removal, otherwise the AI won't attack is you have enough total power on the board to kill any one of it's weenies.
So it's not counting such cards as problems it can remove, even if they're gamebreaking stupidities like that one, but it's counting them as reasons not to attack. Which then makes it behave more stupidly than it needs to and breaks a bunch of cards by making them all basicaly hexproof. Which then makes it misuse creatures, misuse removal and makes cards do stuff which isn't written on them for the player.
It's not a minor thing, or a personal gripe, if this is happening with any 1/3 or 1/4 creature then the AI isn't "working ok-ish" or "stupid but what can you do", it's a huge ass script/code/whatever flaw, which possibly can be fixed, and really should be fixed because if the AI wasn't playing it like this the game would play very differently and much, much closer to how it's supposed to.
My tone is due to the fact that it looked to me like noone noticed it. If I were a MtG engine dev, volunteer or payed, and saw this happening, I'd drop whatever I was doing and get on it right away. And if I noticed this kind of thing happening I certainly wouldn't put cards affected by it on the "implemented" list, because they've been horribly missimplemented (possibly due to a bug or oversight or whatever). I'm just mortified that noone noticed for all this time. If this is working the way it is now, how can you even test the game at all?
Hexproof - a creature can't be target of spells or abilities your opponents control. There is none on that card, and there is supposed to BE none, it's powerful enough as it is. But the way the game treats it is as if it effectively has hexproof, since the AI won't target it even if it's alone on the board and even if the AI has Seal of Doom. Multiple ones. I've won two drafts simply due to this. So the implemented card, in fact, has hexproof for you, and if the AI would be using it then it doesn't as you'd target it.
Now if the AI can ignore Lin Sivvi, who is top removal priority in any match she's in, it means that something about her stats makes a creature have "effective hexproof", meaning it will be passed over for removal. Since the vast majority of creatures, agressive ones especially, have 2 power, getting rid of something with 3-4-5 toughness requires removal, otherwise the AI won't attack is you have enough total power on the board to kill any one of it's weenies.
So it's not counting such cards as problems it can remove, even if they're gamebreaking stupidities like that one, but it's counting them as reasons not to attack. Which then makes it behave more stupidly than it needs to and breaks a bunch of cards by making them all basicaly hexproof. Which then makes it misuse creatures, misuse removal and makes cards do stuff which isn't written on them for the player.
It's not a minor thing, or a personal gripe, if this is happening with any 1/3 or 1/4 creature then the AI isn't "working ok-ish" or "stupid but what can you do", it's a huge ass script/code/whatever flaw, which possibly can be fixed, and really should be fixed because if the AI wasn't playing it like this the game would play very differently and much, much closer to how it's supposed to.
My tone is due to the fact that it looked to me like noone noticed it. If I were a MtG engine dev, volunteer or payed, and saw this happening, I'd drop whatever I was doing and get on it right away. And if I noticed this kind of thing happening I certainly wouldn't put cards affected by it on the "implemented" list, because they've been horribly missimplemented (possibly due to a bug or oversight or whatever). I'm just mortified that noone noticed for all this time. If this is working the way it is now, how can you even test the game at all?
---
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
Re: Forge Beta: 09-12-2014 ver 1.5.26
by drdev » 16 Sep 2014, 21:23
lujo, please understand that it's not that nobody cares. In fact, we currently have a developer that's looking into doing a massive overhaul of the AI and game logic that will allow making the AI much much smarter than it is now, which would hopefully address issues like you're seeing.
In the meantime, would it be enough if we simply hard-coded a list of "kill on sight" cards that the AI should target over any others? We could even assign each a number that determines just how high a priority target that card is in case there's more than one present. That's probably the simplest and quickest thing we could do to hold people over until more involved changes can be done.
If we went that route, are there other cards besides Lin Sivvi you'd suggest we prioritize?
In the meantime, would it be enough if we simply hard-coded a list of "kill on sight" cards that the AI should target over any others? We could even assign each a number that determines just how high a priority target that card is in case there's more than one present. That's probably the simplest and quickest thing we could do to hold people over until more involved changes can be done.
If we went that route, are there other cards besides Lin Sivvi you'd suggest we prioritize?
- drdev
- Programmer
- Posts: 1958
- Joined: 27 Jul 2013, 02:07
- Has thanked: 189 times
- Been thanked: 565 times
Re: Forge Beta: 09-12-2014 ver 1.5.26
by lujo » 16 Sep 2014, 22:02
^ Well, that would do a world of good, yes. I'm currently testing various draft formats (and doing research on block constructed and draft metagames from pro tours and well established meta conventions and trying them out in paractice to see how the AI responds), so if I run into any more cards like that I'll put them up. I generally report them in the AI improvement thread, this looked like a systemic error with certain stat configurations on creatures making them unreasonably low priority which caused a mess in other places.
System wide it basically means efficent walls get passed over way more than they should, I'll end up with a list at some point. It also seems to be deathly afraid of attacking into wither, so wither + low power + high toughnes breaks stuff hard.
As for broken bombs which don't have stats/abilities that draw the curret AI's removal aggro:
Lin Sivvi is the poster child of kill on sight in any format, Arcbound Ravager and Disciple of the Vault are format warping in Mirrodin block (and any constructed decks they're in), and Nightshade Seer and Cinder Seer destroy the AI in Urza's limited. Wild Mongrel is god of Oddisey block, and there's various format defining pingers (in tempest block especially) but I'll need to run a few more of those to see is it as bad as Lin Sivvi. If I run into/remember any more I can tell in the AI thread.
Sorry for the tone, really, I'm enjoying drafting sets I thought I'd never draft again, and anyone who's played masques would get a brain stroke if he saw that of all the "accidentally hexproof" cards that could've happened it had to be that particular p***e of S***t.
EDIT: I made a short thread in the dev section about it I'll do research and post findings for the list there, would that be ok? It would do the game a lot of good.
System wide it basically means efficent walls get passed over way more than they should, I'll end up with a list at some point. It also seems to be deathly afraid of attacking into wither, so wither + low power + high toughnes breaks stuff hard.
As for broken bombs which don't have stats/abilities that draw the curret AI's removal aggro:
Lin Sivvi is the poster child of kill on sight in any format, Arcbound Ravager and Disciple of the Vault are format warping in Mirrodin block (and any constructed decks they're in), and Nightshade Seer and Cinder Seer destroy the AI in Urza's limited. Wild Mongrel is god of Oddisey block, and there's various format defining pingers (in tempest block especially) but I'll need to run a few more of those to see is it as bad as Lin Sivvi. If I run into/remember any more I can tell in the AI thread.
Sorry for the tone, really, I'm enjoying drafting sets I thought I'd never draft again, and anyone who's played masques would get a brain stroke if he saw that of all the "accidentally hexproof" cards that could've happened it had to be that particular p***e of S***t.
EDIT: I made a short thread in the dev section about it I'll do research and post findings for the list there, would that be ok? It would do the game a lot of good.
Last edited by lujo on 16 Sep 2014, 22:49, edited 2 times in total.
---
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
Re: Forge Beta: 09-12-2014 ver 1.5.26
by lujo » 16 Sep 2014, 22:20
And, err, I may have an actual bug to report, though I'm not sure if it's bug or AI. Following situation occurs a lot in Masques draft: I have Kyren Sniper, opponent has either Charm Peddler or Alabaster Wall. It often tries to prevent the damage from the Sniper, but it always fails to prevent any. Charm Peddler might be an AI targeting issue, but I have no clue why Alabaster Wall isn't preventing it (I've killed the AI by just sitting there and passing turns with Kyren Sniper through the Alabaster Wall).
Still the fact that it's even tryign to use Charm Peddler in response to Kyren Sniper triggering is very odd, makes me thing something about either one of them is not scripted right.
Still the fact that it's even tryign to use Charm Peddler in response to Kyren Sniper triggering is very odd, makes me thing something about either one of them is not scripted right.
---
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
My Shandalar deck pack folder is avaliable here:Dropbox
Leave feedback on particular decks here: Google doc
Ask for instructions, give feedback and complaints here: Thread
Re: Forge Beta: 09-12-2014 ver 1.5.26
by excessum » 17 Sep 2014, 02:22
There are cards like Biovisionary and Azor's Elocutors that literally win the game with no specific "prioritizing" targets since all this is dependent on the game-state. Cards like Phage the Untouchable and Master of Cruelties have specific "must be blocked" AI flags since they are very likely to win the game immediately if said action ie. blocking, is not done.drdev wrote:lujo, please understand that it's not that nobody cares. In fact, we currently have a developer that's looking into doing a massive overhaul of the AI and game logic that will allow making the AI much much smarter than it is now, which would hopefully address issues like you're seeing.
In the meantime, would it be enough if we simply hard-coded a list of "kill on sight" cards that the AI should target over any others? We could even assign each a number that determines just how high a priority target that card is in case there's more than one present. That's probably the simplest and quickest thing we could do to hold people over until more involved changes can be done.
If we went that route, are there other cards besides Lin Sivvi you'd suggest we prioritize?
Lin Sivvi, Defiant Hero is only overpowered in your context, it can be safely ignored by pretty much every tournament-standard combo deck and quite a few AI decks so having a specific must-kill priority is not recommended.
Re: Forge Beta: 09-12-2014 ver 1.5.26
by Yosituna » 17 Sep 2014, 02:47
Description: I think I may have hit a limit for creatures and/or stack effects; in a three-player Commander game, I played Storm Herd with 53 life, with Aura Shards and Trostani, Selesnya's Voice (my commander) out. (I already had a few other Knight/Cat Soldier tokens in addition to the newly summoned army of Pegasi.) The next turn, I equipped Rhys the Redeemed with Illusionist's Bracers and used his second ability to double all my tokens twice; the first instance of the ability fired, at which point it promptly gave me this error. I'm not sure whether it's the creation of that many creatures that caused it, or the putting of that many triggers onto the stack (at rough math both together probably would have created ~180 creatures, which would mean ~360 triggers; even the first ability, where it crashed, would have created ~60 creatures and ~120 triggers).
Edit: It didn't actually crash, and selecting "Continue" let me continue the game with seemingly no problem, though there were a few oddities (Trostani changing power/toughness from the boosted 3/6, from Mirari's Wake, back to 2/5, and gaining/losing the ability added by Quest for Renewal).
Edit: It didn't actually crash, and selecting "Continue" let me continue the game with seemingly no problem, though there were a few oddities (Trostani changing power/toughness from the boosted 3/6, from Mirari's Wake, back to 2/5, and gaining/losing the ability added by Quest for Renewal).
- ConcurrentModificationException | Open
Re: Forge Beta: 09-12-2014 ver 1.5.26
by Agetian » 17 Sep 2014, 04:33
Got this NPE using the hotfix 1.5.28 snapshot from this thread.
Not sure if this is already fixed upstream by e.g. r27386, but I decided to report in case it's a different issue.
- Agetian
Not sure if this is already fixed upstream by e.g. r27386, but I decided to report in case it's a different issue.
- NullPointerException | Open
- Agetian
- Agetian
- Programmer
- Posts: 3488
- Joined: 14 Mar 2011, 05:58
- Has thanked: 683 times
- Been thanked: 570 times
Re: Forge Beta: 09-12-2014 ver 1.5.26
by drdev » 17 Sep 2014, 04:37
Pretty sure that issue was fixed by r27386.Agetian wrote:Got this NPE using the hotfix 1.5.28 snapshot from this thread.
Not sure if this is already fixed upstream by e.g. r27386, but I decided to report in case it's a different issue.Screenshot of the battlefield when it happened:
- NullPointerException | Open
- Agetian
- drdev
- Programmer
- Posts: 1958
- Joined: 27 Jul 2013, 02:07
- Has thanked: 189 times
- Been thanked: 565 times
Re: Forge Beta: 09-12-2014 ver 1.5.26
by tasteful » 17 Sep 2014, 04:54
Targeting Artisan of Forms with its own ability caused it to become a permanent with no types. It was no longer a creature but wasn't anything else, and simply stayed on the battlefield but couldn't do anything. Thassa was in play and I had clicked on Thassa when targeting Artisan's ability but it didn't work because Thassa was not a creature, so that might have had something to do with it?
Re: Forge Beta: 09-12-2014 ver 1.5.26
by excessum » 17 Sep 2014, 05:04
I'm pretty sure this problem is still around... viewtopic.php?f=52&t=6333&start=3165#p161207. Here is another example with r27480drdev wrote:Pretty sure that issue was fixed by r27386.
- Code: Select all
Forge Version: SVN
Operating System: Windows 7 6.1 amd64
Java Version: 1.8.0_20 Oracle Corporation
java.lang.NullPointerException
at forge.game.spellability.SpellAbilityRestriction.checkActivatorRestrictions(SpellAbilityRestriction.java:283)
at forge.game.spellability.AbilityActivated.isPossible(AbilityActivated.java:127)
at forge.ai.ComputerUtilMana$ManaProducingCard.<init>(ComputerUtilMana.java:115)
at forge.ai.ComputerUtilMana.sortManaAbilities(ComputerUtilMana.java:139)
at forge.ai.ComputerUtilMana.payManaCost(ComputerUtilMana.java:360)
at forge.ai.ComputerUtilMana.payManaCost(ComputerUtilMana.java:50)
at forge.ai.PlayerControllerAi.payManaCost(PlayerControllerAi.java:798)
at forge.game.player.PlayerController.payManaCost(PlayerController.java:275)
at forge.game.cost.CostPartMana.payAsDecided(CostPartMana.java:133)
at forge.game.cost.CostPayment.payComputerCosts(CostPayment.java:175)
at forge.ai.ComputerUtil.handlePlayingSpellAbility(ComputerUtil.java:104)
at forge.ai.PlayerControllerAi.playChosenSpellAbility(PlayerControllerAi.java:425)
at forge.game.phase.PhaseHandler.startFirstTurn(PhaseHandler.java:1027)
at forge.game.GameAction.startGame(GameAction.java:1559)
at forge.game.Match.startGame(Match.java:81)
at forge.control.FControl$4.run(FControl.java:496)
at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
at java.lang.Thread.run(Unknown Source)
Who is online
Users browsing this forum: No registered users and 25 guests