Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

84 Pages«<1415161718>»
Sepefeets Silverfish update (OG, Kara, MSG)
parakeet
#307 Posted : Sunday, June 26, 2016 2:11:32 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 128

Thanks: 18 times
Was thanked: 83 time(s) in 69 post(s)
@Dre Is this the version you want: http://www.filedropper.com/silverfish4
sepefeets
#306 Posted : Sunday, June 26, 2016 3:54:48 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 899

Thanks: 1028 times
Was thanked: 584 time(s) in 238 post(s)
parakeet;39050 wrote:
v121.2SE

Played Thunderbluff Valiant into Doomsayer. Could have killed the doomsayer, but instead, just ended the turn. Possibly related to the AI incorrectly thinking it had fatal vs a mage with a secret.

http://paste.ee/p/5ZOml

Yeah it doesn't understand iceblock immunity yet but HR's branch added it recently so I'll merge it sometime soon.
parakeet
#308 Posted : Sunday, June 26, 2016 4:05:38 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 128

Thanks: 18 times
Was thanked: 83 time(s) in 69 post(s)
@sepefeet Have you heard anything from Joy regarding the data sync issue? Do you have a sense of what needs to be done? I am tempted to try modifying the decompiled assembly HR code myself to see if we can make things sync.
sepefeets
#309 Posted : Sunday, June 26, 2016 5:23:33 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 899

Thanks: 1028 times
Was thanked: 584 time(s) in 238 post(s)
He has just said that he'll work on it.

To put it simply, HR needs to slow down. It is calling for a new move before the board has finished updating. iirc a year ago it was a minor problem that was barely noticeable (maybe 1/10 of misplays with the slime from sludge belcher deathrattle being the most obvious) but people complained about the bot being slow so Joy made it faster. It also affects temp buffs the most and temp buffs weren't as common in popular decks back then.
Dre
#310 Posted : Sunday, June 26, 2016 5:30:55 AM(UTC)

Rank: Advanced Member

Groups: HSModerator, Registered
Posts: 888
Man
Location: The Netherlands

Thanks: 216 times
Was thanked: 480 time(s) in 315 post(s)
Thanks Sepe and parakeet guiding me to earlier versions, sorry forgot there was a link in the first page for it :)

@sepe do you have any idea how the 2 last updates (121.1 and 121.2) totoaly screwed my WR with my face shammy deck. I'am not the only one. I can give you all my logs from last night again if you like. My settings didnt chance from when i was using v.121.0
sepefeets
#311 Posted : Sunday, June 26, 2016 5:42:32 AM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 899

Thanks: 1028 times
Was thanked: 584 time(s) in 238 post(s)
.1 had that problem with not wanting to attack with doomhammer, if you have problems with .2 then I need more precise info than a pile of logs to dig through. I already waste too much time digging through logs just to find out it was data sync. The changelogs are not just for show, they are things you should pay attention to because although it fixes the exact board that I tested, it could cause other situations with the card to go awry.
parakeet
#312 Posted : Sunday, June 26, 2016 1:42:43 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 128

Thanks: 18 times
Was thanked: 83 time(s) in 69 post(s)
@Dre If you go to github and click "Commits" you can see what changes were made in each version.

If it is just the Shaman decks that were impacted then it was probably the overload related changes. If all decks were impacted then maybe it was that "data = data.Trim();" though I am not sure how.

Dre or Scottbreak, can you test if it was just Shaman that was impacted?
sepefeets
#313 Posted : Sunday, June 26, 2016 3:33:02 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 899

Thanks: 1028 times
Was thanked: 584 time(s) in 238 post(s)
parakeet;39062 wrote:
If it is just the Shaman decks that were impacted then it was probably the overload related changes. If all decks were impacted then maybe it was that "data = data.Trim();" though I am not sure how.

FYI I'm 5 nines sure that if something is wrong that it's neither of those things. The overload stuff wasn't really used anywhere before to begin with (the places that need the info have it hardcoded instead) and the Trim is just to remove a newline in my new curdeck.txt file so that the deckname is matched correctly for the automatic settings switching.
parakeet
#315 Posted : Sunday, June 26, 2016 3:56:35 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 128

Thanks: 18 times
Was thanked: 83 time(s) in 69 post(s)
That makes sense. However, I don't see any other 121.1SE changes that would be so negatively impactive to the shaman decks.
anczej
#316 Posted : Sunday, June 26, 2016 4:33:58 PM(UTC)
Rank: Member

Groups: Registered
Posts: 27

Thanks: 2 times
Was thanked: 5 time(s) in 4 post(s)
In my opinion, its a problem with too many possibilites to play. Bot is trying to find out the best one and its time over. I saw that when me and my opponent had 5 minions each (bot didnt make any move).

I dont think thats the sepe Silverfish problem coz it starts with 5.9 HR update.
Dre
#317 Posted : Sunday, June 26, 2016 5:39:32 PM(UTC)

Rank: Advanced Member

Groups: HSModerator, Registered
Posts: 888
Man
Location: The Netherlands

Thanks: 216 times
Was thanked: 480 time(s) in 315 post(s)
In example what i experience after the 121.1 & 2 upgrade.

I use this combo:

Code:


// [Combo] [1] Tunnel Trogg (LOE_018) > [0] The Coin (GAME_005) + [3] Feral Spirit (EX1_248)
LOE_018,0;nxttrn;GAME_005,0;EX1_248,0;mana:1;bonusfirst:30;bonussecond:60;hero:shaman


Ive checks tens and tens of logs and this allways worked perfect. After the update it suddenly does this:



it ignored my combo and that never happened before in the previous versions. Trust me I checked this so many times. It allways did the combo of Tunnel trog > coin + feral.
anczej
#320 Posted : Sunday, June 26, 2016 6:24:56 PM(UTC)
Rank: Member

Groups: Registered
Posts: 27

Thanks: 2 times
Was thanked: 5 time(s) in 4 post(s)
On my board Totem golem + taunt totem and going with 3/2 totem to kill 2/1 minion.

http://pastebin.com/aM4W6vWx
sepefeets
#321 Posted : Sunday, June 26, 2016 6:51:00 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 899

Thanks: 1028 times
Was thanked: 584 time(s) in 238 post(s)
Dre - To be clear, where did you have your _combo.txt at?
parakeet
#323 Posted : Sunday, June 26, 2016 8:52:31 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 128

Thanks: 18 times
Was thanked: 83 time(s) in 69 post(s)
I will try HR ver5.8.2.0 with sepefeet's 121.2SE and see what happens
scottbreak
#314 Posted : Sunday, June 26, 2016 9:47:52 PM(UTC)
scottbreak


Rank: Advanced Member

Groups: Registered, HSModerator
Posts: 429

Thanks: 291 times
Was thanked: 469 time(s) in 153 post(s)
parakeet;39062 wrote:
@Dre If you go to github and click "Commits" you can see what changes were made in each version.

If it is just the Shaman decks that were impacted then it was probably the overload related changes. If all decks were impacted then maybe it was that "data = data.Trim();" though I am not sure how.

Dre or Scottbreak, can you test if it was just Shaman that was impacted?

Not just Shaman for me; also effected the Zoolock I have been working on.
parakeet
#324 Posted : Sunday, June 26, 2016 11:37:52 PM(UTC)

Rank: Advanced Member

Groups: Registered
Posts: 128

Thanks: 18 times
Was thanked: 83 time(s) in 69 post(s)
Okay sounds like anczej is probably correct then. Maybe Joy did some additional tweaks that are messing up the interaction between HR and silver.dll??

Still testing HR ver 5.8.2.0 with v121.2SE and Midrange Shaman. Starting from Rank 11.

Currently: Winrate 55% 35/63 games. Rank 8.
Dre
#322 Posted : Monday, June 27, 2016 1:57:22 AM(UTC)

Rank: Advanced Member

Groups: HSModerator, Registered
Posts: 888
Man
Location: The Netherlands

Thanks: 216 times
Was thanked: 480 time(s) in 315 post(s)
sepefeets;39075 wrote:
Dre - To be clear, where did you have your _combo.txt at?


Just in the Silverfish folder, I havent worked tet with the data map since iam just working on one deck atm.

So after updated (winrates are all well over 50 games):

121.0 59% winrate on rank 5
121.1 20% winrate drop till 9
121.2 32% winrate drop till 12

Now yesterday i rolled back,

121.0 56% winrate again

I didnt chance or move my settings, mulligan, combo and discovery in that time. Also the HR version in those days didnt change, using the latest one... Mystery :)
anczej
#325 Posted : Monday, June 27, 2016 3:10:01 AM(UTC)
Rank: Member

Groups: Registered
Posts: 27

Thanks: 2 times
Was thanked: 5 time(s) in 4 post(s)
HR version changed 3 days ago.
aznfury2
#327 Posted : Monday, June 27, 2016 4:56:24 AM(UTC)
Rank: Newbie

Groups: Registered
Posts: 5

Was thanked: 1 time(s) in 1 post(s)
I'll second the fact that the changes to the HR version somehow messed with sepefeet's. I tried rolling back but it didn't affect the winrate at all. I also tried using the default AI that's bundled with HR and it made a huge misplay where the enemy had lethal with a sorcerer apprentice and a leeroy, but the bot still attacked face (5 minions on board, any of which could've traded with leeroy or sorcerer to prevent lethal). Going to see if rolling back both will help.
Dre
#326 Posted : Monday, June 27, 2016 4:57:32 AM(UTC)

Rank: Advanced Member

Groups: HSModerator, Registered
Posts: 888
Man
Location: The Netherlands

Thanks: 216 times
Was thanked: 480 time(s) in 315 post(s)
anczej;39082 wrote:
HR version changed 3 days ago.


I know and then I still used 121.0 with 59% wr. V121.1SE released 2 days ago and .2SE 1 day ago. All 4 of my reported winrates are with the latest version of HR 5.9
84 Pages«<1415161718>»
Forum Jump  
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.

Powered by YAF | YAF © 2003-2011, Yet Another Forum.NET
This page was generated in 0.385 seconds.