FAangband 2.0beta

Collapse
X
 
  • Time
  • Show
Clear All
new posts

  • MITZE
    replied
    Bug Spotted?

    Perhaps a stupid question (because I really don't know newFA code), but should line 719 in mon-make.c be false rather than true?

    As far as I understand it, the entire function it's in is returning true or false based on whether it's successfully created a player ghost, but that particular line is part of a check to determine whether the monster is supposed to be a player ghost at all?
    Last edited by MITZE; July 8, 2021, 05:17. Reason: Fixed formatting.

    Leave a comment:


  • Nick
    replied
    Originally posted by Cuboideb
    I tried this fix. Bone files are properly generated now in android.
    Thanks, I've done an overhaul of that code which includes your fix, and I think it now works properly.

    Crashes I will get to another time.

    Leave a comment:


  • MITZE
    replied
    Originally posted by Nick
    OK, I've found a memory error in wilderness and updated - see if that makes a difference.
    Both crashes still happen.
    Last edited by MITZE; July 5, 2021, 20:48. Reason: Fixed formatting.

    Leave a comment:


  • Cuboideb
    replied
    Originally posted by Cuboideb
    In ui-death.c, line 353. file_close should be outside the infinite loop. If the user answers N that line could never be executed. Perhaps the loop should be closed at line 337.
    I tried this fix. Bone files are properly generated now in android.

    Leave a comment:


  • Nick
    replied
    OK, I've found a memory error in wilderness and updated - see if that makes a difference.

    Leave a comment:


  • MITZE
    replied
    Originally posted by Nick
    Did you get any instances of the game hanging?
    No lockups for me, though there were a few instances of the transition taking a non-negligable amount of time to happen; this may or may not be a webserver consequence rather than a game issue though.

    (Incidentally, I had been wondering if Rolanako's lockup had just been him coincidentally running into connection issues at the same time--no way to tell now, though.)
    Last edited by MITZE; July 5, 2021, 07:35. Reason: Fixed formatting.

    Leave a comment:


  • Nick
    replied
    Originally posted by MITZE
    As far as I can tell, the crash that deposits you on the level you were trying to enter seems to have something to do with player ghosts; as far as I've noticed, every time it's happened there's been a player ghost when I've reloaded the game. In case this is valuable information for debugging, I've been scumming for these crashes going between Amon Rudh 15 and West Beleriand 16; the player ghosts I've confirmed involved when the crash happens are 'BEEFCAKE', 'Dex', 'Judias', 'Klassik', 'Rythil', and 'WildKhaine'.

    As for the crash that deposits you on the level you were trying to leave, it's only happened to me once or twice during my bugtesting time, and it hasn't been replicatable; I've been making a habit of saving before attempting to make a transition during this, and when this crash happened it didn't occur when I tried the transition a second time--not sure if saving screwed with the RNG or what.

    EDIT: Confirmed the crash that deposits you on the level you were trying is not replicatable; spammed wilderness transitions without saving for a while and found it again, but it still didn't occur when trying the transition a second time.
    This is all really useful information. I suspect (well, hope) finding the first issue won't be too hard now, as it's almost certainly related to ghost generation, and apparently after level generation. The second one is happening during level generation, so may be harder to find.

    Did you get any instances of the game hanging?

    Leave a comment:


  • Sacksquatch
    replied
    My warrior BEEFCAKE still wrecking shit in the afterlife!

    Leave a comment:


  • MITZE
    replied
    Wilderness Transition Crashes

    Okay, so not sure whether this is good news or bad news, but I've gone back to bugtesting on the webserver, and am having plenty of luck scumming for the wilderness transition crashes now.

    Like the issues already submitted on Git about it seem to reflect, I suspect there are two separate issues (or perhaps one issue with two failure points?) here; one causes a crash that deposits you on the level you were trying to transition to, and the other deposits you on the level you were trying to leave. The latter seems to be rarer than the former.

    As far as I can tell, the crash that deposits you on the level you were trying to enter seems to have something to do with player ghosts; as far as I've noticed, every time it's happened there's been a player ghost when I've reloaded the game. In case this is valuable information for debugging, I've been scumming for these crashes going between Amon Rudh 15 and West Beleriand 16; the player ghosts I've confirmed involved when the crash happens are 'BEEFCAKE', 'Dex', 'Judias', 'Klassik', 'Rythil', and 'WildKhaine'.

    As for the crash that deposits you on the level you were trying to leave, it's only happened to me once or twice during my bugtesting time, and it hasn't been replicatable; I've been making a habit of saving before attempting to make a transition during this, and when this crash happened it didn't occur when I tried the transition a second time--not sure if saving screwed with the RNG or what.

    EDIT: Confirmed the crash that deposits you on the level you were trying to leave is not replicatable; spammed wilderness transitions without saving for a while and found it again, but it still didn't occur when trying the transition a second time.
    Last edited by MITZE; July 5, 2021, 07:40. Reason: Fixed formatting.

    Leave a comment:


  • Nick
    replied
    Originally posted by Cuboideb
    In ui-death.c, line 353. file_close should be outside the infinite loop. If the user answers N that line could not executed. Perhaps the loop should be closed at line 337. I will test a fix in android tomorrow.
    Oops. Thanks

    Leave a comment:


  • Cuboideb
    replied
    In ui-death.c, line 353. file_close should be outside the infinite loop. If the user answers N that line could not executed. Perhaps the loop should be closed at line 337. I will test a fix in android tomorrow.

    Leave a comment:


  • Nick
    replied
    Originally posted by MITZE
    Bah, was hoping some of that stuff would include like a depth, to give me a lead where to look in-game.
    Oh, I see. I failed to mention that the file name gives the depth - bone.060 is level 60, etc.

    Originally posted by MITZE
    FWIW testing offline I am finding no player ghosts, but no crashes/lockups either (OS Windows 10 Home v10.0.19042, on a Surface Go 2 tablet).
    Good (well, mostly). I have been struggling to find crashes/lockups as well, and at some point have to decide to release with some herd-to-find bugs still present.

    Leave a comment:


  • MITZE
    replied
    Originally posted by Nick
    player ghosts
    Bah, was hoping some of that stuff would include like a depth, to give me a lead where to look in-game.

    FWIW testing offline I am finding no player ghosts, but no crashes/lockups either (OS Windows 10 Home v10.0.19042, on a Surface Go 2 tablet).
    Last edited by MITZE; July 5, 2021, 02:21. Reason: Fixed formatting.

    Leave a comment:


  • Nick
    replied
    Yeah, player ghosts involve both writing and reading from files, and attempt to do it across platforms and on single install multiple player systems like live. So there are like a million things that can go wrong.

    The format of bone files is:
    Code:
    Name
    Race index
    Class index
    String index: String
    where the string index is 0 for no string, 1 for string as something the ghost says, 2 for string as a description.

    Leave a comment:


  • Cuboideb
    replied
    New bones are generated with 0 bytes in android. I have to test if its something related to android.

    Leave a comment:

Working...
😀
😂
🥰
😘
🤢
😎
😞
😡
👍
👎