Difference between revisions of "Worms Armageddon ReadMe (English)/v3.7.2.1 Update"

Latest revision as of 21:08, 25 August 2020

Fixes

• [CS] Disabled the unfinished crash log uploader, which was accidentally enabled in v3.7.0.0.
  
• [CS] W:A will now display a warning when attempting to select a Direct3D 9 renderer on systems where it could lead to excessive or continuous screen flicker (see section 4.0). [The root problem has been fixed in the following version, and is no longer listed in section 4.0.]
  
• [CS] W:A will now gracefully handle an integer division by zero crash in VMware's accelerated graphics implementation, and treat it as a graphics initialisation failure.
  
• [CS] W:A will now offer to switch to alternative graphics settings if the default ones result in a graphics initialisation failure.
  
• [CS] W:A will now suggest trying the Direct3D 9 (shader) graphics API to Windows 8 users when launched for the first time.
  
• [CS] W:A will now suggest enabling a Direct3D-based graphics API when Shift+Tab is pressed while using a DirectDraw-based API in the Steam edition. (The Steam overlay does not support DirectDraw.)
  
• [CS] The W:A update installer will now detect and offer to remove WormKitDS (dsound.dll), as its functionality was added in W:A in 3.7.0.0 in the form of the "Load WormKit modules" option on the Advanced Options screen. The built-in option is preferred above WormKitDS for improved stability and interoperability.
  
• [DC] Bug introduced in v3.6.30.0: After returning from a Deathmatch, the default scheme would be incorrectly set in Offline Multiplayer; it would claim to be Intermediate, but was set to 1 round to win instead of 2 rounds.
  
• [DC] In the v3.7.0.0 French translation, extra strings were added to six game comments, with the conditional wrapper "\{>=EE," "\}". This triggered the error message "Unknown character in conditional" due a bug in v3.7.0.0.
  
• [DC] In the v3.7.0.0 French translation, two strings were added to GAME_CRATEDROP_COMMENTS and two to GAME_SUDDEN_DEATH_COMMENTS, but each pair of comments was incorrectly entered, and would have been concatenated into a single comment if displayed.
  
• [DC] Minor spelling corrections have been applied to the English text.
  
• [CS] Minor grammar corrections have been applied to the Russian translation.
  
• The French translation has been extended by LTK, including all new strings added since v3.7.0.0, and a couple of minor corrections.
  
• [DC] The instant replay countdown timer, displayed in the upper-left corner of the screen during an instant replay in a mock-digital-watch font, incorrectly displayed the hundredths of a second as repeatedly counting down from 60 instead of from 100.
  
• [DC] Combo box bugs introduced in v3.7.0.0
  
  • [DC] Clicking the mouse button on the single-pixel-thick border of a combo box's list box closed the list box.
    
  • [DC] Releasing the mouse button on the single-pixel-thick border of a combo box's list box caused a stack overflow crash.
    
  • [DC] In the Host/Join screen, the down-arrow button on the scheme and intrinsic level combo boxes was accidentally and improperly widened by 4 pixels.
    
• [DC] Latent combo box bug introduced in v3.6.19.7: If clicking on an item that was already highlighted caused a dialog box to pop up, it was possible for a crash to happen upon dismissing the dialog box.
  
  • [DC] Bug made possible in v3.6.20.2: It was possible, but rare, for the "Nonstandard Weapon Power levels" warning dialog box to cause a crash upon being dismissed, if it had been triggered by selecting a scheme from the combo box in the Host screen.
    
  • [DC] Bug made possible in v3.6.30.0: It was possible for the "Large map warning" dialog box to cause a crash upon being dismissed, if dismissing it caused a map to be successfully loaded.
    
• [DC] If W:A was installed from the Czech CD-ROM before being Updated, the Team Editor would not work properly due to failing to load the gravestone bitmaps with Czech filenames. It would either crash upon entering the Team Editor, or show blank fields for everything except the team and worm names.
  
• [DC] Bug in a feature introduced in v3.6.29.0: The /dir command line parameter will no longer be overridden by the executable's directory, even if this results in W:A not being able to start properly.
  
• [DC] Bug introduced in v3.7.0.0: In the crash report file "errorlog.txt" there was a line ", version 3.7.0.0" which started with a comma.
  
• [DC] Bug introduced in v3.7.0.0: Most of the non-chat lines in User\Logs\*.log started with "???" instead of "•••".
  
• [DC] Bug in a feature introduced in v3.7.0.0: A potential crash / security vulnerability in custom language files has been fixed.
  
• [DC] Additional sanity checking is now done on data received while connecting to a host.
  
• [DC] Bug in a feature introduced in v3.6.28.0 and v3.6.29.0: In the Join screens, the /version command would show the incorrect version for a host using a version in the range 3.5 Beta 3pre10 - v3.6.27.3q. (Note: Since hosts using these versions do not report their version to the client, a client can only deduce a version range depending on what protocols and formats the host uses.)
  
• [DC] Bug in a feature introduced in v3.6.30.0: Using the characters "&'<>\ in game names on WormNET did not work properly, because support for them in v3.6.30.0 was implemented using translation of the corresponding HTML Entities, but by the time v3.6.30.0 was released, WormNET had changed the way it handled them, no longer outputting standard entities. Using these characters now works properly, and no longer decreases the maximum length of a game name. This is accomplished by translating the characters "&'<>\ to the non-printing characters 0x10, 0x11, 0x12, 0x13, 0x14, and 0x15 respectively.
  
• [DC] Bug in a feature introduced in v3.7.0.0: When starting WA.exe with /langdir parameter(s), combined with either "Auto-register associations" enabled or the /register parameter, the command line registered for wa:// URLs did not include the /langdir parameter(s).
  
• [DC] Bug introduced in v3.6.31.2b: Tall maps satisfying a certain inequality regarding their width and height could cause a crash upon starting a game. The formula in C-like pseudocode is: tilewidth=((height / 30) rounded down to nearest multiple of 8); if ((tilewidth - (width % tilewidth)) is sufficiently large) then crash upon starting a game.
  
• [DC] Bug in a feature introduced in v3.6.30.0: If the user pressed the Escape key in the host/join round-finished lobby but clicked "Cancel" in response to the Quit confirmation dialog box, then subsequent clicks of the "Exit" button would display the confirmation dialog box, even though the default behaviour of the "Exit" button in that screen is to instantly quit.
  
• [DC] Bug in a feature introduced in v3.6.24.2: With "Offline rope knocking" enabled, it was possible for a Deathmatch to be started with rope knocking enabled (which isn't supposed to happen). This would happen if the user started a Quick CPU game, then after finishing it, started a Deathmatch without first exiting the Single Player menu or closing W:A.
  
• [DC] When the "Create new team" button was used from the Training screen, the background particles behind the Team Editor moved twice as fast as they should.
  
• [DC] Bug introduced in v3.7.0.0: Using the /ignore command in the WormNET lobby (IRC) would cause an immediate crash.
  
• [DC] Bug in a feature introduced in v3.5 Beta 1: When minimising in-game, if any keys were held down at the moment of minimisation, then upon restoring, the game would continue to think the keys were held down, even if on the physical keyboard they weren't. Conversely, if any keys that weren't held down at the moment of minimisation were held down upon restoring, the game wouldn't necessarily detect that they were held down.
  
• [DC] Bug in a feature introduced in v3.7.0.0: When using a Direct3D 9 renderer, restoring after minimising in-game in single-step mode (during playback) would result in a black screen unresponsive to keyboard input (subsequently resuming normal operation required clicking a mouse button to restore the display, then again clicking a mouse button to restore keyboard focus).
  
• [DC] Bug in a feature introduced in v3.7.0.0: When using a Direct3D 9 renderer, restoring after minimising in the front end would not give focus to keyboard input until the mouse was clicked. In Windows Vista and later this could happen with all renderers, not just Direct3D, but focus was not always lost, just sometimes.
  
• [DC] Bug introduced in v3.6.30.0: When a host game dialog box was open (either by clicking "Host" in a WormNET channel or clicking "Direct TCP/IP" in the Network Choice screen with the IP address edit box left empty), pressing Alt+Tab once would cause the dialog box to lose focus. Subsequently using Alt+Tab or any other key to switch out of W:A, minimising it, would cause W:A to be in a buggy state upon being restored. In v3.6.30.0 and v3.6.31.0, this buggy state was a black screen; in v3.7.0.0 it was a snapshot of the screen as it was before entering the buggy state. Either way, the only way to get back to normal functionality was to press Esc enough times to exit the current dialog screen (i.e. WormNET or Network Choice, wherever the host game dialog box was opened from).
  
• [DC] Bug in a feature introduced in v3.6.19.17: Pressing Space to skip an instant replay did not work while in single-step mode.
  
• [DC] Bug in a feature introduced in v3.6.28.0: If the current map was a cavern with a width narrower than the current resolution, when the chat panel was visible, an area of matching height on the bottom of the screen would have graphical artefacts in its left and right sides, because the black letterbox area mistakenly excluded those regions.
  
• [DC] Bug in a feature introduced in v3.5 Beta 1: After minimising an offline game using Shift+Escape, animated effects that don't affect game logic (such as smoke and flames from explosions, smoke rising from Petrol, Homing Missile and Super Sheep trails, etc.) continued to be animated; as a result, by the time the game was restored, usually all of the effects would have already been finished and fizzled out, and in many situations would then be recreated all at once, putting the animations into an artificial-looking simultaneous phase until the passage of time allowed them to stabilise again into comfortable chaos. (The same thing has always happened when pressing Escape to pause, but as changing this would significantly affect the look and feel of the game unless done right, it has been left as-is for now.)
  
• [DC] Bug in a feature introduced in v3.5 Beta 1: When playing on a cavern map with a height smaller than the current resolution (i.e., a resolution high enough to result in letterboxing on the top and bottom), rising water caused a fluctuating inconsistency in the on-screen positioning of sprites and land. This effect was particularly obvious when using fast-forward, and when using "Playback at" with "Start Paused" checked.
  
• [DC] Bug in a feature introduced in v3.6.29.0: Export Log did not log the ending of the last turn (along with its turn time, retreat time, and damage dealt), if it was ended by locally quitting the game.
  
• [DC] Bug in a feature introduced in v3.6.30.0: When playing back a replay file that was recorded with a W:A version that records the passage of time between turns, if the file ended abruptly without an "exit" message at a point between turns, the end-of-game fade-out would be unnecessarily delayed until the beginning of the next turn.
  
• [DC] Bug introduced in v3.6.30.0: In-game bitmaps (including the health bars, text boxes, wind bar, and weapon panel) having a width and/or height of an odd number of pixels were drawn one pixel left and/or above where they should be, respectively. In most cases this is an arbitrary distinction, but not in the case of the turn, retreat, and round timers, which were being drawn one pixel up and to the left of where they should be within the timer box.
  
• [DC] Bug in a feature introduced in v3.6.20.1: In an offline replay created by the first version of W:A to save replays (v3.5 Beta 3pre7a), the message "Detected pre-2003.12.07 GMT game file (no checksums)" would be printed every time a checksum was found to be missing (this message was supposed to be printed only once).
  
• [DC] A fix in v3.6.26.4 prevented Mission Attempts counters from going over 127 and wrapping into the negative numbers, however, if playing missions with a version earlier than v3.6.26.4 already caused the counter to exceed 127, the latest version of W:A would still treat the number of Attempts as negative.
  
• [DC] When the Map Editor was in Preview mode, changing the water level did not immediately update the previewed positions of Placement Holes. Toggling the top (cavern) border did not update the preview either, even though it can change the positions of objects.
  
• [DC] In versions prior to v3.6.28.0, the water level of a map was limited to 50%, and setting it anywhere from 51% to 100% in the Map Editor would be equivalent to setting it to 50%. This limit was unwittingly removed in v3.6.28.0. One bug resulted from the presence of this limit in earlier versions, and another bug resulted from the removal of this limit:
  
  • [DC] The water level was never correctly previewed in the Map Editor. The 50% limit was not taken into account, and the on-screen water level preview had been calibrated as if maximum water level (which happened to be 50% of the map height) was only reached at 100%, i.e. as if the water level increased linearly from 0% to 100%; as a result the water level was displayed as if it were half of what it really was.
    
  • [DC] Bug introduced in v3.6.28.0: When a version earlier than v3.6.28.0 was emulated, and the map was set to a water level greater than 50%, the game would desynchronise immediately. Now, when hosting a game, if the water level is greater than 50% and there are client(s) present using a version earlier than v3.6.28.0, the game will not be allowed to start.
    
• [DC] Bug introduced in v3.6.28.0: When emulating a version earlier than v3.6.28.0, it was possible for a desynchronisation to happen under the following circumstances:
  
  • [DC] A Strike weapon is used, and by the time the plane disappears, all of the objects released by the strike are gone, and nothing else is in motion. There are at least two known ways this can happen:
    
    • The scheme is set to zero retreat time, and a Strike weapon, set to "zero power" in the scheme, is used. Since it releases zero missiles, no trickery is needed to get them to disappear quickly. The next weapon use will desync.
      
    • The scheme is set to less than 3 seconds retreat time; an Air Strike with 1 star of power is aimed at three towers of Frozen Worms piled at the top of the map near the left edge, and all four missiles hit the towers near their tops. By the time the plane disappears, the bouncing motion of the frozen worms has damped down to zero. The next weapon use will desync.
      
  • [DC] A missile with "zero power" is created in a position where it already intersects with something that would make the missile explode upon impact, after the worm responsible for firing the weapon is already idle. There is no known way this could happen.
    
• [DC] Fixes affecting game logic
  
  • [DC] It was possible to place a Girder partially underwater if the water level was higher than the bottom of the map. Subsequent water rising would only erase the part of the girder that was initially above water, leaving an invisible obstacle in the water.
    
  • [DC] The game would crash if a CPU Worm, while enumerating possible places to which to walk, reached its maximum search depth at a point in its path where it would fall and retain control. (See Footnote 2.)
    
  • [DC] Bug introduced in v3.5 Beta 2: CPU Worms would always shoot Strike weapons to the right, and never to the left.
    
    • [DC] Bug in a feature introduced in v3.6.26.4: The absence of this bug in v3.0 was not emulated.
      
  • [DC] The glitch in which targetable weapons (such as Homing Missile, Air Strike, and Teleport) could be fired without being targeted has been modified. It now only works in the form of pre-targeting, which must be done on the same turn.
    
  • [DC] The v3.6.31.0 Fix, "The game would sometimes crash during the worm death explosion animation, if the Russian language was selected", changed what comments in GAME_LAND_DEATH_COMMENTS would be randomly displayed for players who had the Russian language selected, but did not implement emulation for what comments were displayed in earlier versions. Emulation for this is now implemented.
    
  • [DC] Bug in a feature introduced in v3.6.28.0: On a map of width greater than about 32304, shooting a rope off the right edge could result in the endpoint of the rope wrapping around to the left side, resulting in its line segment being incorrectly drawn. (The same thing could happen on a map of height greater than about 32304, shooting a rope down into the ocean.) (Strictly speaking this fix doesn't affect the game logic, but it is tied to the logic version so that all players will see the same thing.)
    
  • [DC] Bug in a feature introduced in v3.6.28.0: When playing on a map of height 32568 or greater, objects falling into the water would sink forever without drowning, preventing the turn from ending. Now, the visible height of the ocean will be decreased in such situations to allow objects to drown normally.
    
    • [DC] When this bug occurred, it was always accompanied by another one: The current worm was not tracked by the camera. This bug is now fixed without being tied to the logic version (i.e., it won't be emulated when the other bug is emulated).
      

Features

• [DC] Reopening games hosted on WormNET
  
  • It is now possible to reopen a WormNET game after a match has been finished, using the /open command in the main Host screen. A notification message will be shown to all players.
    
  • The host is now notified when 5 minutes have passed since the game has been hosted, at which point WormNET automatically deletes the game. The /open command can be used to reopen it in this case as well. (The notification messages are shown only to the host in this case.)
    
  • Note that third-party WormNET servers may not implement an automatic timeout, or may have a timeout of a different duration. W:A currently assumes any server it connects to has a 5 minute timeout, and behaves accordingly.
    
• [DC] The names of the animated gravestones (e.g. "[ Gravestone 1 ]") in the Team Editor are now localised.
  
• [DC] The output of the Host/Join /version command is now logged in User\Logs\*.log files.
  
• [DC] Export Log now logs the total number of turns of each team, now printed alongside the turn time and retreat time totals.
  
• [DC] The game no longer checks for the Tab key being pressed unless the current worm is in worm selection mode, and the key is now checked before all other keys, allowing the selected worm to be moved on the same frame as it is selected.