Angband 4.2.4

[Pete Mack]

The message only occurs when you cross the threshold. Going further over it is silent.

[malcontent]

Originally Posted by Pete Mack

The message only occurs when you cross the threshold. Going further over it is silent.

Ah, yes, my Knight's Shield of resistance had already sent me past the threshold. Thanks!

[backwardsEric]

Originally Posted by Sphara

Hi Nick.
Trap immunity (Wormtongue boots) doesn't apply when disarming chest traps. Summoning trap went off on my face. Intended or not?

If you had directly opened it (or, to use wobbly's phrase, booted it) the trap immunity would have bypassed the trap. To me, that's another argument, besides how it works for floor traps, for applying it when disarming.

[PowerWyrm]

Item ignoring setup is bugged, the tags are incorrect. Commit 3126eb4 missed a I2A -> all_letters_nohjkl change in tag_options_item() and added a pointless menu.selections in do_cmd_options_item() since the menu is managed by a menu_iter.

(expected: a-b-c-d-e-f-g-i-m-n-o-p-q-r-s-t-u-(blank)-Q-E-T; displayed: a-b-c-d-e-f-g-i-m-n-o-p-q-r-s-t-u-(v)-w-x-y)

[PowerWyrm]

Commit a318f61 removed the call to open_audio_hook(), surely this disabled sound completely no?

Also if you use ogg files and ogg format is supported but not mp3 format, code in open_audio_sdl() will now check for mp3 and reject sound instead of loading the ogg files.

[backwardsEric]

Originally Posted by PowerWyrm

Commit a318f61 removed the call to open_audio_hook(), surely this disabled sound completely no?

Also if you use ogg files and ogg format is supported but not mp3 format, code in open_audio_sdl() will now check for mp3 and reject sound instead of loading the ogg files.

There were two calls to open_audio_hook(), one after the other, in sound-core.c. That change you referenced only removed one of them. As for the sound format support, the added call to Mix_init() was to check that the SDL Mixer library has MP3 support because that's what Vanilla's and FAangband's sound files use. If a variant uses other formats, then it's likely best to 'formats = ...;' line to reflect the formats that the variant expects the SDL mixer to support or to use 'formats = 0;' to disable checking for any specific format (at the risk of triggering file descriptor leaks, see https://github.com/angband/angband/issues/5313 , if the sound system later tries to load a format not supported by the SDL mixer).

[PowerWyrm]

Originally Posted by backwardsEric

There were two calls to open_audio_hook(), one after the other, in sound-core.c. That change you referenced only removed one of them.

Check the change, that's not the case. First line is to check if the hook (pointer) exists:

if (!hooks.open_audio_hook) return 1;

Second line actually calls the hook and checks the return value:

if (!hooks.open_audio_hook()) return 1;

These are not the same calls. Removing the second line actually removes the call to open_audio_hook() and only leaves the pointer check.

[PowerWyrm]

Effect READ_MINDS returns false if no monster with a mind is found, which is not the same as any other detection/mapping effect which always return true. This has a weird side effect of preventing putting the effect on any object, because if there's no monster with a mind around, the object will not be consumed.

[PowerWyrm]

Originally Posted by backwardsEric

As for the sound format support, the added call to Mix_init() was to check that the SDL Mixer library has MP3 support because that's what Vanilla's and FAangband's sound files use.

The problem is that reading MP3 using the SDL Mixer library doesn't require calling Mix_init. In fact, on my WinXP VM, calling Mix_init(MIX_INIT_MP3) returns 0 (Does not support the required music formats) while I can play with sounds using MP3 files perfectly.

If there are file descriptor leaks, it must be another non related issue. For example, load_sample_sdl() calls Mix_LoadMUS() and stores the return value in the "loaded" boolean. Here we can't tell samples not loaded from samples where loading caused an error, maybe the boolean should be changed to an enum (not loaded, loaded, error). In case an error occured and the game tries to reload the sample, the "error" state should tell that the sample cannot be loaded and prevent calling Mix_LoadMUS() again.

[backwardsEric]

Originally Posted by PowerWyrm

The problem is that reading MP3 using the SDL Mixer library doesn't require calling Mix_init. In fact, on my WinXP VM, calling Mix_init(MIX_INIT_MP3) returns 0 (Does not support the required music formats) while I can play with sounds using MP3 files perfectly.

Out of curiosity, is that because that environment has a default external command configured to play the MP3 file (or perhaps it's related to this part of the changes list for SDL Mixer 2.0.3: "Fixed regression where Mix_Init() would return 0 for available music formats")? In any case, trying to Mix_Init() to workaround the resource leaks doesn't seem useful.