Re: [AD] possible bug in override_config_file

[Daniel Schlyder <daniel@xxxxxxxxxx>]

Eric Botcazou:
>> If I use set_config_file(/path/to/var/allegro.cfg) midi is not
>> initialised, as expected, but it is initialised if I instead use
>> override_config_file().

Correction: by not initialised, I meant set to MIDI_NONE.

> Could you post the sketch of your initialization code?

Of course, but when I tried to duplicate the problem in a simple recipe, it
turned out to be my code at fault. The problem was caused by me using both
override_config_file() and, later on, override_config_data(), as follows:

allegro_init();
override_config_file("/path/to/overrides.cfg");
std::string setting("[sound] \n flip_pan = 0");
override_config_data(setting.c_str(), setting.length());

After the call to override_config_data() the program "forgets" the values in
overrides.cfg. I guess this is intended behaviour. Maybe it wouldn't hurt to
spell this out in clear text in the docs, or maybe I just need a new brain. :)

The following code achieves what I wanted:

allegro_init();
push_config_state();
set_config_file("/path/to/overrides.cfg");
set_config_int("sound", "flip_pan", 0);
pop_config_state();
override_config_file("/path/to/overrides.cfg");

A bit ugly, but I can live with it.


BTW, I noticed that the code example in override_config_data() doc uses
multiline string. AFAIK, that feature is deprecated, and if I'm wrong, it's
still easy to miss the fact that it's a multiline string, and therefore not
think you need to separate category and variable names using newlines. The
attached patch (against 4.1.14) removes possible confusion.

Kind regards,
Daniel Schlyder
--
http://bitblaze.com/

Attachment: override_config_data_doc.diff
Description: Binary data