*** ricardocrudo has quit IRC | 00:13 | |
*** ricardocrudo has joined #lv2 | 00:14 | |
*** ricardocrudo has quit IRC | 01:23 | |
*** falktx has quit IRC | 02:12 | |
*** artfwo has quit IRC | 02:44 | |
*** deva has joined #lv2 | 04:58 | |
*** sigma6 has joined #lv2 | 07:12 | |
*** falktx has joined #lv2 | 07:54 | |
*** ssj71 has quit IRC | 08:13 | |
*** edogawa has joined #lv2 | 09:15 | |
falktx | with 800 lv2 plugins lilv world gets quite big | 10:40 |
---|---|---|
falktx | testing some new code I get | 10:40 |
falktx | total heap usage: 938,349 allocs, 938,349 frees, 67,763,681 bytes allocated | 10:40 |
rgareus | falktx: I think it's presets that cause most of this. | 11:01 |
rgareus | and it's a lot better than it used to be. here's maybe 100 plugins with old sord tree: http://robin.linuxaudio.org/tmp/lv2_mem_dbg2.txt (115MB) | 11:04 |
rgareus | and new sord http://robin.linuxaudio.org/tmp/a3_mem_dbg_newsord1.txt (40MB) | 11:04 |
rgareus | ^^ SVN | 11:05 |
rgareus | ardour drops the lilv world after initial scan and caches relevant parts. hence the steep drop in the lv2_mem_dbg2.txt | 11:05 |
*** ricardocrudo has joined #lv2 | 12:07 | |
*** artfwo has joined #lv2 | 12:30 | |
drobilla | Presets shouldn't be loaded unless actually needed | 15:10 |
drobilla | If you load absolutely everything, then yeah, it'll get pretty big | 15:10 |
falktx | I need to know the uri and name of the presets | 15:48 |
falktx | can you know the name without loading the preset? | 15:48 |
drobilla | A few things would help that I can't do without breaking compatibility. There's at least one useless field in SerdNode, currently Lilv allocates its own nodes rather than exposing the Sord ones which are pretty much equivalent (though this is quite low on the profile), and there's usually more indexing around than is actually necessary | 15:48 |
drobilla | A profile for this case would be interesting (ardour is big) | 15:49 |
drobilla | falktx: No, but if it really matters you can load, get name, and unload | 15:49 |
drobilla | Requiring label in manifest may have been a good idea, but with translation that could blow up, and it seems like a slippery slope... | 15:49 |
*** sigma6 has quit IRC | 16:12 | |
*** jvfc678 has joined #lv2 | 16:34 | |
*** falktx has quit IRC | 16:34 | |
*** jvfc678 has quit IRC | 16:35 | |
*** son0p has quit IRC | 16:39 | |
*** ssj71 has joined #lv2 | 16:40 | |
*** son0p has joined #lv2 | 16:40 | |
*** ricardocrudo has quit IRC | 17:11 | |
*** falktx has joined #lv2 | 17:32 | |
*** uncle-j_j has joined #lv2 | 17:41 | |
*** uncle-j_j has quit IRC | 17:53 | |
*** uncle-j_j has joined #lv2 | 17:53 | |
*** son0p has quit IRC | 19:01 | |
*** son0p has joined #lv2 | 19:03 | |
*** deva has quit IRC | 19:07 | |
*** ricardocrudo has joined #lv2 | 19:52 | |
*** rncbc has joined #lv2 | 20:36 | |
*** ricardocrudo has quit IRC | 21:02 | |
*** falktx has quit IRC | 22:01 | |
*** edogawa has quit IRC | 22:12 | |
*** rncbc has quit IRC | 22:26 | |
*** Magnus_RM has joined #lv2 | 23:23 | |
*** uncle-j_j has quit IRC | 23:36 |
Generated by irclog2html.py 2.13.0 by Marius Gedminas - find it at mg.pov.lt!