| *** edogawa has quit IRC | 00:04 | |
| *** NickSB2 has joined #lv2 | 02:58 | |
| *** NickSB2 has quit IRC | 04:26 | |
| *** velho has quit IRC | 07:12 | |
| *** grejppi_ has joined #lv2 | 07:51 | |
| *** edogawa_ has joined #lv2 | 08:23 | |
| *** edogawa_ is now known as edogawa | 08:54 | |
| *** falktx has joined #lv2 | 10:20 | |
| *** falktx has quit IRC | 10:29 | |
| *** NickSB2 has joined #lv2 | 11:15 | |
| *** gianMOD has joined #lv2 | 12:00 | |
| *** grejppi_ has quit IRC | 12:52 | |
| *** gianMOD has quit IRC | 13:12 | |
| *** velho has joined #lv2 | 13:36 | |
| *** NickSB2 has quit IRC | 13:44 | |
| *** grejppi has quit IRC | 14:04 | |
| *** gianMOD has joined #lv2 | 16:22 | |
| *** HarryHaaren has joined #lv2 | 16:56 | |
| *** gianMOD has quit IRC | 17:09 | |
| *** gianMOD has joined #lv2 | 17:20 | |
| *** gianMOD has quit IRC | 17:25 | |
| *** gianMOD has joined #lv2 | 17:41 | |
| *** gianMOD has quit IRC | 17:59 | |
| *** drobilla has quit IRC | 18:06 | |
| *** drobilla has joined #lv2 | 18:07 | |
| *** mlpug has joined #lv2 | 18:10 | |
| *** HarryHaaren has quit IRC | 18:13 | |
| *** HarryHaaren has joined #lv2 | 18:30 | |
| HarryHaaren | drobilla, https://github.com/harryhaaren/pugl there is Cairo double buffering available there. Its a small patch: but there's no option of single-buffering anymore... I don't think anybody will miss it? | 18:35 |
|---|---|---|
| HarryHaaren | there are some transparency issues to discuss; when do we clear the background? Drawing transparent regions over eachother will "add-up" alpha | 18:36 |
| HarryHaaren | but I'm drawing a bg rectangle using Cairo, so it doesn't affect me right now (aka hack-fixed) | 18:37 |
| HarryHaaren | the pugl_cairo_test isn't affected, as no alpha is used in drawing stuff | 18:37 |
| HarryHaaren | that's all for now, chat anoter time! | 18:37 |
| * HarryHaaren away | 18:37 | |
| drobilla | HarryHaaren: I guess it could be slower, but I doubt anyone will miss it | 18:38 |
| drobilla | Worse case, puglInitDoubleBuffering, which also applies to GL | 18:38 |
| drobilla | HarryHaaren: I'll check it out, thanks | 18:39 |
| drobilla | Probably merge that first then try to better separate the cairo/GL stuff ala earlier conversation with rgareus | 18:39 |
| HarryHaaren | its a single paint, so slowdown will be minimal. Cool. You're welcome | 18:39 |
| rgareus | HarryHaaren: does your pactch support partial exposure? or do you always clear the whole thing? | 18:41 |
| HarryHaaren | right now, we're blitting the whole, just like the previous cairo implementation | 18:41 |
| HarryHaaren | there's no reason why in future it can't be made to partial-blit though | 18:41 |
| HarryHaaren | the framework is in place, it just needs doing | 18:42 |
| rgareus | HarryHaaren: I've already done that :) | 18:42 |
| HarryHaaren | rgareus, branch or master? your .git or merged in drobilla's master? | 18:42 |
| rgareus | no | 18:43 |
| HarryHaaren | rgareus, so are you handling things at widget-level then? RobTK i'm guessing? | 18:43 |
| * HarryHaaren just built AVTK "2.0".. no partial exposure yet, but not far off either | 18:43 | |
| rgareus | HarryHaaren: I diverged a lot of upstream pugl - or rather upstrem pugl diverged. | 18:43 |
| HarryHaaren | ah yes, i remember that discussion | 18:44 |
| HarryHaaren | rgareus, double-buffering is literally a few lines: https://github.com/harryhaaren/pugl/commit/9b805430f5a5f5081be7e857bc635772cd5cd429 | 18:44 |
| rgareus | mmap a cairo software surface to a GL texture. let GL do the double buffering. | 18:44 |
| HarryHaaren | this doesn't require GL at all, Cairo -> Xlib | 18:44 |
| *** NickSB2 has joined #lv2 | 18:44 | |
| rgareus | HarryHaaren: you'll find that cairo_surface_create_similar performs poorly on most systems | 18:44 |
| rgareus | sadly we're not living in an ideal world. | 18:45 |
| HarryHaaren | it "feels" fine here. Yeah that. | 18:46 |
| HarryHaaren | but in future it could be changed to the double-buffering method you've described: and it would be transparent to the user of the PUGL API (assuming your + master gets merged somehow-eventually?) | 18:46 |
| drobilla | You'll also find that GL performs poorly on many systems. See above Re: ideal world :) | 18:47 |
| rgareus | HarryHaaren: yeah it it fine here on my laptop as well. - but ask the_CLA to run it on his AMD graphics card and oofus with nvidia binary blob.. | 18:47 |
| HarryHaaren | so right now I'm gonna focus on making shiny UI's.. and then if the problem of speed arises i'll engage then | 18:47 |
| rgareus | HarryHaaren: then check on a windows system (no 32bit RGBA, cairo does BitBlt) | 18:47 |
| drobilla | But yeah, that. If cairo-via-GL is a better way, easy enough. | 18:47 |
| HarryHaaren | "i'll burn that bridge when I get to it" ;) | 18:48 |
| rgareus | but windows + GL can do ARGB32 | 18:48 |
| *** zth has joined #lv2 | 18:48 | |
| HarryHaaren | this can be solved in future: right now, its more important to me to make Fabla 2.0 happen | 18:48 |
| rgareus | yep, as long as your UI uses cairo, you'll be fine | 18:49 |
| rgareus | what's under the hood cairo -> hardware can change | 18:49 |
| HarryHaaren | yep. Awesome. | 18:49 |
| * HarryHaaren back to coding stuff | 18:49 | |
| rgareus | the cairo guys are working on improving their 'native' surfaces | 18:49 |
| bgola | do you think it's a bad thing if mod-host uses jack to manage the routing between CV ports? | 18:51 |
| rgareus | bgola: quite the opposite | 18:52 |
| rgareus | bgola: in an ideal world, there'd never have been jack-midi-ports but jack-control-ports (which could include MIDI) | 18:52 |
| rgareus | bgola: in your case it may even be easier to re-use audio float ports | 18:53 |
| rgareus | (as long as that jack is confined and a user won't be able to connect those ports to audio-out) | 18:54 |
| HarryHaaren | drobilla, little "warning" note that currently my Cairo stuff doesn't clean-up properly, i'm just grabbing latest git from you (which needs an <stdint.h> and some other minor fixes).. and i'll include some cleanup for the Cairo stuff too | 18:54 |
| drobilla | CV isn't events | 18:54 |
| bgola | rgareus: that's the thing... for the mod hardware it is not a problem, user never touches jack connections | 18:55 |
| HarryHaaren | not on the MOD, but if MOD is running on desktop... | 18:55 |
| drobilla | Worked around to get pseudo-CV with metadata in the same way we worked around the brainded MIDI fuck-up, though | 18:55 |
| bgola | HarryHaaren: exactly | 18:55 |
| rgareus | drobilla: why not? semantically N events key/value (with N <= samples-per-period) does model it best. (though audio-ports more pactical) | 18:56 |
| drobilla | rgareus: With CV metadata is sensible, though potentially dangerous since unaware clients will consider it audio | 18:58 |
| drobilla | rgareus: with MIDI it's literally MIDI. to get anything else you have to pack other-thing in sysex | 18:58 |
| drobilla | rgareus: For CV you basically need a straight up audio buffer. There'd be an 'event' per sample | 18:59 |
| rgareus | using audio-ports also has the advantage that you can connect external equipment | 18:59 |
| drobilla | That's essentially what CV is. However, yes, in the overwhelming majority of cases, events are the proper thing | 19:00 |
| rgareus | IFF you have a soundcard that allows to bypass the hi-pass filter. | 19:00 |
| drobilla | Yes. Even in LV2 a different "type" was kinda sorta a mistake, maybe, ish | 19:00 |
| *** mlpug has quit IRC | 19:00 | |
| bgola | ... i suppose there is no way to tell jack not to expose a port to other clients.. apparently it's only a problem if users try to connect cv to audio | 19:06 |
| HarryHaaren | nope. I guess it could be included in the metadata of the port, but that API is pretty new and I don't think any connection-manager (patchage, QjackCtl etc) would understand the meta-data | 19:09 |
| rgareus | bgola: the most pragmatic is to name the ports "do_not_connect_manuall_or_risk_the_wrath_of_culu__CV_port-XXX" :D | 19:10 |
| drobilla | Patchage (um, at least in svn, I forget if I released that) does know what they are and colour them, but makes no attempts at safety | 19:11 |
| HarryHaaren | ah my bad drob, i'll admit i dont have latest, and I didn't test it but assumed. | 19:11 |
| HarryHaaren | on a side note, meta-data can be used to understand "more" about a port, but should not generally be used in the "do-less with this" way, for backwards compat with existing systems | 19:12 |
| drobilla | In a patch bay it could be an option or a warning dialog or whatever | 19:14 |
| drobilla | But in general, no real way of getting around it. Jack "audio" ports exist that aren't audio | 19:14 |
| *** mlpug has joined #lv2 | 19:14 | |
| drobilla | If Jack had metadata from the get go lots of things would be better, but c'est la vie | 19:15 |
| bgola | is this the reason jalv doesnt support cv ports? | 19:18 |
| drobilla | Nah, I'm just lazy. :) | 19:27 |
| drobilla | Although sort of, in some sense, because they require explicit support. | 19:27 |
| drobilla | (I decided this was a Good Thing for the "danger" reasons, and since inputs you sometimes want to show as controls, not audio) | 19:27 |
| *** HarryHaaren has quit IRC | 19:49 | |
| *** zth has quit IRC | 20:04 | |
| *** ricardocrudo has joined #lv2 | 20:06 | |
| *** mlpug has quit IRC | 20:59 | |
| *** ricardocrudo has quit IRC | 22:23 | |
| *** NickSB2 has quit IRC | 22:37 | |
| *** falktx has joined #lv2 | 22:43 | |
| *** falktx has quit IRC | 22:49 | |
| *** grejppi has joined #lv2 | 23:11 | |
| *** edogawa has quit IRC | 23:15 | |
Generated by irclog2html.py 2.13.0 by Marius Gedminas - find it at mg.pov.lt!