| *** edogawa has quit IRC | 00:13 | |
| *** ricardocrudo has quit IRC | 00:41 | |
| *** Magnus_RM has quit IRC | 00:43 | |
| *** drobilla has joined #lv2 | 00:47 | |
| drobilla | rgareus: Should fil4 chew a massive amount of CPU? | 01:03 |
|---|---|---|
| drobilla | rgareus: It acts really strangely in jalv.gtk here, show up, but all the Jack ports are nuked immediately | 01:03 |
| drobilla | then stalls on quit | 01:03 |
| drobilla | Hm, works in Ingen though | 01:09 |
| *** artfwo has quit IRC | 02:40 | |
| *** Guest78532 has joined #lv2 | 05:09 | |
| *** Guest78532 has left #lv2 | 06:08 | |
| *** falktx has joined #lv2 | 06:52 | |
| *** falktx has quit IRC | 07:15 | |
| *** sigma6 has joined #lv2 | 08:03 | |
| rgareus | drobilla: on your machine, I'd say it should not take more that 3% for all buffersizes > 64fpp. (unless you enable spectrum analysis) | 08:44 |
| *** falktx has joined #lv2 | 08:44 | |
| *** EntropySink has quit IRC | 09:12 | |
| *** EntropySink has joined #lv2 | 09:16 | |
| *** edogawa has joined #lv2 | 09:35 | |
| *** uncle-j_j has joined #lv2 | 09:37 | |
| *** uncle-j_j has quit IRC | 11:03 | |
| *** uncle-j_j has joined #lv2 | 11:04 | |
| *** ricardocrudo has joined #lv2 | 11:21 | |
| *** ricardocrudo has quit IRC | 11:23 | |
| *** uncle-j_j has quit IRC | 12:23 | |
| *** uncle-j_j has joined #lv2 | 12:24 | |
| *** ricardocrudo has joined #lv2 | 12:44 | |
| *** ricardocrudo has quit IRC | 12:47 | |
| *** artfwo has joined #lv2 | 12:47 | |
| *** aombk2 has joined #lv2 | 12:59 | |
| *** aombk has quit IRC | 13:02 | |
| drobilla | rgareus: Odd. Something funny about how the UI loop is driven in Jalv, I'd guess, though Jack dropping it suggests something audio. I'll check into it later | 14:07 |
| drobilla | rgareus: Looks shiny though, and that's what really matter :) | 14:07 |
| rgareus | uhm. I beg to differ | 14:09 |
| rgareus | the fancy UI is only a bonus. | 14:09 |
| drobilla | That's a joke :P | 14:09 |
| rgareus | I'll make an animated gif with waterfall for lv2plug.in frontpage. | 14:10 |
| * rgareus tests in jalv again | 14:11 | |
| rgareus | works fine | 14:12 |
| rgareus | with recent jalv svn | 14:13 |
| drobilla | ... Though on that, visual nitpick on said ultra important trivialities, the light gradient buttons stand out and don't really fit in | 14:14 |
| drobilla | Maybe it's the dual ramps, a linear one like the knobs used on the buttons as well (with different colours as appropriate) would make things look better I think | 14:15 |
| drobilla | rgareus: Could be my pending control changes, but I don't see how | 14:15 |
| drobilla | rgareus: Buuuuuuuuuut work time. | 14:15 |
| * drobilla is soooooooooo incredibly tired of not being able to work on the things he wants to work on | 14:16 | |
| falktx | machina time? | 14:16 |
| drobilla | falktx: Turning that into an LV2 is something that needs to happen, but that's somewhere down the hierarchy in the "things that are cool but nobody else cares about yet" section :) | 14:20 |
| rgareus | drobilla: control-changes? | 14:20 |
| falktx | we're lacking midi generators/sequencers | 14:20 |
| drobilla | Currently control change stuff, yes, which I need to commit already | 14:21 |
| falktx | I do care about those, but they are very hard to do right | 14:21 |
| drobilla | ... though thinking about putting that in Ingen gave me pause, because all of a sudden plugins have controls that don't have symbols | 14:21 |
| drobilla | and inputs/outputs having symbols is a pretty key LV2 thing | 14:21 |
| rgareus | drobilla: the fil4 GUI performs a hand-shake with the DSP (using float control-ports) to hand over and acknowledge peak-hold. | 14:22 |
| drobilla | Also compilation in Ingen, where I currently have a thing that compiles a parallelism-aware graph, but no execution yet... but thinking about conditional switches | 14:22 |
| drobilla | and proper graph presets | 14:22 |
| drobilla | Somewhere off in fantasy land I want a control panel builder, but I'm trying not to think about that :) | 14:23 |
| drobilla | This parameter thing is a bit important since I'm not sure I really even want Ingen to be making control ports at all, at least not internally (i.e. only as an external compatibility layer sort of thing) | 14:24 |
| drobilla | buuuuut the whole thing is completely designed around the "everything has a path" idea | 14:25 |
| drobilla | Which, as it happens, came largely from OSC, which is also a topic of interest around lately | 14:25 |
| drobilla | http://drobilla.net/files/jalv_parameters.png | 14:27 |
| drobilla | (spring is a control that slowly bounces back to 0 when you change it) | 14:27 |
| falktx | so a slowtrigger? | 14:29 |
| drobilla | Hm? | 14:31 |
| falktx | lv2 already has a trigger. but it's for 1 single run | 14:32 |
| drobilla | Ah | 14:33 |
| drobilla | Well, this is just a float control that uselessley changes internally for the purposes of testing | 14:33 |
| drobilla | But good point, an "impulse" is a missing one. Current discussion on lv2-dev about that | 14:33 |
| drobilla | (We don't have a fundamental atom type for that, kind of on purpose) | 14:34 |
| drobilla | So it'd be a momentary button | 14:34 |
| drobilla | But that one's a bit weird... a parameter has a value, but an impulse doesn't "stick" like a value | 14:35 |
| drobilla | Which is a distinction totally lacking in the spec, come to think of it | 14:36 |
| drobilla | It *could* apply to other value types, too. You could have a sequence that accepts float, but the float is not sticky like a value | 14:36 |
| drobilla | Actually, no, that's the difference between a port that takes a <whatever> straight up and using parameter:Set | 14:39 |
| drobilla | But the thing needs a name and so on to be displayed in UIs | 14:39 |
| drobilla | So perhaps Impulse is-a Command or some such | 14:40 |
| drobilla | Which is also how plugins could build something best presented as a menu and so on | 14:40 |
| drobilla | Blah blah blah | 14:42 |
| drobilla | What we need is an LV2 summit where we can all sit in a room with a whiteboard for a few hours and figure out things like HTF event-based controls should work | 14:43 |
| drobilla | Rather than the 5 years it takes otherwise :) | 14:43 |
| drobilla | One more thought Re: symbols | 14:44 |
| drobilla | Most URIs have one tacked on the end anyway. http://example.org/whatever/suck | 14:44 |
| drobilla | Maybe we can just do what I wish we did for ports and just dictate rules for tranforming the URI | 14:45 |
| drobilla | Uniqueness guarantees are a touch more of a problem than ports if you want to use parameters defined all over the place, but can be resolved easily enough | 14:45 |
| drobilla | This is what we should have done for ports, really. Them not having a URI is a huge PITA | 14:45 |
| drobilla | It's only that way because Turtle gives incentive to *not* give things URIs because you can't abbreviate them like blanks (which is really shitty) | 14:46 |
| falktx | that has already bitten us at mod | 14:47 |
| falktx | we want to add some new small properties to a specific port, but it's not possible without modifying the original ttl file | 14:47 |
| falktx | the modgui can be in a separate file, but extra port stuff can't :( | 14:48 |
| *** edogawa has quit IRC | 14:48 | |
| drobilla | falktx: Yep, it sucks :/ | 14:48 |
| drobilla | (Ports *can* have URIs, of course, but most don't) | 14:49 |
| falktx | with port uris we can get rid of indexes | 14:50 |
| drobilla | falktx: Lilv could probably support that, though, if it's a blocker | 14:50 |
| drobilla | falktx: Not really, at least not without some new mechanism; those are used at run-time | 14:50 |
| falktx | yes, the way I see it connect_port would have an URID instead of index | 14:50 |
| drobilla | (they are what ties the ports in the code to the ports in the ttl together) | 14:51 |
| drobilla | falktx: If we could do an LV3, yes. URIDs being fundamental and in the core would make a lot of things better | 14:51 |
| drobilla | Though, then again, connect_port is stupid and would die anyway | 14:51 |
| drobilla | but some map-to-index bit would take its place | 14:52 |
| *** NickSB2 has quit IRC | 14:52 | |
| drobilla | Maybe we can just enforce parameter URIs are symbol convertible according to some simple rule, and that a plugin's set of parameters must map to a unique set of symbols (unique in the same namespace as ports) | 14:54 |
| drobilla | If lilv checked this, seems fine. It's easy enough to adhere to, tools just need to make it obvious to devs | 14:54 |
| drobilla | Then even our new fancy sample accurate parameters have symbols and all the existing UI/binding/etc stuff that depends on this can still work with them | 14:54 |
| drobilla | mm. das ist gut | 14:57 |
| drobilla | I'll bake it into the spec and add support for jalv's console UI before kicking out an RFC of sorts for that stuff | 14:57 |
| falktx | sadly existing wrappers can't handle this. plugin specs usually don't have this :( | 14:58 |
| falktx | I can make some changes in dpf to allow this. but juce is out of the question | 14:59 |
| drobilla | can't handle this how? | 14:59 |
| falktx | juce doesn't really have an api for this | 14:59 |
| drobilla | They don't have contol ports, either... | 14:59 |
| falktx | yes, but control ports can be worked around | 14:59 |
| drobilla | So can parameters | 15:00 |
| falktx | if they change mid-processing we either split the process or ignore until next run | 15:00 |
| drobilla | Easy enough | 15:02 |
| drobilla | This is actually what I do inside Ingen for plugins with control ports that are connected to sequences of float | 15:02 |
| drobilla | (unless there's a block size restriction) | 15:02 |
| falktx | yeah, I was just saying that for juce plugins it will be no different from what hosts do right now (kinda) | 15:03 |
| falktx | the process is still split for events, rather than passing the events in a fixed audio block | 15:04 |
| falktx | I don't think juce will change this. although they are planning on releasing juce4 next month | 15:04 |
| drobilla | Sure. You don't *have* to, but you still can. | 15:05 |
| falktx | nothing better than pure lv2 plugins | 15:06 |
| drobilla | A clearly biased opinion I nonetheless obviously agree with :) | 15:07 |
| drobilla | As far as just chucking floats around, the sample accuracy thing is nice sometimes, but not a massive win in general | 15:07 |
| drobilla | It's that plus a bunch of other benefits together that make this a thing that really needs to happen | 15:07 |
| drobilla | The transition really sucks, though | 15:10 |
| drobilla | The model of just having one sequence of ordered time-stamped events for your control is so much vastly better | 15:10 |
| drobilla | ... but when you also have these ControlPort things around that everyone is used to at the same time, bleh | 15:11 |
| drobilla | Some veneer on the host side that lets you use the same API to change any control would be nice | 15:13 |
| drobilla | Maybe in the plugin too, though I'm not sure what form that would take | 15:13 |
| * drobilla ends up rolling in to work after noon, as usual | 15:18 | |
| drobilla | I was updating some dates in copyright headers and whatnot earlier, while thinking about how I would really like the time to do some bottom-up overhauls of things | 15:19 |
| drobilla | and realized there is a very distinct range of dates when I throw out all the slv2 crap and made serd, sord, lilv, atoms and related things, and all that | 15:19 |
| drobilla | ... the last time I wasn't in fucking grad school | 15:20 |
| drobilla | surprise! | 15:20 |
| drobilla | ttysl :) | 15:20 |
| *** deva has joined #lv2 | 16:47 | |
| *** falktx has quit IRC | 16:56 | |
| *** sigma6 has quit IRC | 17:20 | |
| *** falktx has joined #lv2 | 17:38 | |
| *** falktx has quit IRC | 18:25 | |
| *** falktx has joined #lv2 | 18:30 | |
| *** deva has quit IRC | 19:47 | |
| *** rncbc has joined #lv2 | 19:50 | |
| *** NickSB2 has joined #lv2 | 19:55 | |
| *** falktx has quit IRC | 21:57 | |
| *** aombk3 has joined #lv2 | 22:12 | |
| *** aombk2 has quit IRC | 22:16 | |
| *** Xzu has joined #lv2 | 23:12 | |
| *** Xzu has quit IRC | 23:15 | |
| *** ssj71 has quit IRC | 23:16 | |
| *** uncle-j_j has quit IRC | 23:22 | |
| *** edogawa has joined #lv2 | 23:43 | |
Generated by irclog2html.py 2.13.0 by Marius Gedminas - find it at mg.pov.lt!
