| | <<<Back 1 day (to 2014/12/21) | 20141222 |
| Robin_Watts | Hi kens. You're here today then? | 11:44.58 |
| kens | Yes, adn tomorrow also | 11:45.12 |
| Robin_Watts | Ah, so I can ignore the support query that just arrived with a clear conscience :) | 11:45.35 |
| kens | Indeed, I'm dealing with it | 11:45.44 |
| Robin_Watts | (as opposed to ignoring it, and feeling vaguely guilty :) ) | 11:45.55 |
| kens | THough I don't plan to make a patched C file for them, either they can learn to do it themselves or wait for Marcos. | 11:46.03 |
| | I'm a bit puzzled why someone who is capable of rebuilding Ghostscript from source would have difficulty integrating sucha simple patch into an earlier version of the code. | 11:52.16 |
| Robin_Watts | kens: The latest comment on bug 695704 set my sarcasm detector off, but I think it was a false-positive :) | 12:46.28 |
| avih_ | tor8: looks like we'll support both mujs and duktape. i managed that with a wrapper which allows duktape to compile the existing code (which uses mujs api), almost 1:1. so you could still have some PR on it ;) but we'll probably make duktape the official support and mujs an "you're on your own" kind of thing. right now they seem to work identically, and also interestingly in my tests - with similar speed. though my test case does excessive use of array | 12:50.44 |
| | manipulations and duktape is more efficient than mujs with non sparse arrays, so maybe that's that. | 12:50.45 |
| Robin_Watts | avih_: Well, that sounds like a great compromise. | 12:51.39 |
| avih_ | Robin_Watts: yeah, i agree. i also didn't want to throw away the code i have and patch it beyond recognition. and the wrapper actually as clean as it could get IMO. | 12:52.29 |
| | that's the initial patch, and there are two more after it, and i still need to solve an ugly issue of the wrapper where i declare all the functions again, but that's not a hard problem to solve. https://github.com/avih/mpv/commit/340ba82b641c5aed9cdc233ad5383db4beb13e08 | 12:53.40 |
| | this revision actually has an off by 1 bug which was fixed in a later patch | 12:54.46 |
| | oh, and 2 more minor bugs. one for duktape and one for mujs. 2 commits later both are fixed. | 12:56.43 |
| Robin_Watts | I can't look at the mo (too busy), and it wouldn't mean much to me if I did. | 12:57.31 |
| | tor8 may or may not be about this week - not sure of his movements around Xmas. | 12:57.59 |
| avih_ | Robin_Watts: would you have an issue with me using the mujs api while compiling with duktape? i don't think you should, though i've now learned to ask in advance ;) | 12:58.12 |
| Robin_Watts | avih_: Would that involve using a mujs header file? | 12:58.37 |
| avih_ | no | 12:58.42 |
| nsz | only comments are copyrightable material in a headerfile.. | 12:59.06 |
| Robin_Watts | Then you're not using any MuJS specific code? | 12:59.07 |
| avih_ | the wrapper provides the definitions | 12:59.08 |
| nsz | and those are easy to remove | 12:59.11 |
| Robin_Watts | I don't think we could possibly object then :) | 12:59.24 |
| | nsz: I don't believe that for a second. | 12:59.41 |
| avih_ | nsz: i don't use mujs.h when compiling with duktape | 13:00.10 |
| Robin_Watts | Inline code and Macro definitions are certainly copyrightable. | 13:00.15 |
| avih_ | you can look at the patch i provided hew few mins ago | 13:00.30 |
| nsz | ok well add '..in reasonable header files' | 13:00.47 |
| Robin_Watts | You may be right in saying that a simple list of prototypes is not copyrightable. | 13:00.51 |
| nsz | you dont put code into header files normally only api | 13:00.58 |
| Robin_Watts | Even then I don't know that I agree with you. | 13:01.27 |
| avih_ | 90% of the mujs api has an identical api under a slightly different name | 13:01.32 |
| nsz | this afaik has been tried several times on us courts with no luck.. so at least there it works | 13:01.45 |
| avih_ | at duktape. so it's just #define js_pushstring to duk_push_string etc | 13:01.55 |
| kens | I'm reasonably sure there was a recent case involving copyright and an API< but I don't recall which way it went. | 13:02.16 |
| Robin_Watts | I believe cases in the US have proved that *PUBLIC* apis are not copyrightable. | 13:02.26 |
| | There was the java header thing. | 13:02.40 |
| avih_ | kens: you serious you don't remember? oracle cs google? :) | 13:02.48 |
| | vs* | 13:02.54 |
| nsz | linux allegedly copied errno numbers from sco and that was not an issue either | 13:03.07 |
| avih_ | iirc the last decision was that oracle had the rights and they got 1$, didn't they? :) | 13:03.16 |
| | or #1m, the same for them | 13:03.28 |
| kens | I have no memory, this is well known to people who know me | 13:03.28 |
| Robin_Watts | http://en.wikipedia.org/wiki/Oracle_v._Google | 13:04.22 |
| | So in the final shakeout, an API *is* copyrightable. | 13:04.33 |
| kens | According to what I just read, the court's current decision is that API's can be copyrighted. | 13:04.44 |
| | But its due for appeal to the Supremes | 13:04.59 |
| avih_ | Robin_Watts: anyway, the mpv docs will say that we support both interchangably but officially support one. but with links etc. that should give mujs useful exposure imo without getting into any licensing issues. | 13:05.31 |
| Robin_Watts | avih_: Sounds good to me. | 13:07.04 |
| avih_ | k. good. | 13:07.41 |
| Robin_Watts | I would imagine that, given the duktape and mujs APIs are so similar, that we'd be hard pressed to make a case that there was significant IPR in the API. | 13:07.45 |
| avih_ | i'd think so too. | 13:08.08 |
| | it's basically almost a 1:1 translation of the api names | 13:08.34 |
| | (not the other way around though since duktape has more features and a _lot_ more api functions) | 13:09.01 |
| kens | Robin_Watts tor8 : have you looked at flights yet ? | 13:50.49 |
| Robin_Watts | kens: Vaguely. | 13:51.06 |
| kens | Looks like BA, US or United for non-stop flights | 13:51.19 |
| kens | has no intentionof flying Iberian | 13:51.33 |
| Robin_Watts | https://www.google.co.uk/flights/#search;f=LHR;t=DEN;d=2015-03-05;r=2015-03-09 | 13:51.45 |
| | BA/Iberia/American are a codeshare I believe. | 13:51.59 |
| kens | THey are, so you need to be careful to see is it s BA aircraft | 13:52.16 |
| Robin_Watts | But the 630 quid one is a direct one. | 13:52.42 |
| kens | Yes and I think its actually a BA flight | 13:53.01 |
| | United is cheaper but seems to have stops | 13:53.17 |
| | DOn't want stops :-( | 13:53.22 |
| Robin_Watts | Helen/I and I think Paul/Linda are staying a while. | 13:53.32 |
| kens | brb door | 13:53.38 |
| | I'm planning to head over to Vail | 13:54.47 |
| | Currently thinking I might spend 2 or 3 days there then back to Denver and home. | 13:55.25 |
| Robin_Watts | kens: We plan to go to Longmont on the saturday/sunday nights, then to Vail on the monday/tuesday. | 13:55.33 |
| | kens: The Vail Cascade seems to be the Hotel that has caught Helens eye. | 13:55.57 |
| kens | Might meet you in Vail then :-) | 13:55.58 |
| | Hmm, I don't remember seeing that one, let me look | 13:56.13 |
| Robin_Watts | Actually in Vail (as opposed to beaver creek etc), not stupidly expensive, and gets good reviews. | 13:56.35 |
| kens | Oh I was only looking at hotels in Vail itself | 13:56.47 |
| | Ideally walking distance of the lift, but that looks expensive on the whole, so maybe I'll take the ski bus | 13:57.07 |
| Robin_Watts | Most hotels in Vail seem VERY expensive. | 13:57.16 |
| kens | Oh boy yes. | 13:57.22 |
| | Cheapest was ~$250 per night at that time | 13:57.34 |
| | OK so that's another one on the ski bus route | 13:58.00 |
| Robin_Watts | There is a lift from the hotel. | 13:58.16 |
| kens | Hmm doesn't say that ont he fromt page, kust says ski shuttle. | 13:58.40 |
| Robin_Watts | https://www.google.co.uk/maps/place/Vail+Cascade+Resort+%26+Spa/@39.6180382,-106.3879095,14z/data=!4m2!3m1!1s0x0:0xfd0d64442e1a1023 | 13:58.48 |
| kens | Hmm, interesting, I presume you can ski down from the head of the lift to the makred trails. | 13:59.29 |
| | Ah so its outside the actual ski resort, I see nopw | 13:59.58 |
| | Aha, its not in the ski resort, that's why | 14:00.14 |
| Robin_Watts | The resort being around "The Lodge At Vail" ? | 14:01.06 |
| kens | I guess, I was looking at the map where it syas 'Vail Ski Resort' | 14:01.24 |
| | I think I need a piste map..... | 14:02.29 |
| Robin_Watts | Vail Ski Resort, on that map, is at the top of a mountain? | 14:02.35 |
| kens | Either that or the bottom, there's no terrain marked | 14:02.51 |
| Robin_Watts | Are the ski runs not the things in dark green ? | 14:02.58 |
| kens | But there are 3 lifts there | 14:03.05 |
| | runs are green, blue or black | 14:03.18 |
| | (well dark gray) | 14:03.24 |
| Robin_Watts | Right, using the satellite imagery, I think there *are* no hotels there. | 14:04.18 |
| | They are all along the main road. | 14:04.23 |
| kens | :-) I guess if its the top of the mountain that makes some sense | 14:04.36 |
| | Let me just find a piste map so I can see where that lift goes | 14:04.50 |
| | OK found it | 14:06.44 |
| Robin_Watts | https://www.google.co.uk/maps/place/Vail+Ski+Resort/@39.4978911,-106.3698265,3902a,20y,80.4t/data=!3m1!1e3!4m2!3m1!1s0x876a7060c630dbdb:0xa7c2cf2d4b45ada2 | 14:06.58 |
| kens | Its pretty low, so I guess its up the lift, down the green to the main lift and go up that | 14:07.05 |
| Robin_Watts | Tilting the map shows terrain. | 14:07.12 |
| | And Vail Ski Resort is indeed at the top. | 14:07.34 |
| kens | Yep, it was clear when I saw the trail map | 14:07.52 |
| | That URL just shows mw blue | 14:08.21 |
| | Oh, I htiunk it got truncated | 14:08.33 |
| | Nope just saws nothing useful for me, oh well. | 14:11.03 |
| Robin_Watts | So, I think these are our flights: https://www.google.co.uk/flights/#search;f=LHR;t=DEN;d=2015-03-05;r=2015-03-11;sel=LHRDEN0BA219,DENLHR0BA218;px=2 | 14:12.14 |
| | But Miles hasn't said to book yet, right? | 14:12.24 |
| kens | Hmm, let me check | 14:12.36 |
| | He said 'let me know your itineraries' so I would say yes | 14:12.57 |
| tor8 | kens: briefly looked at the available options. looks like BA is my best option yet again... | 14:13.53 |
| kens | THere don't seem to be many options otr8 | 14:14.09 |
| | tor8^^ | 14:14.12 |
| | FOr us, anyway | 14:14.18 |
| tor8 | kens: well, there is that or lufthansa :) | 14:14.23 |
| | and I prefer BA, 1 hour shorter second leg | 14:14.34 |
| kens | I think Lufthansa would be a change for us, so I'm thinking BA like Robin | 14:14.47 |
| tor8 | but a 30 minuter longer first leg | 14:14.52 |
| | kens: yeah. might make sense to get on the same flights, if we're staying the same days | 14:15.11 |
| Robin_Watts | tor8: sounds like kens/paul and I are all staying longer. | 14:15.40 |
| tor8 | skiing could be fun, but I'm terribly out of shape and I'd be at huge risk of breaking something :/ | 14:16.52 |
| | not having been on a pair of skis for coming on 10 years, and 10 years before that... | 14:17.10 |
| kens2 | My internet is not happy again today :-( | 14:17.46 |
| Robin_Watts | tor8: Helen and I will not be skiing. | 14:18.10 |
| | Helen because she a) has never tried it, b) wouldn't try if it she could, c) wouldn't be able to even if she tried it. (That sounds harsh, but really, we tried cross country skiing once, and she made it 3 yards before giving up). | 14:19.10 |
| | I dry slope ski'd once, and didn't enjoy it much. | 14:19.54 |
| kens2 | Dry slope ski-ing is horrible | 14:20.04 |
| Robin_Watts | I've done cross country skiing and just thought it was hard work. | 14:20.11 |
| | Am more tempted to try boarding, but won't do that on this trip as it would mean leaving Helen etc. | 14:20.40 |
| kens2 | tor8 I definitely plan to go do some ski-ing, I'd be happy for you to come too, safer to ski in a party | 14:21.20 |
| Robin_Watts | I believe mvrhel was considering skiing too. | 14:21.53 |
| kens2 | I see the return flight is the evening and its 2 hours from Vail to Denver, so technically I could ski half a day. OTOH that would also mean a days kift pass, which is *also* expensive | 14:22.08 |
| tor8 | kens2: is Stella coming as well? | 14:22.44 |
| kens2 | tor8 almost certainly not | 14:22.51 |
| tor8 | kens2: so at the moment it's only you going skiing? | 14:23.07 |
| kens2 | She isn't that great a skier and doesn't fancy an unknown location, at high altitude | 14:23.11 |
| | tor8 so far, just me yes | 14:23.16 |
| tor8 | in that case it'd be bad form not to join you :) | 14:23.19 |
| kens2 | Don't feel pressured, I'm comfortable ski-ing solo | 14:23.34 |
| tor8 | I guess I'll have to work on getting into shape; but I'm not that great a skier anymore so you'll likely take off solo anyway :) | 14:24.10 |
| kens2 | Don't count on it, I'm only intermediate, and I don;'t ski enough to get any better each year | 14:24.31 |
| | Melanie is a better sjier than me, but I'm more confident :-) | 14:24.53 |
| tor8 | I'm comfortable with red pistes, if they're not too icy. I avoid black pistes, they're just not enjoyable since I go down them at a crawl... | 14:25.30 |
| kens2 | It looks like if we stay at the cascades the lift goes up to a green run that takes you down to the lift at lionhead, and that has a long lift up to a mountain | 14:25.50 |
| | tor8 seems to be green/blue/black in the US as opposed to European blue/red/black | 14:26.09 |
| | 3 nights at the cascades (sun,mon, tues nights) is £654 | 14:28.01 |
| | FOr a standard room | 14:28.06 |
| Robin_Watts | Beaver Creek *can* be cheaper I think, but Helen found that the Cascade was as cheap as anywhere you'd want to stay in Beaver Creek. | 14:29.36 |
| | or at least, as cheap as anywhere *she'd* want to stay. | 14:29.53 |
| kens2 | I didn't look anywhere except Vail, the ski-ing didn;t look so good at beaver creek | 14:30.03 |
| | I wonder if I can get up to Vail pn Saturday, then I could Ski Sunday Monday and Tuesday | 14:31.07 |
| | Will have to find a ski hire place too of course. | 14:31.35 |
| tor8 | kens2: I would hope that the resort/hotel place has ski hire places | 14:32.01 |
| kens2 | I'm sure th resort does, byt the hotel is one end of the resort, so might need to catch the shuttle to a hire shop | 14:32.28 |
| | I guess I need to talk to Michael, and unfortunately I have to be off in good time tonight, got guests for dinner. I'll try and catch him later. | 14:39.48 |
| | Hmm going from 3 nights to 4 at that hotel drops the standard rooms and requires a condo, so the price shoots right up | 14:40.14 |
| | I guess it'll be check in on the SUnday then | 14:40.24 |
| | OK so tentatively, I'm planning to arrive in Denver on Thursday, meetings Friday and Saturday, head for Vail Sunday morning, stay there until the Wednesday and fly home Wednesday evening. Return shuttle from the airport to Vail is ~$188, 3 nights in the cascades is $1204, lift pass is $129 per day. | 14:48.34 |
| henrys | vail is pretty expensive I thought - haven't looked recently - keystone, copper and steamboat are all popular amongst the locals here. | 15:18.03 |
| kens2 | Vail does seem tobe expensive, yes | 15:18.19 |
| tor8 | avih_: js_setlength is actually js_pushnumber+js_setproperty("length") ... which is used if for example truncating an array | 15:20.45 |
| | the length of an Array object is also automatically updated | 15:21.11 |
| Robin_Watts | kens2: "Vail Cascade Resort & Spa provides a variety of Vail Mountain ski rentals and equipment for guests, from snowshoes and skis to goggles and gloves." | 15:21.12 |
| kens2 | Missed that, thanks RObin | 15:21.23 |
| Robin_Watts | Prof Google knows everything. | 15:21.39 |
| kens2 | will need skis, poles and a hard hat (to keep my ears warm ;-) | 15:22.03 |
| tor8 | Robin_Watts: Google is our networked brain memory extension ;) | 15:22.06 |
| henrys | these rankings match what I've heard fairly well: http://www.powderhounds.com/USA/Ski-Colorado/Best-Ski-Resorts.aspx | 15:23.13 |
| kens2 | Hmm, a major criterion there seems to be 'off piste' | 15:24.01 |
| | Maybe snowmass or copper then | 15:24.50 |
| | Dumb question, how far is it from Denver to Aspen ? | 15:28.44 |
| henrys | I've been to keystone, vail, beaver and a-basin and it all seemed the same to me. poor man's skiing is eldora. rent a dump in boulder to stay at and take the bus to the mountain ... | 15:28.53 |
| kens2 | Hmm so copper is nearer denver. | 15:30.08 |
| henrys | well over 3 hours closer to 4 | 15:30.15 |
| kens2 | Vail looks prettier :-) | 15:31.13 |
| henrys | it certainly is. | 15:31.28 |
| kens2 | I have to push off early today, so I'm heading out now. I'll try and come back later to catch up with Michael. Goodnight all | 16:26.37 |
| avih_ | tor8: yeah, i know that's how you implement it, and it would most probably work in duktape too. anyway, i didn't really have a need for it. | 19:43.04 |
| avih | nevertheless, it should probably either implemented correctly or we should drop the code which uses it. | 19:44.00 |
| tor8 | avih: I mostly reflected over the comment which isn't exactly true | 19:44.44 |
| | or at least rather misleading | 19:44.50 |
| avih | tor8: which comment isn't true? | 19:45.05 |
| | also, the comments at the last patch on that branch are etter, though stillcan be corrected a bit | 19:45.27 |
| | for instance for userdata i ignore both tag and prototype, but only mentioned tag there' | 19:45.51 |
| tor8 | "duktape doesn't really has an equivalent api. It sets the length according to the highest index pushed." near js_setlength | 19:45.57 |
| avih | oh | 19:46.15 |
| tor8 | with the 'empty' js_setlength might be misleading in terms of what js_setlength is supposed to do | 19:46.23 |
| avih | i meant as an equivalently functioning call ;) | 19:46.32 |
| | not two! ;) | 19:46.48 |
| tor8 | does duk_push_pointer take the prototype object from the stack? | 19:47.41 |
| avih | tii was a bit surprised to see that on my system it performs very close to mujs, especially with the overhead of the shift 'this' on every call | 19:47.42 |
| tor8 | I reckon the vast majority of time is going to be spent in property lookups | 19:48.02 |
| avih | it diesn't. that but is fixed at one of the two followup commits | 19:48.05 |
| | same as mud_push_next_key for the mujs case | 19:48.17 |
| tor8 | a few memory shifts of the stack which sits in hot cache is going to be irrelevant | 19:48.20 |
| | right | 19:48.28 |
| | I've done some more benchmarks, and the performance difference between mujs and duktype is all over the map | 19:49.08 |
| avih | yeah, and also, for non sparse arrays with numeric indexes it uses a c array as far as i understand and doesn't convert the intex to a string | 19:49.12 |
| | sort of a fast path | 19:49.18 |
| tor8 | some things, mujs is magnitudes faster, some things, slower | 19:49.22 |
| | yeah, mujs doesn't do that fast path | 19:49.34 |
| avih | interesting | 19:49.35 |
| tor8 | overall I reckon the performance is going to be close enough not to worry about it | 19:49.58 |
| avih | (wtf is wrong with my typing today??!?) | 19:49.59 |
| | paybe, but even if one of them is x2-x3 slower than the other, that's still plenty for scripting of a video player. it mostly has to do with wait for key presses and do something, or handle other very simple stuff | 19:50.56 |
| tor8 | yeah, I don't think any performance diff < 100x is going to be relevant for the sorts of things you intend to use it for | 19:51.31 |
| avih | yeah | 19:51.45 |
| tor8 | which is why performance has never been a major concern for mujs | 19:52.15 |
| avih | btw, another project which i found very interesting is CASTL https://github.com/PaulBernier/castl . it's a javascript engine implemented on top of lua, possibly luajit | 19:52.34 |
| | the code looks solid, as well as the approach. it has some parts in js to bootstrap it, but it's self hosted since this stuff is also at the sources already compiled to lua to support eval and Function | 19:53.25 |
| tor8 | interesting, but very heavy weight dependencies... spidermonkey and pcre | 19:53.46 |
| avih | not at all | 19:53.55 |
| | it's fully self hosted, i.e. you supposedly don't really need any of it to actually run it, since everything is lua eventually | 19:54.29 |
| | for development though, maybe. but i didn't get that far, just read his blurbs and browsed the code a bit | 19:54.57 |
| tor8 | for bootstrapping then I guess | 19:55.35 |
| avih | and since mpv supports lua and luajit, i thought it could actually have some very interesting js performance. but i don't think it's mature enough for us yet, i don't think it has lua binding yet, and it's LGPL3 which we probably prefer to not use | 19:56.19 |
| | still, very interesting :) | 19:56.54 |
| tor8 | brb, need to relog | 19:59.23 |
| avih | i really wish there was some reference js implementation out there | 19:59.28 |
| | i'd say mujs could come close if it straighten out its kinks | 19:59.49 |
| tor8 | spidermonkey is the closest there is, but that's still full of extensions and nonstandard behaviour | 19:59.50 |
| avih | yeah. and huge :) | 20:00.03 |
| | OTOH, if it was distributed with distros as a common enough component, then hugeness wouldn't have been an issue, but afaik it isn't | 20:01.22 |
| tor8 | avih: we'd need to add "use strict" to be really ES5 compliant | 20:03.23 |
| avih | true, though you said it shouldn't be too hard. | 20:03.40 |
| tor8 | but it's such a hack, I'm thinking a flag to js_newstate whether to be "use strict" globally or not | 20:03.47 |
| | the interactions between strict and non-strict code are a nightmare | 20:03.59 |
| avih | and probably also accept null inside strings, and manage gc for everything if that's still not the case | 20:04.11 |
| | oh | 20:04.28 |
| tor8 | zeros in strings are a major issue right now, but should be a fairly minor fix but would need to add another C function in the api | 20:04.45 |
| avih | such as? | 20:04.57 |
| tor8 | const char *js_tolstring(js_State *J, int *lenp) | 20:04.58 |
| avih | right | 20:05.02 |
| | tbh though, the duktape api is not fun to use. mujs is minimalistic and works. /me like | 20:05.37 |
| | which is also why i prefered to take the wrapper approach. initially i wanted to use a wrapper to get things working and then actually apply the wrapper. but it ended up decent enough to keep the mujs style api use, and just make it run in duktape too | 20:06.43 |
| | and i also gained to keep compatibility with both, which is nice | 20:07.00 |
| tor8 | the benefits of both having been inspired by the lua stack based api :) | 20:14.17 |
| | how does duktape handle 'this', both for setting when calling a function and then using? | 20:14.35 |
| avih | duk_push_this | 20:17.09 |
| | inside a function | 20:17.16 |
| | and to can a functoun bound to an object you use duk_call_method | 20:17.35 |
| | call* | 20:17.43 |
| | and there's another one where you can call a property of an onject by name | 20:18.05 |
| | i'm guesing he tried to optimize each case, but tbh it's too many function to use regularly | 20:18.48 |
| tor8 | right. I preferred a more direct approach. the only gripe I have is with how it can be tricky to juggle the prototype objects to create userdata | 20:19.51 |
| | but I can't really see a better approach | 20:20.04 |
| avih | what would that be used for? | 20:20.08 |
| | it also has buffer userdata, both fixed and dynamic, though i inly took what i needed to make my existing code work. | 20:20.38 |
| tor8 | buffer userdata, as typed array style stuff? | 20:21.19 |
| avih | no, c biffer | 20:21.40 |
| | buffer | 20:21.43 |
| tor8 | use the prototype object to create a userdata object with methods | 20:21.51 |
| avih | oh | 20:21.59 |
| tor8 | mujs.com/docs/examples, the "Complete userdata example" | 20:22.25 |
| | f = new File("input.txt"); f.readLine() | 20:22.48 |
| avih | right, yeah, that could be useful, though i didn't need it so i didn't learn its capabilities. i only use userdata in two places (one if i use setcontext), and in both i only needed a pointer | 20:22.52 |
| | i see. interesting | 20:23.37 |
| | the mpv avi internally is rather flat. it's only arranged into object for namespacing of sorts | 20:24.21 |
| | api* | 20:24.25 |
| | and only for the scripts. internally it's all mpv_do_this and mpv_do_that | 20:26.15 |
| tor8 | just a lot of global state then, not a tangle of objects? | 20:28.41 |
| avih | the api which the lua binding expose is this http://mpv.io/manual/master/#lua-scripting and the js one implements it to the letter except for very internal stuff | 20:28.54 |
| | the state is of the script and of the client instance (which maps to the player itself) | 20:29.35 |
| | each script is completely isolated from other scripts except through mpv apis (like one script can change the volume and another script can observe the volume for changes and do something when it changes) | 20:30.24 |
| | and each script runs in its own thread and its own lua/js instance | 20:31.16 |
| tor8 | yeah, so from each script's pov it's global state | 20:31.32 |
| | which makes for a simple scripting API | 20:31.41 |
| avih | yeah, for the most part. the biggest functions are to convert between the mpv json-like c structure and js/lua native objects | 20:33.00 |
| | and there's also a rather chunky piece of code to implement all the state at the client script (this is done in js/lua default file which is loaded before each script). like registrations of events or observers, timers, the key binding stuff needs quite some code to support, etc | 20:34.34 |
| | all in all, it provides very thorough control of the player via scripting | 20:35.17 |
| | if you download and try it, you'll see that it has a sort of GUI which is called OSC (on screen control). so this is implemented in lua which uses libass to draw the stuff | 20:36.15 |
| | Forward 1 day (to 2014/12/23)>>> | |
IRC Logs