| 01:47.37 | Repo | imhobags: Amaedeus master * 0.16beta2-6-g66ab50a (2 files in 2 directories): [+1 commit] The search box in the title bar will now search in the same way that the "search database" window has always worked: case-insensitive by item name. |
| 02:43.56 | seebs | Wow, Shinies just gave me three blues in a row, two of them from the Dead Simon set. |
| 03:32.14 | Noshei | ZorbaBeta: what kind of math is behind the curves on canvas? |
| 03:33.03 | Noshei | I want to make as close to a perfect circle as possible, but I'm not entirely sure how the control point is being factored in. |
| 03:36.50 | ZorbaBeta | noshei, http://help.adobe.com/en_US/AS2LCR/Flash_10.0/help.html?content=00001288.html |
| 03:37.14 | Noshei | can we do cubic bezier curves? |
| 03:37.22 | Noshei | or only quadratic? |
| 03:37.30 | ZorbaBeta | I don't believe flash supports that - I think you're limited to quadratic |
| 03:45.32 | Ivnedar | flash 11 for cubic |
| 03:50.04 | Ivnedar | hrrm, there are some routines for approximating a cubic curve with 2+ quadratic curves |
| 03:52.04 | Ivnedar | hrrm, could probably rig up a helper function for it |
| 03:52.15 | Ivnedar | fixed midpoint approximation looks close enough |
| 03:53.11 | Noshei | I'm just trying to figure out hot to get it setup for Proportional as well |
| 03:54.27 | Noshei | and fit it into rifts rather strange coordinate system |
| 03:57.55 | Ivnedar | proportional you work off a square with 0,0 at the top left and 1,1 at the bottom right |
| 03:58.06 | Ivnedar | any rotation is done around 0,0 so you need to translate first |
| 03:59.13 | Noshei | yeah I know how the coords are setup, just makes everything a bit harder, especially since I'm not familar with the math involved |
| 04:07.08 | seebs | One of the cursor-decorating addons I used in WoW would crash the Mac client for a while because it had a 512x512 circle texture. |
| 04:12.14 | seebs | You know, I really have to say. Purely in terms of Gameplay Options. Rift does not suck at all these days. |
| 04:12.28 | seebs | Like, at any given time I can just sort of log in and there's usually something fun to do. |
| 04:19.32 | seebs | I so want, |
| 04:19.42 | seebs | a way to mark a character as "no, really, I'm not spamming myself." |
| 04:20.00 | seebs | I have two accounts. Even a mere "can send to one characters" wouldn't quite solve it for me. |
| 04:20.28 | Ivnedar | or a "if I have them on my friends list, they can send as many as they want to me" thing |
| 04:23.43 | Ivnedar | crazy, people saying they'd buy a black riding squirrel for 20k credits |
| 04:23.54 | Ivnedar | they've already spent more than that with no squirrel |
| 04:24.01 | Ivnedar | that's >$100 |
| 04:30.54 | seebs | Yeah. |
| 04:35.36 | Ivnedar | facepalms, guy in our guild blew through 45k credits on vaults ... |
| 04:36.47 | Ivnedar | apparently likes the feeling of opening them, doesn't want the rewards ... |
| 04:55.21 | Noshei | lol |
| 04:55.56 | Noshei | I'm wondering how many colors I should set as a maximum for this addon |
| 04:56.22 | Noshei | I'm planning on it cycling through colors, I'm thinking 5 or 6 |
| 05:58.41 | seebs | 6! So we can have rainbows. |
| 06:53.42 | *** join/#riftuidev Imhothar (~Imhothar@ppp-93-104-191-151.dynamic.mnet-online.de) |
| 07:35.46 | Noshei | ok wow |
| 07:36.10 | Noshei | so ZorbaBeta I think I just decided to put my current addon on hold due to performance issues |
| 07:37.11 | Noshei | so all I have right now is a single canvas with no fill that creates a circle around the mouse and will follow the mouse when it is moved |
| 07:37.40 | Noshei | when not moving the mouse I'm getting 60-67 fps, when moving the mouse I'm getting 25-35 fps |
| 07:38.44 | Noshei | with my addon turned off I see about a 10 fps drop when moving the mouse, compaired to a 30 fps drop. |
| 07:43.21 | Noshei | well this sucks |
| 07:44.01 | Noshei | not much use in my finishing this addon at this point, a 20 fps difference is pretty significant |
| 07:52.07 | Ivnedar | Hrrm, apparently running rift from steam ignores the documents path in rift.cfg ... |
| 07:52.49 | Ivnedar | even though the rift.cfg is still a shortcut to roaming data, and that file points document folder at my documents |
| 07:53.06 | Ivnedar | it still uses the interface directory in the game area ... |
| 10:47.03 | Repo | completionist: xenyre * r7 (3 files in 1 directory): v1.2 |
| 10:47.03 | Repo | - Placed quests with an empty zone in the None zone |
| 10:59.47 | Ivnedar | lol, found an amusing way to insanely pad dps on eggtenders |
| 11:00.04 | Ivnedar | picking up harpoon resets cd on harpoon |
| 11:09.19 | seebs | That's a pretty huge drop from mouse movement regardless. |
| 11:12.53 | Ivnedar | how often was it updating? every frame? |
| 11:13.35 | seebs | I'd be unsurprised if merely *having* a thing on the cursor were expensive because of the movement. Hmm. |
| 11:13.54 | Ivnedar | having a frame move 10 times a second shouldn't cause issues |
| 11:14.17 | seebs | Hmm. |
| 11:14.21 | seebs | I wonder. |
| 11:14.44 | Ivnedar | I'm going to play with it once raid ends |
| 11:14.56 | Ivnedar | Topped eggtenders dps meters by a huge margin tonight |
| 11:15.01 | Ivnedar | so much fun :D |
| 11:15.26 | Ivnedar | running 61 tempest, so whenever wavelength was up, I was on boss/eggs |
| 11:15.54 | Ivnedar | whenever it was on cooldown and hatchlings were up, I was standing next to a harpoon rack click spamming it |
| 11:17.55 | Ivnedar | (harpoon does 90% of hatchling's hp in damage as an instant ability with massive range |
| 11:17.58 | seebs | Hmm. You can't :SetPoint something to the mouse cursor, it seems. |
| 11:18.10 | seebs | So you have to get the mouse's location and :SetPoint your frame from that. |
| 11:18.11 | Ivnedar | has a longish cooldown, but picking up the item again resets the cooldown |
| 11:18.13 | Ivnedar | so spam spam spam) |
| 11:18.20 | Ivnedar | yup |
| 11:18.55 | Ivnedar | hook into mouse move event, on update, check if last update was >0.1s ago, if so, move frame to cursor position |
| 11:19.23 | Ivnedar | if noshei was calling setshape every frame, I can see how that would be a fps loss |
| 11:19.35 | seebs | Well, if you check for that, you're going to have a very visible lag, I think. Hmm. |
| 11:19.53 | Ivnedar | 0.05s then |
| 11:19.56 | Noshei | Ivnedar: only setPoint on mouse move event |
| 11:20.19 | Ivnedar | are you guaranteed to only get one move event per frame? |
| 11:20.22 | Ivnedar | hrrm |
| 11:20.24 | seebs | What frame are you checking for mouse move events? |
| 11:20.38 | Noshei | not a frame event |
| 11:20.42 | seebs | Yeah, I was thinking something like mousemove(event, x, y) { updated = true; mx = x; my = y } |
| 11:21.13 | seebs | Ahh, Event.UI.Input.Mouse.Cursor.Move? |
| 11:21.23 | Noshei | aqEvent.Mouse.Move |
| 11:21.57 | Ivnedar | how bad does it look if you only move it every 0.05s ? |
| 11:23.01 | Noshei | haven't tried that Ivnedar |
| 11:23.23 | Noshei | and seebs there is a global event, it isn't frame based at all |
| 11:24.24 | seebs | There is? |
| 11:24.35 | Ivnedar | yeah |
| 11:24.48 | seebs | Hmm. All I can find is Event.UI.Input... which is frame-specific. |
| 11:24.52 | Ivnedar | gives the new x,y coords of the mouse cursor |
| 11:24.59 | seebs | Event.Cursor, of course. |
| 11:25.04 | Ivnedar | Event.Mouse.Move(x,y) |
| 11:25.06 | seebs | Wait, no. That's different. |
| 11:25.29 | Ivnedar | cursor shows the cursor picking stuff up |
| 11:26.14 | seebs | Hmm. Seems to me there is never more than one Event.Mouse.Move per frame, so far. |
| 11:26.26 | Ivnedar | goodo |
| 11:26.38 | Noshei | yeah I think it is frame limited |
| 11:27.38 | seebs | Okay. So, on my machine, I get 50fps in a dimension with nothing happening. |
| 11:28.06 | seebs | And no change if I add a plain Frame to the cursor. |
| 11:28.23 | seebs | Although. It is very obviously lagging by a couple of frames (!). |
| 11:28.37 | Noshei | path = {{xProportional = 0.5, yProportional = 0}, |
| 11:28.37 | Noshei | {xProportional = 1, yProportional = 0.5, xControlProportional = (61/64), yControlProportional = (3/64)}, |
| 11:28.37 | Noshei | {xProportional = 0.5, yProportional = 1, xControlProportional = (61/64), yControlProportional = (61/64)}, |
| 11:28.37 | Noshei | {xProportional = 0, yProportional = 0.5, xControlProportional = (3/64), yControlProportional = (61/64)}, |
| 11:28.38 | Noshei | {xProportional = 0.5, yProportional = 0, xControlProportional = (3/64), yControlProportional = (3/64)}} |
| 11:28.40 | Noshei | CUCI.Cirlce:SetShape(path, nil, stroke) |
| 11:28.44 | Noshei | thats the path I'm using |
| 11:28.50 | seebs | Okay, this is really odd. |
| 11:29.07 | seebs | I have a SetPoint("TOPLEFT", UIParent, "TOPLEFT", mx, my) |
| 11:29.16 | seebs | where mx/my are the last mouse move event coordinates seen. |
| 11:29.26 | seebs | And I run this at Update.Begin. |
| 11:29.49 | seebs | And there is a very consistent frame or so of lag. |
| 11:30.04 | Ivnedar | how's it look if you hook into event.mouse.move instead ? |
| 11:30.05 | Noshei | yeah |
| 11:30.20 | seebs | Hmm. |
| 11:30.42 | seebs | Well, that's an interesting question. I'd been logging the values from Event.Mouse.Move, then SetPoint in Update.Begin. |
| 11:30.55 | Ivnedar | why log them? just use them to set position :D |
| 11:31.22 | Ivnedar | then you don't have to worry about the order the events fire in |
| 11:31.46 | seebs | Because I originally thought there might be more. Anyway, no change. |
| 11:32.04 | seebs | And it really feels like it's *more than one* frame of lag. !. |
| 11:32.26 | Imhothar | what order? update.beign comes before all other events in the frame? |
| 11:32.42 | seebs | No. System.Update.Begin is stated to be your last change to make UI changes for this frame. |
| 11:32.45 | seebs | ... I think. |
| 11:33.05 | seebs | "Signals the beginning of a frame render. This is your last chance to make UI changes for this frame." |
| 11:33.22 | Imhothar | ah well ok, then the last before update.end |
| 11:33.22 | Imhothar | anyway, I see no ordering problem |
| 11:33.34 | seebs | No ordering problem, just a couple-frame lag. |
| 11:36.24 | Ivnedar | fair enough, I'd never seen any guarantee anywhere, so had to assume it's possible the order could be different |
| 11:37.09 | Imhothar | One could probably just call Inspect.Mouse in Update.Begin and ignore the move event, since it's your last chance to do any visual changes anyway |
| 11:37.27 | Ivnedar | then you need to store the result and compare every frame |
| 11:37.36 | Ivnedar | even if the mouse is untouched |
| 11:38.01 | Imhothar | so what, at least you get a constant frame drop instead of spikes ;) |
| 11:38.10 | Ivnedar | lol |
| 11:38.13 | Noshei | I just realised I am also calling Inspect.Mouse just before I do the setpoint, so that may be causing some of the lag, have to test without that |
| 11:38.19 | Imhothar | and comparing 2 numbers for equality should really not be of any impact |
| 11:38.54 | Ivnedar | there'd be the inspect overhead too |
| 11:39.25 | Imhothar | "overhead" |
| 11:39.32 | seebs | Hmm. I get a usage error from :SetShape, with that path, and a stroke of { r = 1, g = 0, b = 0 }. |
| 11:39.39 | seebs | I wonder what I am doing wrong. |
| 11:40.01 | Ivnedar | alpha required? |
| 11:40.15 | seebs | Doesn't seem to be. |
| 11:40.30 | Noshei | http://pastebin.com/V9cBLLyg |
| 11:45.02 | seebs | Huh. I am typing manually, but just in case, I dropped the extra control things, and with just those five xProportional/yProportional values, it still errors. And I have no real clue why. |
| 11:45.12 | Ivnedar | what's the error? |
| 11:45.14 | seebs | They *look* lik ethey are spelled correctly. |
| 11:45.24 | seebs | It's the generic Usage Error, so I have no data other than the complete usage docs. |
| 11:45.40 | Ivnedar | but usage error tells you the parameters needed and what you gave it doesn't it? |
| 11:45.43 | seebs | Well, it does confirm that my parameter types are userdata, table, nil, table. |
| 11:45.51 | Ivnedar | hrrm, ok |
| 11:45.56 | seebs | Which seems right. Foo:SetShape() |
| 11:46.09 | Ivnedar | you're missing thickness |
| 11:46.09 | Noshei | thickness: Line thickness. Required. |
| 11:46.31 | seebs | agh |
| 11:46.38 | seebs | That's not in the in-game docs for stroke, I think. |
| 11:46.48 | Ivnedar | down the bottom of stroke |
| 11:46.49 | Noshei | it is |
| 11:46.49 | seebs | Or rather. Not in the error message's usage table. |
| 11:46.58 | Noshei | what i just posted is what is in the docs |
| 11:47.03 | Ivnedar | 3 required at the top, then some optionals, then another required |
| 11:47.05 | seebs | Yup. |
| 11:47.17 | seebs | But the error message's usage list doesn't mention any of the miter things, or thickness. |
| 11:47.37 | Ivnedar | does it go off the bottom of the frame or something? |
| 11:47.47 | seebs | No, this is in the console window. |
| 11:48.03 | seebs | ZorbaBeta: The usage error for SetShape appears to be truncated right after "Endcap behavior"? |
| 11:48.09 | seebs | I wonder whether it's a length limit. |
| 11:49.06 | Ivnedar | <PROTECTED> |
| 11:49.09 | Ivnedar | that shows it at least |
| 11:49.43 | seebs | Hmm. Try a canvas:SetShape() call with wrong args. |
| 11:50.00 | seebs | oh! yes, it's length limit. |
| 11:50.19 | seebs | Because if I change the inputs such that the length of the initial diagnostic changes, the documentaiton truncates differently. |
| 11:51.34 | seebs | Huh. I get no frame rate drop at all from just :SetPoint() on the path every iteration. |
| 11:52.56 | Amadeus- | omg ..I just drove myself insane figuring out how to take a table of strings...and, being as efficient as possible, place certain (known) strings in a particular order at the beginning ...and then sorting the remaining and putting them at the end of the table. |
| 11:53.12 | Amadeus- | so easy in so many languages ...but a pain in the ass with LUA |
| 11:53.34 | seebs | Hmmm. |
| 11:54.15 | seebs | Okay, a data point, I updated it so my mouse move event handler does a SetShape(path, nil, { r = 1, g = 0, b = math.random(), thickness = math.random(5) }) |
| 11:54.18 | Amadeus- | oh..and it's important that the final list not have any nil values in it :) |
| 11:54.23 | Ivnedar | for each string in the first list, if it's in the table, insert into new table, then sort the remaining list? |
| 11:54.25 | seebs | And I still get no visible slowdown from moving the mouse. |
| 11:55.05 | seebs | Noshei, I'm thinking there may be something else slowing it down, because just drawing a canvas somewhere around the cursor seems not to be hurting performance any. |
| 11:55.54 | Imhothar | First things first, it's Lua not LUA. And second, yeah |
| 11:56.42 | seebs | Hmm. |
| 11:58.26 | Noshei | seebs: what are you using to check framerate? |
| 11:58.31 | seebs | Gadgets. |
| 11:58.52 | Noshei | i'll pull up the gadgets one, using fraps right now |
| 11:58.59 | seebs | oh! |
| 11:59.12 | seebs | I betcha fraps has to pay some horrific cost for cursor stuff. |
| 11:59.36 | Noshei | na, gadgets is showing the same thing |
| 12:01.43 | seebs | Huh, that's very odd. |
| 12:02.19 | seebs | I've got one taking that basic path, and pulsing it in size and color every frame, with larger sizes when dx and dy are larger, and it's not affecting frame rate at all. |
| 12:02.34 | seebs | I wonder whether it matters whether anything else is going on. |
| 12:02.36 | Noshei | and even more interesting is that fps continues to go down the longer I am constantly moving the mouse |
| 12:03.01 | seebs | oh! you have ghosts |
| 12:03.06 | Ivnedar | seebs: pastebin yours and get noshei to try it? see if there's soemthing different you're doing? |
| 12:03.50 | Noshei | if you mean like the mouse trail thing, I certainly have that going on |
| 12:04.06 | Repo | imhobags: Amaedeus master * 0.16beta2-7-g75f377c (2 files in 2 directories): [+1 commit] Item enchantments such as runes, whetstones, and insoles now have their own category. |
| 12:04.49 | seebs | http://pastebin.com/NHcAJCHD |
| 12:05.10 | seebs | The reason it's such strange code is that it's being maintained as a snippet in RiftRC. |
| 12:05.49 | seebs | ooh. |
| 12:05.54 | seebs | I DID manage to get two mouse updates! |
| 12:06.16 | seebs | Only... like three times in five minutes. |
| 12:06.37 | Repo | imhobags: Amaedeus master * 0.16beta2-8-ga0c9178 RiftAddon.toc: [+1 commit] Update TOC file for new tag (0.17beta) |
| 12:07.59 | Noshei | thats pretty much what I'm doing |
| 12:08.16 | Repo | imhobags: Amaedeus 0.17beta * feb780b /: [new tag] Tagging as 0.17beta |
| 12:08.25 | Noshei | in face you are doing more stuff within the update event than I am |
| 12:09.11 | Amadeus- | god this damn system drives me nuts |
| 12:09.48 | *** join/#riftuidev Imhothar (~Imhothar@ppp-93-104-191-151.dynamic.mnet-online.de) |
| 12:10.17 | Ivnedar | ooh, 2 updates! |
| 12:10.35 | Repo | imhobags: Amaedeus 0.17beta * /: [TAG DELETED] |
| 12:10.37 | Ivnedar | it does look cool :) |
| 12:11.11 | Imhothar | Amadeus- deleting a tag won't prevent Curse from packaging |
| 12:11.18 | Repo | imhobags: Amaedeus 0.17beta * 80469a4 /: [new tag] Tagging as 0.17beta |
| 12:11.21 | Amadeus- | it wasn't packaging it though :( |
| 12:11.28 | Amadeus- | so, was going to try again |
| 12:11.34 | Amadeus- | there |
| 12:11.59 | Amadeus- | learned lesson: You have to tag/push at once, and with no other commits ..for it to work properly ...at least that's what I find |
| 12:12.15 | Ivnedar | interesting that if you click and drag, the cursor vanishes, the events still fire, and then the cursor reverts back to where it was when you finish dragging |
| 12:12.16 | Amadeus- | I tried commiting ...then tagging (but not pushing) ....THEN push ...and the packager just ignored the tag |
| 12:12.24 | Imhothar | First you push your entire code, wiat for the alpha build, and then push the tag separately for the beta |
| 12:13.17 | Amadeus- | yea...know that now ..haha |
| 12:14.23 | Noshei | yeah seebs I just tested yours and it doesn't have the same fps issue |
| 12:14.34 | Noshei | so something else in my code is the issue |
| 12:15.00 | seebs | if mousemoved then local _g = deep_copy(_G) end |
| 12:16.11 | Noshei | lol |
| 12:16.26 | Ivnedar | you missed the loop around that |
| 12:17.00 | seebs | BTW, I am seriously impressed with the quality of that circle. |
| 12:17.27 | Ivnedar | this is amusing |
| 12:17.39 | Ivnedar | switched to deepwood shard and now I'm regularly getting 2 updates |
| 12:17.46 | Noshei | ah thanks, it gets a bit iffy around 100-150px |
| 12:17.53 | Ivnedar | back on laethys I barely ever get it |
| 12:19.17 | seebs | hmm. |
| 12:19.33 | seebs | I wonder if there is a way to detect that the cursor is being hidden. |
| 12:20.07 | Ivnedar | maybe if you compare event.mouse.move with UIParent's mouse move event |
| 12:20.15 | Ivnedar | though it may not see if it it's over another frame |
| 12:21.10 | Noshei | I have a lot of other shit going on in my addon, but most of it should be an issue at all |
| 12:21.31 | seebs | There is no cursor event for that, anyway. There's a type=nil if you drop an item, but camera movements don't show up. |
| 12:21.43 | Noshei | http://pastebin.com/wCS7DsxT |
| 12:22.11 | Noshei | most of that is ui for the options |
| 12:22.21 | seebs | Cirlce? |
| 12:22.52 | Noshei | lol |
| 12:23.05 | Imhothar | Oh man, how many times would I mistype that |
| 12:23.27 | seebs | The first thing I'd suggest would be just if false then ... end around the MouseMove stuff other than the setpoint for the cursor. |
| 12:23.33 | seebs | Then you can re-enable them and see what's mattering. |
| 12:23.51 | Imhothar | Can't count how many times I mistyped category in IB |
| 12:23.54 | seebs | I do note: You have multiple nested tests there which should nearly always be true. |
| 12:24.06 | seebs | <PROTECTED> |
| 12:24.07 | seebs | <PROTECTED> |
| 12:24.09 | seebs | <PROTECTED> |
| 12:24.11 | seebs | <PROTECTED> |
| 12:24.17 | seebs | I am pretty sure at least three of these will stay true forever once they become true. |
| 12:24.44 | Noshei | yeah I could clean that up a bit, is mostly copy/pasted from another project |
| 12:24.48 | seebs | Also, I think you are not nearly aggressive enough about setting .dragging = false |
| 12:25.03 | seebs | It at least used to be fairly easy to miss a LeftUp event. |
| 12:25.04 | Noshei | I commented all of that part out thoug and still had the fps drop |
| 12:25.36 | Noshei | haven't had an issue with leftup and leftupoutside triggering for me |
| 12:25.46 | seebs | Hmm. |
| 12:25.49 | seebs | That is so odd. |
| 12:26.15 | seebs | Okay. What if you set the mousemove handler to start with |
| 12:26.18 | seebs | if true then return end |
| 12:26.19 | seebs | ? |
| 12:26.31 | seebs | If that's not faster, then it's not related to the shape or anything. |
| 12:26.44 | Noshei | not sure what you mean |
| 12:27.00 | seebs | function CUCI.Function.Event.MouseMove(handle, xcoord, ycoord) |
| 12:27.01 | seebs | <PROTECTED> |
| 12:27.03 | seebs | <PROTECTED> |
| 12:27.05 | seebs | <PROTECTED> |
| 12:27.07 | seebs | Right before that if: |
| 12:27.10 | seebs | if true then return end |
| 12:27.15 | seebs | Obviously, it won't *work* anymore. |
| 12:27.19 | seebs | But will it be faster? |
| 12:28.02 | Noshei | ah as in no fps drop |
| 12:28.13 | seebs | Right. |
| 12:28.20 | seebs | And/or comment out the Event.Attach for it. |
| 12:28.47 | seebs | Or, heck, try it with this addon not loaded. |
| 12:29.00 | Noshei | same fps drop I get with the addon disabled |
| 12:29.13 | seebs | Hmm. |
| 12:29.17 | seebs | So that SetPoint is being super expensive. |
| 12:29.20 | seebs | Okay, next experiment. |
| 12:30.10 | seebs | Command.Event.Attach(Event.System.Update.Begin, function() CUCI.Cirlce:SetPoint("CENTER", CUCI.ContextCircle, "TOPLEFT", 23, 23) end, "horrible update handler") |
| 12:30.18 | Ivnedar | lol |
| 12:30.21 | seebs | Use that instead of the mousemove handler. |
| 12:30.29 | seebs | If you get horrific framerates, it's something to do with SetPoint. |
| 12:30.31 | Noshei | lol, except I didn't have the addon loaded |
| 12:30.31 | seebs | Hmm. |
| 12:30.33 | seebs | Hang on. |
| 12:30.43 | seebs | I see a difference. |
| 12:30.56 | seebs | I am doing SetPoint("CENTER", UIParent, "TOPLEFT", mx, my) |
| 12:31.02 | seebs | You're using your context. |
| 12:31.05 | seebs | I wonder whether that matters. |
| 12:31.49 | seebs | Doesn't seem to on my end, anyway. |
| 12:32.01 | Noshei | trying that |
| 12:32.13 | Noshei | no difference |
| 12:32.33 | Noshei | which it shouldn't |
| 12:33.19 | seebs | Right. |
| 12:33.49 | seebs | I am super happy with the visual effect of a thing which has a thickness and size scaling with cursor speed. |
| 12:34.33 | seebs | Sorta wanna do a thing where when the cursor stops there's a secondary animation or something. Hmm. |
| 12:35.20 | Ivnedar | have a delay on when the pulsing stops? |
| 12:35.33 | Ivnedar | ooh, or give the circle momentum, so it bounces like a spring? |
| 12:35.35 | seebs | Hmm. That's a thought. |
| 12:36.02 | seebs | Right now it's just a 20-frame cycle of (x+1)%20, and then math.abs(x - 10) for a modifier on circle size. |
| 12:36.29 | seebs | Man, though. The frame or two of lag is really annoying. |
| 12:36.50 | seebs | Thinking about it, a ton of UI stuff has lag like that. |
| 12:37.02 | seebs | Like, my addon that puts stuff behind chat windows, there's a frame or so of delay. |
| 12:37.26 | seebs | And the item tooltips, Event.Tooltip shows up, and it's not until a frame or so later that UI.Native.Tooltip:GetFoo() show meaningful values. |
| 12:37.31 | Noshei | really wish I knew why seebs's code doesn't cause fps drop but mine does |
| 12:37.36 | seebs | Me too! |
| 12:37.39 | Imhothar | many games have a mouse lag, that's a result of how they do message handling in the main loop |
| 12:38.18 | seebs | So as I understand it, if you comment out all the code other than the :SetPoint(), you still get that performance hit. |
| 12:38.32 | Noshei | yup |
| 12:38.55 | Noshei | maybe it is because it is global |
| 12:38.59 | seebs | And if you drop the :SetPoint, it doesn't. |
| 12:39.23 | seebs | The lookup cost of a global shouldn't be that big, not by several orders of magnitude. |
| 12:39.38 | seebs | Well, easy to check. |
| 12:39.44 | seebs | s/CUCI = {}/local CUCI = {}/ |
| 12:40.30 | Noshei | yeah just did that, no difference |
| 12:40.33 | seebs | Hmm. Okay, and it's not the strata difference (I did "topmost", but "modal" is no different). |
| 12:41.34 | Noshei | I've also had it set to hud |
| 12:41.40 | Noshei | no difference there |
| 12:41.50 | seebs | Hmm. |
| 12:42.57 | Imhothar | you could get rid of many of those ifs everywhere if you registered to Mouse.Move after everything is loaded |
| 12:42.59 | Noshei | thought maybe it was SetBackgroundColor(0,0,0,0), but just removed that with no difference |
| 12:43.30 | Noshei | I've got all that commented out right now |
| 12:44.45 | seebs | hmm |
| 12:44.59 | seebs | Okay, how about this. Put the .toc and .lua somewhere and I'll try them here. |
| 12:46.48 | Noshei | https://drive.google.com/file/d/0B9iQSZB7cVyEUDAyTGg1aVFrZzA/edit?usp=sharing |
| 12:47.06 | Noshei | just hit file download |
| 12:47.11 | Noshei | will download a zip |
| 12:51.39 | seebs | oooh |
| 12:51.40 | seebs | I get it too |
| 12:52.21 | seebs | And I don't with mine, even if I have mine set to do three canvases and update a different one each frame to make trails. |
| 12:52.48 | seebs | hmm |
| 12:52.56 | seebs | Takes a while of moving the mouse before I notice it, though... |
| 12:53.11 | seebs | hang on |
| 12:53.19 | seebs | took longer the first time, then repeating it it happened sooner |
| 12:53.41 | seebs | So like 10 seconds of moving, no effect, at first. |
| 12:53.53 | seebs | But once it's started, moving the mouse drops frame rate almost immediately. |
| 12:59.04 | seebs | Inspect.Addon.Cpu is showing a huge bump in ccContextCircle render time. |
| 12:59.43 | Noshei | yeah it is |
| 13:01.15 | seebs | hmm |
| 13:01.25 | seebs | I don't know anything about LibSimpleWidgets. But I noticed a thing. |
| 13:01.32 | seebs | Which is that when I screwed something up, it was in the call trace. |
| 13:02.20 | seebs | So CUCI.CreateCircle() appears to be getting called in a context that may have implications not obvious to me. |
| 13:02.34 | seebs | Hmm. |
| 13:03.06 | Ivnedar | I'm still amused that Track Fish is a pvp buff :D |
| 13:03.14 | seebs | Okay. CUCI.Circle = UI.CreateFrame(...) |
| 13:03.17 | seebs | Break this line. |
| 13:03.24 | seebs | Like, change it to CUCI.ContextCircleNot or something. |
| 13:03.37 | seebs | And... get an error from LibSimpleWidgets/uicreateframe.lua |
| 13:03.39 | seebs | Why? |
| 13:04.19 | Ivnedar | because libsimplewidgets is naughty |
| 13:04.22 | seebs | Huh. Well, that affects my code too. |
| 13:04.26 | Ivnedar | and replaces UI.CreateFrame |
| 13:04.29 | Noshei | ^ |
| 13:04.37 | seebs | gah |
| 13:04.39 | seebs | that is horrible |
| 13:04.41 | Ivnedar | :D |
| 13:05.17 | Ivnedar | what was the error? assertion about parameters? |
| 13:05.50 | seebs | Yeah. I just didn't expect to see something else in the way. |
| 13:06.02 | seebs | Okay, what it's doing looks like it should be harmless. Hmm. |
| 13:07.42 | seebs | Is there any kind of addon memory check we can do? |
| 13:08.02 | seebs | hmm |
| 13:08.33 | seebs | Interesting. If I change the SetPoint to 23, 23 instead of xcoord, ycoord, it seems to get better. |
| 13:12.38 | seebs | Okay. |
| 13:12.53 | seebs | Imagine that I change the SetPoint line to (given a nearby local o which is file scope and thus persistent) |
| 13:13.03 | seebs | SetPoint("CENTER", UIParent, "TOPLEFT", 270, 270 + o) |
| 13:13.15 | seebs | If that's it, I get no frame rate drop. |
| 13:13.20 | seebs | If I follow that by: |
| 13:13.27 | seebs | o = (o + 1) % 24 |
| 13:13.42 | seebs | I get the frame rate drop. So :SetPoint() with varying locations is resulting in Something. |
| 13:15.19 | seebs | AH HAH |
| 13:15.21 | seebs | I found a thing |
| 13:15.31 | Noshei | ? |
| 13:15.57 | seebs | Your AddonLoaded() will call CreateCircle every time ANY addon loads. This may be harmless, but it's Suspicious. |
| 13:16.29 | seebs | Because your context has then a lot of things in it, most of which we have no lua handles for. |
| 13:17.35 | seebs | hmm but fixing that doesn't seem to change things |
| 13:17.43 | Noshei | same result for me |
| 13:20.34 | Noshei | though fps drop is better |
| 13:21.29 | Noshei | oh I know why seebs |
| 13:21.48 | seebs | Oh, wait. Because that's on Addon.Startup.End, so it does only get called once. |
| 13:21.51 | Noshei | I'm using Event.Addon.Startup.End, which is the end of ALL addons loading |
| 13:21.52 | seebs | Nevermind. |
| 13:23.39 | seebs | Meanwhile, my addon can have NINE cursors and still not visibly slow down. |
| 13:23.48 | seebs | There is NO REASON FOR THIS. |
| 13:24.51 | Noshei | I agree |
| 13:25.29 | Ivnedar | lol, pug in this raid is qq'ing that they weren't healed through standing in bad stuff since they're melee |
| 13:25.35 | Noshei | I'm going to try adding the SetShape to the mouse move event, see if that makes a difference |
| 13:25.38 | Ivnedar | we had to point out that they had soul sickness and 7k hp ... |
| 13:26.55 | Noshei | WTF |
| 13:27.03 | Noshei | WTF |
| 13:27.05 | Noshei | WTF |
| 13:27.47 | seebs | Lemme guess, :SetShape + :SetPoint is faster than :SetPoint? |
| 13:27.57 | Noshei | yes |
| 13:28.04 | Noshei | but SetShape is after SetPoint |
| 13:28.11 | Noshei | like, WTF |
| 13:28.18 | seebs | Huh! |
| 13:28.56 | Noshei | maybe redrawing is faster than moving |
| 13:29.00 | seebs | Okay, I'll drop my :SetShape and see. |
| 13:29.49 | seebs | oooh this is interesting |
| 13:29.53 | Noshei | yeah |
| 13:30.30 | Noshei | only thing I can think of is that it is faster to completely redraw it than to do the translations on everything to move it |
| 13:30.39 | seebs | what the hell |
| 13:30.46 | seebs | Okay so. |
| 13:30.59 | seebs | I changed it so I do a :SetShape() on each of 9 cursors once up front, and don't do them thereafter. |
| 13:31.10 | seebs | All of them are coming out as the SAME thing which is sort of a cumulative set of paths. |
| 13:31.44 | Noshei | hu? |
| 13:31.48 | Ivnedar | you need to setshape again after resizing |
| 13:32.26 | seebs | Okay. But the thing that's confusing me is, I'm ending up with a thing which looks a lot like each of the 9 canvas objects has a shape which is roughly what I would have expected from the sum of all the shapes next to each other. |
| 13:32.42 | Noshei | lol |
| 13:33.10 | Noshei | okwe ll I'm going to AFK for a bit |
| 13:33.17 | Noshei | GAH, WTF |
| 13:35.05 | seebs | Okay, yeah, doing SetWidth or SetHeight without a SetShape produces madness. |
| 13:35.17 | Ivnedar | yup yup |
| 13:35.19 | seebs | But... With none of those, just a setpoint, I still never see a performance hit. And I have no idea why. |
| 13:36.33 | Noshei | so I was about to go somewhere real fast, been really nice out the last week or so and weather is like mid 50's for the next week. So I just throw on some sandles, open my door, and there's a nice 1/5 inch of snow outside |
| 13:36.56 | seebs | <-- Minnesotan does not see how this is not nice sandles weather. |
| 13:37.37 | seebs | Okay, I *can* get the performance drop. |
| 13:37.41 | seebs | With my code. |
| 13:37.52 | seebs | It just takes a really long time, and also no SetShape in the loop, I think. |
| 13:38.27 | Noshei | lol seebs |
| 13:38.54 | seebs | I'm talking like a *minute* spent watching the pretty shapes I could get from a sort of not-very-good design. |
| 13:39.06 | Noshei | I doubt you guys have to deal with the crazy weather changes we get here in colorado |
| 13:39.08 | seebs | And finally I saw the frame rate drop. Added SetShape back in, will see if I can do it. |
| 13:39.10 | seebs | ... Haha. |
| 13:39.33 | seebs | I think MN is crazier weather than Colorado. And yes, I have spent time there, I have coworkers there and my mom lived in Denver for a decade. :) |
| 13:39.48 | Noshei | my guess is it will br sunny all day and there wont be any of this 1/2 inch left by the end of the day |
| 13:40.38 | Noshei | anyways, now that I have shoes on, afk a bit |
| 13:41.15 | seebs | I think my theory is, there's some clever caching going on for the case where you :SetPoint a canvas without :SetShape, and it is becoming pathological in some edge case. |
| 13:43.37 | seebs | Ayup! |
| 13:43.49 | seebs | If I :SetPoint on all nine cursors every frame, and don't :SetPath, I get horrible performance. |
| 13:44.40 | seebs | And it does get noticably worse over time. Hmm. |
| 13:45.38 | seebs | And if I :SetPoint on all nine, and :SetPath on one every frame, no degradation. |
| 13:45.50 | seebs | Zorba! THIS IS MADNESS. Just so you know. |
| 13:47.45 | Ivnedar | SetShape creates a new scaleform object |
| 13:48.25 | Ivnedar | so if there's something building up inside that object, creating a new one will clear it out |
| 13:48.59 | seebs | A lookup table of locations the object has previously held, each with a pointer to a full-screen bitmap of how to draw it? |
| 13:49.07 | seebs | ... nah, rotating through a small number of positions triggers it too. |
| 13:49.19 | Ivnedar | each object does a deep copy of _G every frame |
| 13:49.28 | Ivnedar | hides |
| 13:50.13 | seebs | In any event, mystery basically solved. |
| 13:50.23 | seebs | :SetPath shapes when setting their points and life is okay. |
| 14:05.31 | Noshei | lol, watching where the mouse is when I'm holding down right mouse is kinda funny |
| 14:27.57 | Imhothar | wish there was a way to punish every single developer who thinks its a great idea to drop all their nonsense into "My Documents" |
| 14:34.13 | Imhothar | it's really terrible to see how heavily this folder is misused |
| 14:34.23 | Imhothar | they don't do it on macs, so why on windows |
| 15:13.41 | Thorarin | Dunno about macs, but yeah it's annoying :P |
| 15:15.00 | Thorarin | According to some developers, we might as well consider My Documents the new C: |
| 15:15.35 | Imhothar | it sucks. there's ProgramData, AppData, Saved Games and whatnot |
| 15:15.50 | Imhothar | on my mac everything in the Documents folder was placed there by me |
| 15:19.39 | *** join/#riftuidev nywef (~fewyn@fewyn.net) |
| 15:19.55 | *** join/#riftuidev Jydan (marcel@213.125.209.53) |
| 15:19.55 | Jydan | 0.o |
| 15:20.01 | *** mode/#riftuidev [+v fewyn] by ChanServ |
| 15:20.25 | Jydan | Do you know of any Game that actually uses "Saved Games"? |
| 15:20.55 | Imhothar | there are some |
| 15:21.23 | Imhothar | AC Brotherhood, C&C 3, Crysis2, Rage |
| 15:21.40 | Imhothar | interestingly it's the only AC game there |
| 15:21.45 | Jydan | Because if people used it, that would take care of a substantial part of the crap in my My Documents :P |
| 15:21.59 | Thorarin | C&C 3 dumps it in My Documents |
| 15:22.25 | Thorarin | Or at least it creates a directory there too |
| 15:22.41 | Thorarin | Never played Brotherhood... |
| 15:22.49 | Imhothar | the steam version doesn't it seems |
| 15:22.55 | Thorarin | Only AC3 and AC4, and they both use My Documents I think :< |
| 15:23.15 | Thorarin | For some INI file it seems |
| 15:23.19 | Thorarin | Like wth :P |
| 15:24.09 | Thorarin | The problem has started because people sort of know how to back up their My Documents, but AppData is much more difficult to handle properly |
| 15:24.44 | Thorarin | In the process of reinstalling GFs PC today, so I can relate to that somewhat :S |
| 15:37.20 | *** join/#riftuidev Thorarin (marcel@D57DD135.static.ziggozakelijk.nl) |
| 15:37.34 | Thorarin | What's up with freenode :S |
| 15:47.14 | *** join/#riftuidev NuSuey_ (uid3556@gateway/web/irccloud.com/x-orsmogixwdcyzpjf) |
| 16:01.31 | Repo | raidcooldowns: quickgric * r35 (2 files in 1 directory): 1.2 fixes |
| 17:06.27 | *** join/#riftuidev HechtiQ (~yoram@5357C46A.cm-6-8d.dynamic.ziggo.nl) |
| 17:10.34 | *** join/#riftuidev Cairenn (~Owner@MMOI/Administratrix/Cairenn) |
| 17:10.34 | *** mode/#riftuidev [+o Cairenn] by ChanServ |
| 17:42.27 | ZorbaBeta | I'll give microsoft credit: I now have five separate versions of visual studio installed, and they don't step on each other's toes *too* much. |
| 17:45.58 | HechtiQ | sounds like a cause of celebration, pop the champagne! :) |
| 18:14.34 | seebs | Zorba, did you see the strange discoveries pertaining to :SetPoint on Canvas objects? |
| 18:23.05 | ZorbaBeta | nope, what were they? wouldn't surprise me if there's some oddness |
| 18:26.12 | seebs | Basically, if you make a canvas object, and in Event.Mouse.Move, canvas:SetPoint("CENTER", UIParent, "TOPLEFT", xcoord, ycoord), and then you move the mouse around a lot, your frame rate will suddenly drop catastrophically. |
| 18:26.42 | seebs | If you do the SetPoint, and then follow it immediately by :SetShape(path, fill, stroke) with the exact values used to create it, the frame rate will not drop within the first minute or two anyway. |
| 18:27.02 | seebs | So *something* accumulates if a Canvas is moved around a lot without being re-:SetShape()d. |
| 18:27.15 | ZorbaBeta | Huh. Interesting. Alright, good repro steps there, I'll take a look at it :) |
| 18:28.23 | seebs | I note in passing: It might be sort of cool if a canvas could have its color multiplied without needing to re-specify the path. |
| 18:28.53 | seebs | ... this may be possible, come to think of it. |
| 18:29.20 | seebs | Hmm. No, because :SetBackgroundColor sets the color of the space behind the object. |
| 18:29.49 | seebs | Oh, also. |
| 18:29.59 | seebs | There is a very interesting couple-frames-of-lag. |
| 18:30.43 | seebs | If you are :SetPoint()ing something to the coordinates of mouse move events in Event.System.Update.Begin, it doesn't just appear to lag the cursor, but appears to lag the cursor *by more than one frame*. |
| 18:31.18 | seebs | If you have a rotating series of objects you're :SetPoint()ing, and you move the mouse at a steady pace, it's quite noticeable that the one closest to the mouse is two or three "steps" back from the mouse's location. |
| 18:33.50 | ZorbaBeta | there's a bunch of things that would be kind of cool, but given that I need to reconstruct the path each time, it'd be a lot of backend complexity without a lot of performance gain |
| 18:33.59 | ZorbaBeta | so, if you want that, implement it with your own interface ;) |
| 18:34.24 | ZorbaBeta | yeah that wouldn't surprise me - if you have a good graphics card and low detail settings, the graphics card will be able to queue up a few frames |
| 18:34.34 | ZorbaBeta | I may try to clamp the lookahead to one frame at some point in the future |
| 18:38.55 | seebs | This is on nigh-maximum settings, as I recall. GeForce 670 or so. |
| 18:39.14 | ZorbaBeta | well, okay, if you're in a non-combat situation at least ;) |
| 18:39.18 | seebs | I would have assumed that at frame rates well below refresh rate, the card wouldn't be far ahead. |
| 18:39.25 | seebs | And I would kill for a frame rate limiter. |
| 18:39.37 | ZorbaBeta | actually you know what I've got this backwards - if it's clamping on the GPU, not on the CPU. which means high detail (in a non-combat situation) |
| 18:39.37 | seebs | e.g., a non-background equivalent of the minimum frame time when backgrounded. |
| 18:39.55 | ZorbaBeta | wouldn't be too hard to make that, but why are you looking for it? |
| 18:40.01 | seebs | Laptop fans. |
| 18:40.06 | ZorbaBeta | ah, makes sense |
| 18:40.19 | seebs | I am frequently on machines which can manage 20fps smoothly, or 29fps while making a noise like a jet engine. |
| 18:40.37 | seebs | Also, in some cases, a *steady* 20 is better than an erratic 30ish. |
| 18:41.44 | seebs | I am thinking to make a cursor-trails addon which tries to sort of highlight where the cursor goes, with low-alpha display when it's not moving, and then when it stops, a little burst of higher alpha. |
| 18:42.07 | seebs | And with multiple separate bits doing this, so they provide a visible sense of direction. |
| 19:15.25 | *** join/#riftuidev Cairenn (~Owner@MMOI/Administratrix/Cairenn) |
| 19:15.25 | *** mode/#riftuidev [+o Cairenn] by ChanServ |
| 20:43.54 | *** join/#riftuidev Cairenn (~Owner@MMOI/Administratrix/Cairenn) |
| 20:43.54 | *** mode/#riftuidev [+o Cairenn] by ChanServ |
| 21:07.35 | seebs | Hmm. |
| 21:07.47 | seebs | ZorbaBeta, or anyone else: Any suggestions on a way to determine that someone is Moving The Camera? |
| 21:08.17 | seebs | I'd like to suppress display of my mouse-tracking stuff when the cursor has gone invisible, basically. But there's no Event.Cursor for that. |
| 21:10.35 | ZorbaBeta | hmm, I suspect there isn't |
| 21:12.18 | seebs | BTW, after a year or three of messing with it, I have concluded: I think it would have been better for saved variables to live in some addon-specific namespace. |
| 21:12.49 | seebs | Because naming saved variables is a source of mild clutter, and once there's 20+ addons in the system, they can get sorta distracting. |
| 21:12.55 | seebs | Also, references into _G are slow. :P |
| 21:17.02 | ZorbaBeta | Possibly, yeah. In that case I basically said "WoW's method is good enough, I'll just stick with that" |
| 21:17.17 | ZorbaBeta | note that luajit uses a hash table so adding more things into _G probably isn't slowing it down much :) |
| 21:17.22 | seebs | http://www.riftui.com/downloads/info428-HereMousey.html |
| 21:17.44 | seebs | Part of the issue is just that if I forget what I named a variable, it's a pain to find it. :) |
| 21:19.18 | seebs | For the "if there's a ton of free time some day" list: Deprecate their presence in _G, make a table with those members an additional parameter to the variables loaded event. |
| 21:20.02 | seebs | So, HereMousey now exists and is of course configured by default to make SHINY RAINBOWS around cursors. |
| 21:20.42 | seebs | This is not going to *decrease* the frequency with which one of the people I know in game accuses me of being a "fag". (Note: He also accuses me of being a "nerd" for trying to do science on game mechanics. And spends hours helping me do said science.) |
| 21:21.31 | ZorbaBeta | problem is, I could "deprecate" them but I'd probably never be able to remove them. the event redesign was justified because things were absolutely broken |
| 21:21.40 | ZorbaBeta | this redesign is more of a "this is slightly annoying" deal |
| 21:21.57 | ZorbaBeta | and I'd rather have one slightly-dubious design than a slightly-dubious design in parallel with a good design, both of which now have to be maintained :) |
| 21:23.43 | seebs | Point. |
| 21:24.15 | seebs | Although honestly, I'd love to have that table *anyway*. :) |
| 21:26.34 | *** join/#riftuidev Cairenn (~Owner@MMOI/Administratrix/Cairenn) |
| 21:26.34 | *** mode/#riftuidev [+o Cairenn] by ChanServ |
| 21:47.10 | *** join/#riftuidev HechtiQ (~yoram@5357C46A.cm-6-8d.dynamic.ziggo.nl) |
| 22:37.32 | Thorarin | <seebs> BTW, after a year or three of messing with it, I have concluded: I think it would have been better for saved variables to live in some addon-specific namespace. |
| 22:37.46 | Thorarin | Hmm, concluded that after looking about 3 minutes at the addon API tbh :S |
| 22:40.00 | seebs | Oh, I suggested that in one of my first few threads. But after time I've concluded that it's a slightly bigger nuisance than I anticipated. |
| 23:07.01 | *** join/#riftuidev Noshei (~quassel@c-24-9-171-99.hsd1.co.comcast.net) |