Chat log from the meeting on 2018-09-18

From OpenSimulator

Jump to: navigation, search

[11:17] Arielle.Popstar anyway imo dev's and testers should have mesh avi's to be able to test how well they rezz
[11:18] Andrew.Hellershanks Arielle, are you going to supply the mesh avis?
[11:18] Arielle.Popstar sure
[11:18] Sheera Khan: unfortunately there are few female devs and the male mesh bodies aren't ready yet ^^
[11:18] Bill Blight: legitmate ones?
[11:18] Arielle.Popstar Sheera already did last year anyway
[11:18] Kayaker Magic: Isn't Ruth 2.0 available yet?
[11:18] Arielle.Popstar brought the avastar avi's to a meeting
[11:18] Bill Blight: yep
[11:18] Arielle.Popstar that too Kay
[11:19] Sheera Khan: Ruth yes... I'm wearing one
[11:19] Bill Blight: and works great
[11:19] Arielle.Popstar aand Hyacinth has a working roth
[11:19] Sheera Khan: with alphas?
[11:19] Arielle.Popstar cant remember as i dont wear male mesh bodies
[11:20] Arielle.Popstar but can use rigged mesh clothing for him
[11:21] Andrew.Hellershanks I haven't tried using an avi made using the Bastioni LAB system
[11:24] Arielle.Popstar whats this about Npc's not working anymore?
[11:24] Arielle.Popstar or properly in .9
[11:24] Bill Blight: there is one sitting at the beach here, they work fine
[11:24] Andrew.Hellershanks What do you consider "not working"?
[11:25] Arielle.Popstar not exactly sure but have heard several sources mention it in the last few weeks
[11:25] Bill Blight: I use NPC's daily, not sure what is not working
[11:25] Arielle.Popstar between .8.2 but also not the same in different versions of .9
[11:26] Andrew.Hellershanks Arielle, get them to file a bug report or tell you more specifically what is the problem.
[11:26] Bill Blight: My bunker back home has them roaming around
[11:26] Sheera Khan: ideally file a Mantis ...
[11:27] Andrew.Hellershanks Depends whether it is a real problem or not. Either way, not much can be done without details.
[11:27] Bill Blight: The only thing I can think of that has changed with npc's is the fact that parcel perms work the way they are supposed to and if there is NO access to the parcel or region the NPC's will end up in the corner
[11:28] Arielle.Popstar no access?
[11:28] Bill Blight: but that is easily fixed by setting group land and rezzing the NPC's under group
[11:28] Arielle.Popstar and whose interpretation of "how they should work"?
[11:29] Bill Blight: oh here we go
[11:29] Arielle.Popstar ~smiles~
[11:34] Andrew.Hellershanks I've had too many things on my plate this past week so I haven't finished the updates to the OSSL wiki pages which I want to make.
[11:36] Andrew.Hellershanks Kayaker, did you get your (test?) grid setup?
[11:37] Ubit Umarov: about changes, there was issue with the xmlrpc library on master, not allowing utf8
[11:37] Ubit Umarov: i could not find the sources of the one we did use :)
[11:37] Ubit Umarov: but seems fixed now
[11:38] Kayaker Magic: Yes! I think I reported something last week. Have "Snail Dev" running with YEngine on OSGrid. Working fine, but I have not seriously tested it yet. One big script worked flawlessly.
[11:38] Andrew.Hellershanks Ubit, It isn't in opensim-libs or libOMV?
[11:38] Bill Blight: I did submit a patch to stop avatar names from containing any HG address delimiter such as @ : and .
[11:38] Ubit Umarov: no andrew
[11:38] Ubit Umarov: the one in use on master is at -libs now
[11:39] Ubit Umarov: it is almost as original, just did not had the uft8 patches older core ppl made
[11:39] Ubit Umarov: so i made my own :)
[11:40] Andrew.Hellershanks :)
[11:40] Andrew.Hellershanks I see it in -libs.
[11:40] Andrew.Hellershanks I wonder if the source for any other DLL's we use is missing
[11:41] Ubit Umarov: i did recompile most of them, or got newer versions for .net4.6
[11:41] Ubit Umarov: don't remember i missed one
[11:42] Andrew Hellershanks: ok, good.
[11:42] Andrew Hellershanks: I wonder if the xmlrpc.dll file originally came from the XML-RPC.NET website.
[11:42] Ubit Umarov: someone told opensim wants .net3.5 id so some missing, but can't test now
[11:42] Andrew Hellershanks: Doesn't matter too much now that we have source in -libs.
[11:43] Ubit Umarov: no code does not match
[11:43] Andrew Hellershanks: ok
[11:43] Ubit Umarov: guess a older comum root
[11:43] Andrew Hellershanks nods
[11:44] Ubit Umarov: as usual we don't use it all.. and the *rpc* as even more things we don't use
[11:45] Ubit Umarov: in fact i seem mails from back 2007 core considering using that one, but deciding not
[11:47] Ubit Umarov: there is a report of TP issues if using gloebits.. but i've no clear idea why still :(
[11:48] Andrew Hellershanks: Hm... that would be a matter for the Gloebit people.
[11:48] Ubit Umarov: hmm not sure if last week or before, but secure http request was not working, by mistake i had a line commented to do main region ssl testing :)
[11:49] Ubit Umarov: yes i had a little chat with colosi
[11:49] Ubit Umarov: at least they are going to light the code on some time critical events, doing it async
[11:50] Bill Blight: They are also going to opensource their module soon
[11:50] Andrew Hellershanks: They are? That's good. The Virwox people did the same.
[11:51] Ubit Umarov: i did fix that LSL secure httprequest, not in i a way also compatible with main use of SSL and money module ( that also did broke it )
[11:51] Bill Blight: Chris said they will be releasing it soon,
[11:52] Andrew Hellershanks: Ubit, did your change also affect the DTL/NSL money module?
[11:53] Ubit Umarov: that did collide with the secure lsl http yes
[11:53] Andrew Hellershanks: ok
[11:53] Ubit Umarov: and still does collide with main region use of ssl
[11:54] Ubit Umarov: well i did not look to recent versions
[11:55] Ubit Umarov: got a few reports of issues, but where just changes from 0.82
[11:55] Ubit Umarov: like those npcs issues arielle told..
[11:55] Ubit Umarov: of that funny problem fred had :)
[11:55] Andrew Hellershanks: I've noticed that someone has been updating that money module so I would expect they will make any required adjustments need for the lastest version of 0.9
[11:56] Ubit Umarov: all npcs piling up at region edge :)
[11:56] Ubit Umarov: that was as Bill told
[11:57] Ubit Umarov: NPCs are under region access control.. not rights to be on parcel, they get pushed to edge
[11:57] Andrew Hellershanks: Just a few minutes left before the top of the hour. Any one else have any other OS issues to discuss that hasn't been talked about yet?
[11:57] Bill Blight: I just think it's funny, people want it like SL where NO-ACCESS means NO-ACCESS ... you fix that and they are like ,"but but but but but" ...
[11:57] Andrew Hellershanks: hehe
[11:58] Bill Blight: simple fix, set land to group , rez NPC's with group .. and presto
[11:58] Andrew Hellershanks nods
[11:59] Bill Blight: Oh a side note, very happy with YEngine ... No issues that were not badly written script related
[12:00] Andrew Hellershanks: Bill, good to hear.
[12:00] Andrew Hellershanks: I've been working through my ToDo list so I haven't gotten to testing YEngine yet.
[12:01] Bill Blight: the regions I have switched it to, the users say scripts are running faster and "better" ... Not sure about the better part but I do see a speed improvement , or should say a lack of "loosing speed"
[12:02] Bill Blight: I threw 500 fast timer scripts at YEngine, and it did not even blink, everything fired as it should , as far as I could tell .. hard to watch 500 scripts run
[12:02] Andrew Hellershanks: Bill, do you see much in the way of improvements from YEngine when crossing between regions running YEngine?
[12:02] Bill Blight: Y to Y is faster
[12:03] Bill Blight: than X ever was
[12:03] Andrew Hellershanks: Bill, combine a fast timer with a call to a texture change in the timer handler. That will cause really high runtimes with XEngine. It would be another way to test YEngine to see how it behaves.
[12:03] Bill Blight: region startup is much faster too, where 2000 scripts may take a couple of minutes to fire up, with Y it is a few seconds at most ..
[12:04] Andrew Hellershanks: That is impressive.
[12:04] Dev Random: Does YEngine remove the built-in delays in LSL? e.g. some ops pause for a time.....
[12:04] Bill Blight: no
[12:04] Dev Random: kk
[12:05] Bill Blight: It loads the same API as X
[12:05] Andrew Hellershanks: Bill, how do the script runtime numbers compare between X and Y for a given script?
[12:05] Bill Blight: well, in the viewer they suck, as it does not seem they ever reset with Y
[12:06] Bill Blight: but they seem to be right on the console
[12:06] Andrew Hellershanks: Hm... interesting that you would get two different numbers. I would expect the source for the numbers to be the same place. Perhaps an update isn't being pushed out to the viewer.
[12:06] Bill Blight: I'd say Y has the advantage but as far as that goes it adds up so every little bit helps
[12:07] Andrew Hellershanks nods

[12:08] Andrew Hellershanks: If the performance of YEngine is noticably better it might also mean that you can have more people in a given region at the same time before lag starts to become noticable.
[12:08] Bill Blight: I mean 100 scripts may not appear to run faster but you get over a 1000 and there is a noticible diff with Y
[12:08] Dev Random: Unrelated topic... Is anyone emotionally attached to the "Command" object in code? It enforces parameter count, and is unnecessarily restrictive.
[12:09] Dev Random: e.g. When I added the "terrain modify" operations, I couldn't use the Command because I had optional params.....
[12:09] Andrew Hellershanks: Bill, In XEngine I tried to keep runtimes close to 1 or less with YEngine under 0.8 but I still haven't determined what the new target number should be under 0.9
[12:09] Dev Random: (sorry for the late question... I can wait 'til next week to discuss...)
[12:10] Bill Blight: well, Andrew again that is subjective, because it depends on the type of scripts
[12:10] Andrew Hellershanks: Dev, Can you give an example of how it is a problem for you?
[12:11] Andrew Hellershanks: Bill, subjective perhaps but under 0.8 if you start getting runtimes that exceed 1 by too much the script may need changes.
[12:12] Dev Random: Andrew: In the "terrain modify" operations, I wrote it so if you want a circle, you can specify a single size param for "ellipse" intead of having to specify both... Command enforces param count, so optional params can't happen
[12:13] Dev Random: The Terrain module is written using Command class...
[12:14] Aaack Aardvark: Sorry if Im silent I'm pesting ubit for two reasons, one is that it seems im perpetually child agent in some regions and two I just love to pest ubit =P
[12:14] Bill Blight: Aaack Aardvark ec243d82-cb0d-44c0-b424-33f9e070dd25 Root Dev Outreach <359.5694, 406.9327, 24.05366>
[12:14] Bill Blight: you immediately became root here
[12:14] Aaack Aardvark: yes I can't understand but its persistent and reproducible on my sim and on another with another server connection
[12:15] Andrew Hellershanks: ok. I know there are some commands I've used on occasion where I have had to put in all parameters. To make parameters optional would mean you may need a way to specify which of the optional arguments you are supplying.
[12:16] Dev Random: Andrew: Yes... e.g. if you don't specify the final param, then the second-last one is used for it... as in the case with ellipse --> circle
[12:16] Dev Random: specify radius, as opposed to radius-x and radius-y
[12:16] Andrew Hellershanks: If a command could take 6 args and you provide only 4 then it would have to assume the last two are not being used. If the last of the 4 you provided was meant to be one of the last two of the 6 possible args that could lead to problems.
[12:17] Andrew Hellershanks: hm... I haven't looked at the terrain modify command that closely. I didn't remember it could take ellipse or circle parameters.
[12:17] Sheera.Khan I thought OOP would solve that problem by overloading the method calls?
[12:19] Andrew Hellershanks: Sheera, that may be true within the code. It isn't the same on the command line.
[12:20] Dev Random: Andrew: e.g. for a "fill" command: fill <height> [ -rec=x1,y1,dx[,dy] [-taper=<height2>] ... dy is optional, and if not supplied, then dx is used. The "-taper" param is optional.
[12:20] Andrew Hellershanks: It could become "interesting" to know if the user has supplied all the required arguments for their particular use of a command.
[12:20] Dev Random: Most of OS does NOT use Command..... I think it's only in 2 modules
[12:21] Andrew Hellershanks: I think you would need something in each command line that lets you make sure which variant of a command was being invoked
[12:22] Andrew Hellershanks: Dev, I'm not familiar with the details of that section of the OS code. You could discuss the matter further on the mailing list or suggest a code change via the bug tracking system.
[12:22] Dev Random: Andrew: LOL... IIRC, Justin hated Command.
[12:23] Ubit Umarov: ive the feeling i hate it 2 :p
[12:23] Andrew Hellershanks: That is encouraging. It sounds like a patch to remove Command would be welcome.
[12:24] Dev Random: heh.. that's what I was wondering about... Before I start into it, I wanted to see if anyone would miss it.

Personal tools
About This Wiki