Chat log from the meeting on 2023-04-04

From OpenSimulator

Jump to: navigation, search
[11:02 AM PT]  Lyr Lobo: by blow up, what do you mean?
[11:02 AM PT]  Andrew Hellershanks: Hello, everyone.
[11:02 AM PT]  Orbert.Tatham I use the profile stuff, and it is broke
[11:02 AM PT]  Orbert.Tatham Andrew
[11:02 AM PT]  Gavin.Hird to block scripted agents to estate or region
[11:03 AM PT]  Gavin.Hird Hi Andrew
[11:03 AM PT]  Ubit Umarov: i will not add that one
[11:03 AM PT]  Ubit Umarov: i seen it..
[11:03 AM PT]  Gavin.Hird why not?
[11:03 AM PT]  Cuga.Rajal file a Jira, hopefully they fix in next release
[11:03 AM PT]  Ubit Umarov: the block bots, right?
[11:03 AM PT]  Gavin.Hird it does, if you set it
[11:04 AM PT]  Ubit Umarov: you do mean the "block bots" flag right?
[11:04 AM PT]  Gavin.Hird scripted agent is a bot I suppose
[11:04 AM PT]  Ubit Umarov: yes
[11:04 AM PT]  Ubit Umarov: ok those are special sl things
[11:05 AM PT]  Gavin.Hird probably a good thing to have
[11:05 AM PT]  Vincent.Sylvester How is it meant to even detect that?
[11:05 AM PT]  Gavin.Hird viewer has code for it
[11:05 AM PT]  Ubit Umarov: if you use a viewer that is actually a viewer side bot, you need to flag that on the SL account
[11:06 AM PT]  Vincent.Sylvester XD "Let's rely on people to be honest" type deal then
[11:06 AM PT]  Gavin.Hird that was just to set it from the UI
[11:06 AM PT]  Ubit Umarov: so that flaq will only work og accounts that did declare to be a bot
[11:06 AM PT]  Orbert.Tatham Sounds like wishful thinking
[11:06 AM PT]  Lyr Lobo: Those are for accounts that are driven only by a bot script or chatbot service or a custom scripted server.
[11:06 AM PT]  Ubit Umarov: the only thing to force that is SL TOS
[11:07 AM PT]  Vincent.Sylvester Oh this is hilarious
[11:07 AM PT]  Lyr Lobo: we used them for our Learning Center for an automated tour guide that is an avatar
[11:07 AM PT]  Lyr Lobo: long ago
[11:07 AM PT]  Ubit Umarov: ofc i consider that just.. useless..
[11:07 AM PT]  Lyr Lobo grins and nods
[11:07 AM PT]  Vincent.Sylvester SL, full of copybots, playing some 1d chess there
[11:07 AM PT]  Ubit Umarov: so we do not have such account flag
[11:08 AM PT]  Ubit Umarov: so no point adding it to regions, right?
[11:08 AM PT]  Vincent.Sylvester Only use we could possibly have for that is to deny parcel owners the use of NPCs on a estate level
[11:08 AM PT]  Ubit Umarov: in principle viewerrs will be happy without region setting it
[11:09 AM PT]  Lyr Lobo: In commercial worlds, people may be using avatar-account bots to inflate their traffic counts, which spurred a change in ToS a few days ago about it. But that does not affect us here
[11:09 AM PT]  Ubit Umarov: well yeah we could do it for npc..   but confusing
[11:09 AM PT]  Orbert.Tatham Would that no block animesh as well?
[11:09 AM PT]  Ubit Umarov: npc do not tp
[11:09 AM PT]  Ubit Umarov: no
[11:10 AM PT]  Ubit Umarov: the sl flag is only for client side bots, as i said and  Lyr explained
[11:10 AM PT]  Ubit Umarov: or accounts that declare to be a "bot account"
[11:10 AM PT]  Ubit Umarov: see?
[11:10 AM PT]  Orbert.Tatham Thank you for clarifying
[11:10 AM PT]  Jagga Meredith: the mantis about oars was me.  I still find the loading message a bit ambiguous but can live with it.  You need to modify mantis to allow for version 9.2.2.yeti.  it still says DEV.
[11:10 AM PT]  Lyr Lobo: In SL, you had to identify in the profile page that your account is a Bot so people would not think tha t the account was driven by a human. No such confusion for animesh
[11:11 AM PT]  Ubit Umarov: ahh yes i need to update mantis adding release and
[11:11 AM PT]  Ubit Umarov: sorry forgot .. ooops :)
[11:11 AM PT]  Jagga Meredith: np
[11:11 AM PT]  Ubit Umarov: well me or andrew :)
[11:12 AM PT]  Vincent.Sylvester Depending on what the flag is called in the viewer then it could be used for NPC stuff, that I could actually see being a useful toggle to have since that is easier than changing the ossl perms and restarting the simulator
[11:12 AM PT]  Ubit Umarov: and ofc as i told last week, i did release
[11:12 AM PT]  Vincent.Sylvester Could be neat if you want to remove NPCs you lost the key of
[11:12 AM PT]  Jagga Meredith: I've got 9.2.2 on Trondheim.  works fine.
[11:12 AM PT]  Ubit Umarov: moved the dotnet6 branch to version Dev Nessie
[11:13 AM PT]  Gavin.Hird The Viewer UI has it named as "Must not be scripted agent" on the Estate tab
[11:13 AM PT]  Ubit Umarov: at this  point master branch is on old
[11:14 AM PT]  Ubit Umarov: it has the code of the release and still says version Dev yeti
[11:14 AM PT]  Gavin.Hird grouped with payment info and 18  and such
[11:14 AM PT]  Ubit Umarov: im on that dilema of  merging dotnet6 on top of it, or keeping it for mono a bit longer
[11:15 AM PT]  Jagga Meredith: at some point we're gonna have to say "mono is dead, long live mono"
[11:15 AM PT]  Gavin.Hird yes
[11:15 AM PT]  Cuga.Rajal that will be a major merge
[11:16 AM PT]  Cuga.Rajal better to rename branches than merge?
[11:16 AM PT]  Gavin.Hird I was thinking the same
[11:16 AM PT]  Ubit Umarov: merge is rename :)
[11:16 AM PT]  Ubit Umarov: git has no rename operation
[11:17 AM PT]  Gavin.Hird rename Master to mono master or something and then dotnet6 back to Master
[11:17 AM PT]  Gavin.Hird of course it has
[11:17 AM PT]  Ubit Umarov: not is has not
[11:18 AM PT]  Ubit Umarov: google it
[11:18 AM PT]  Misterblue Waves: I thought it was "git mv"
[11:18 AM PT]  Ubit Umarov: well i had not find it
[11:18 AM PT]  Vincent.Sylvester I can rename branches in git extensions just fine, so it must have something for it
[11:18 AM PT]  Ubit Umarov: ahh mb i did type on irc im
[11:18 AM PT]  Gavin.Hird git branch -m  
[11:19 AM PT]  Ubit Umarov: well ill look
[11:19 AM PT]  Cuga.Rajal
[11:19 AM PT]  Gavin.Hird -m for mv  move to make it confusing
[11:19 AM PT]  Ubit Umarov: never the less my thing is how long i keep master as is
[11:19 AM PT]  Cuga.Rajal it might cause havok with those who are updating same branch name before renaming
[11:20 AM PT]  Vincent.Sylvester Heh paying attention does help in that case
[11:20 AM PT]  Cuga.Rajal well, maybe not
[11:20 AM PT]  Vincent.Sylvester Never blindly git pull rebase
[11:20 AM PT]  Cuga.Rajal git clone is safest
[11:21 AM PT]  Andrew Hellershanks: As long as you don't have locally uncommited changes.
[11:21 AM PT]  Cuga.Rajal if they do, they should already know about the branch naming plans :)
[11:21 AM PT]  Andrew Hellershanks: There are naming plans? ;)
[11:21 AM PT]  Vincent.Sylvester I wrote an article on maintaining forks with least amount of issues from merging a while back, patches are your friend
[11:22 AM PT]  Misterblue Waves: rebase can really mess you up... I then to get diffs and apply them separately... "git apply-patch" and "git format-patch"
[11:22 AM PT]  Misterblue Waves: then => tend
[11:22 AM PT]  Andrew Hellershanks: Doing a git stash before a pull or application of patches can also mess you up
[11:23 AM PT]  Vincent.Sylvester Best to maintain a repo of just your changes and then stuff that onto master. You can update master independently and if the changes fail to patch over you just have to fix those
[11:23 AM PT]  Misterblue Waves: could tag 'master' as "legacy-framework" or something and then merge 'dotnet6'
[11:23 AM PT]  Misterblue Waves: is that the sort of thing being discussed?
[11:24 AM PT]  Ubit Umarov: well ill do something
[11:25 AM PT]  Ubit Umarov: one day master will be on Nessie, only for dotnet >=6
[11:25 AM PT]  Andrew Hellershanks: I was expecting master to be the latest version with changes that was just released as but I see that isn't the case.
[11:25 AM PT]  Ubit Umarov: and current Net4.x/mono will be stored on anothe branch
[11:26 AM PT]  Ubit Umarov: yes but there is dotnet6 that is a breaking change
[11:26 AM PT]  Vincent.Sylvester Hopefully before that day comes dotnet6 will at least some tests setup to make sure it works
[11:26 AM PT]  Misterblue Waves: I will be checking in some small changes for BulletSim over the next few weeks... at the moment should I put them into both 'master' and 'dotnet6'?
[11:27 AM PT]  Ubit Umarov: as git is today.. yes
[11:27 AM PT]  Ubit Umarov: im me at irc, so i do notice it..
[11:28 AM PT]  Misterblue Waves: I'm running the dotnet6 branch on my region in a Docker container in a Droplet on DigitalOcean... no problems found so far
[11:28 AM PT]  Ubit Umarov: we are on it here also
[11:28 AM PT]  Gavin.Hird I converted all my 57 regions to it
[11:28 AM PT]  Ubit Umarov: as you can see above that bottle to my right
[11:28 AM PT]  Gavin.Hird runs fine
[11:28 AM PT]  Misterblue Waves: k... irc is only good when I happen to be online (traveling at the moment and my laptop is my connection)
[11:29 AM PT]  Ubit Umarov: about the release of yeti.. so far only have one issue on mantis... some problems with mysql that i can't repo
[11:29 AM PT]  Ubit Umarov: possilble because they are using mysql .5.5 ?  still not sure
[11:31 AM PT]  Vincent.Sylvester Starting to look more like that's the cause, still strange, but ultimately keeping things up to date is a good idea
[11:31 AM PT]  Ubit Umarov: the code changes where commited not long ago, but where changes i made locally long ago.. possilble when i was using ,ysql 5.x also, but can't tell
[11:31 AM PT]  Gavin.Hird I am on 8.0.23
[11:31 AM PT]  Ubit Umarov: yeah im on a 8.x also now
[11:31 AM PT]  Gavin.Hird 8.0.32
[11:32 AM PT]  Ubit Umarov: hmm did i miss any question, so far?
[11:32 AM PT]  Andrew Hellershanks: FYI, I have changed to in Mantis and added
[11:32 AM PT]  Ubit Umarov: Oh tanks andrew
[11:32 AM PT]  Ubit Umarov: and thanks also :)
[11:33 AM PT]  Andrew Hellershanks: :)
[11:34 AM PT]  Ubit Umarov: i asked last week.. but , you see any major reason to not just move all on to dotnet, stopping adding new code to net4.x/mono ?
[11:35 AM PT]  Gavin.Hird not really
[11:35 AM PT]  Vincent.Sylvester You know I'm gonna say, tests
[11:35 AM PT]  Ubit Umarov: ( it is still possible we do a Yeti for mono.. if we see a big bug )
[11:35 AM PT]  Gavin.Hird you might want to fix some bugs on the old mono version
[11:35 AM PT]  Gavin.Hird as they appear I mean
[11:35 AM PT]  Ubit Umarov: yes just typed that gavin.Hird
[11:36 AM PT]  Ubit Umarov: like that mysql thing ofr example.. lets see
[11:36 AM PT]  Ubit Umarov: wel there is a 0,9,2,2 branch also... with the released code
[11:36 AM PT]  Ubit Umarov: as our tradtion
[11:37 AM PT]  Ubit Umarov: so even is master changes.. there is that :)
[11:38 AM PT]  Ubit Umarov: breaking changes, always bad and nasty :(
[11:38 AM PT]  Vincent.Sylvester I already told Ubit that if there are any things that are potentially useful for mono to point me at the commit and I'd create a backport patch for that. I feel like dotnet6 should only go ahead if we have some basic tests for it to make sure things like login, inventory, etc. the most crucial operations are tested for to make sure code changes don't cause regressions
[11:38 AM PT]  Ubit Umarov: we did lost 32bit on intel,  Xengine, and by my decision, old ODE
[11:39 AM PT]  Misterblue Waves: by 'tests', what do you mean?
[11:39 AM PT]  Vincent.Sylvester automated tests run by Jenkins
[11:39 AM PT]  Ubit Umarov: the nunit ones
[11:40 AM PT]  Ubit Umarov: that we do need to update at least to nunit3
[11:40 AM PT]  Ubit Umarov: oe xunit
[11:40 AM PT]  Misterblue Waves: ok.... so we need to find someone who will adapt the existing tests?
[11:40 AM PT]  Ubit Umarov: or
[11:40 AM PT]  Ubit Umarov: guess simpler path is nunit3
[11:40 AM PT]  Vincent.Sylvester nunit3 is kinda out of support already and the test adapter for visual studio is kinda funky
[11:40 AM PT]  Ubit Umarov: as long we can make github and vs2022  ( at least ) run them
[11:40 AM PT]  Cuga.Rajal they will have tro be re-written for dotnet-test eventually
[11:41 AM PT]  Ubit Umarov: well that is a big thing
[11:41 AM PT]  Cuga.Rajal so better to rewrite once than rewrite twice?
[11:41 AM PT]  Ubit Umarov: and i wil not stop main devel bc of those
[11:41 AM PT]  Vincent.Sylvester We discussed that last week
[11:41 AM PT]  Andrew Hellershanks: Based on how I have been writing unit tests for a small(ish) project I'd hate to think of how many would wind up being written for OpenSim.
[11:42 AM PT]  Cuga.Rajal I guess nunit3 better if we need something sooner
[11:42 AM PT]  Gavin.Hird how many tests are we talking about?
[11:42 AM PT]  Ubit Umarov: a ton LOL
[11:42 AM PT]  Vincent.Sylvester A couple hundred
[11:42 AM PT]  Gavin.Hird ic
[11:42 AM PT]  Lyr Lobo grins
[11:42 AM PT]  Ubit Umarov: and sadly most of them, just useless
[11:42 AM PT]  Gavin.Hird but you can do it over time
[11:42 AM PT]  Andrew Hellershanks: My smallish project already has 162 tests.
[11:42 AM PT]  Gavin.Hird start with the most critical ones
[11:43 AM PT]  Ubit Umarov: almosty like   make a = b.. then right a test to see if a == b
[11:43 AM PT]  Ubit Umarov: bahh
[11:43 AM PT]  Ubit Umarov: write even ;)
[11:43 AM PT]  Andrew Hellershanks: Some tests may seem useless but can still have value. I have been testing the heck out of my functions by seeing that they handle data they should never get passed but if they ever do they will handle it properly.
[11:44 AM PT]  Vincent.Sylvester At most we have 562 tests that I managed to get somewhat working, these include performance tests though, which we don't really need
[11:44 AM PT]  Vincent.Sylvester The most critical things testing for regressions in script engine, basic operations like login, inventory and so on
[11:45 AM PT]  Andrew Hellershanks: Performance tests are still useful to catch code changes that make performance worse than before.
[11:45 AM PT]  Vincent.Sylvester Starting with just a few the rest just turn into busy work really
[11:46 AM PT]  Misterblue Waves: or we can do it the Internet way and just release it and fix the complaints that come back :)
[11:46 AM PT]  Andrew Hellershanks: :)
[11:46 AM PT]  Misterblue Waves: but you probably already had that discussion :)
[11:46 AM PT]  Gavin.Hird :-)
[11:46 AM PT]  Vincent.Sylvester The Microsoft way
[11:46 AM PT]  Orbert.Tatham We try too hard to do it the Microsoft way - we actually care
[11:47 AM PT]  Gavin.Hird where is a good AI when you need one
[11:47 AM PT]  Vincent.Sylvester For that the mantis would need to be less of a mess really
[11:47 AM PT]  Andrew Hellershanks: Which version of nunit was used for the current set of tests?
[11:47 AM PT]  Vincent.Sylvester 2
[11:47 AM PT]  Misterblue Waves: "testing" is so "waterfall" and so "yesterday"
[11:47 AM PT]  Misterblue Waves will be quiet
[11:48 AM PT]  Andrew Hellershanks: 2 dot what?
[11:48 AM PT]  Vincent.Sylvester I can recall some instances where tests failing did catch some potential issues though, just most are a bit plainly written, naively
[11:48 AM PT]  Gavin.Hird some of the code they release on the net seems like freefall
[11:48 AM PT]  Vincent.Sylvester NUnit-Console version
[11:49 AM PT]  Andrew Hellershanks: I just found some pages about updating 2.6 to 3.8
[11:49 AM PT]  Vincent.Sylvester When I asked the nunit folks about the test adapter for visual studio they said 3 was already going out of support and there would be no more test adapter, then something about nuget and I stopped listening at that point
[11:50 AM PT]  Andrew Hellershanks: One of those pages said that the switch from 2.x to 3 was a big one.
[11:51 AM PT]  Vincent.Sylvester Depending on the scale of that it might be easier to go to dotnet-tests directly given that has support going forward and is part of dotnet
[11:51 AM PT]  Vincent.Sylvester Seems logical to test with the thing that was specifically written for it
[11:51 AM PT]  Vincent.Sylvester Though I am not sure what the test adapter, if there is one, looks like for that
[11:51 AM PT]  Orbert.Tatham And it has a better chance of being supported going forward
[11:51 AM PT]  Ubit Umarov: Vincent
[11:52 AM PT]  Ubit Umarov: dotnet.tests can be NUnit, XUnit, or ms test
[11:52 AM PT]  Ubit Umarov: so not that clear to talk abotu dotnet tests
[11:53 AM PT]  Andrew Hellershanks: For the move to nunit 3 there is this page ->
[11:54 AM PT]  Vincent.Sylvester dotnet test primarily aimed at xunit from the looks of things, if you want to pedantic
[11:55 AM PT]  Ubit Umarov: and about dotnet-test
[11:55 AM PT]  Andrew Hellershanks: Bah! The page was moved to
[11:55 AM PT]  Ubit Umarov: if fact may suport others
[11:56 AM PT]  Vincent.Sylvester It wouldn't run the nunit tests we currently have, so not sure what version it will run on that, 3 or maybe 4
[11:58 AM PT]  Selby.Evans bye all
[11:58 AM PT]  Andrew Hellershanks: We are almost at the top of the hour. Does anyone have some other question(s)/comment(s) to make on something other than unit testing before we continue on that topic?
[11:58 AM PT]  Motoko.Karu cya Selby
[11:58 AM PT]  Andrew Hellershanks: ok, bye Selby. Thanks for coming.
[11:58 AM PT]  Andrew Hellershanks: You were sitting in a different chair today.
[11:59 AM PT]  Vincent.Sylvester Just having a single test written in xunit would be nice, to test how that integrates with visual studio and jenkins
[11:59 AM PT]  Cuga.Rajal if there was one well-written test, others could be cloned form it
[11:59 AM PT]  Andrew Hellershanks: Will Jenkins on the machine be able to run nunit 3 based tests?
[12:00 PM PT]  Ubit Umarov: current is dead
[12:00 PM PT]  Ubit Umarov: can only compile  with mono 5.12 ( that andrew did manage to compile) and nunit 2
[12:01 PM PT]  Ubit Umarov: recent jeckins could to any of those mentioned i guess
[12:01 PM PT]  Ubit Umarov: well it basicly runs scripts..
[12:01 PM PT]  Vincent.Sylvester I recall looking into nunit3 and couldn't really work out initially how to get the newer version on there to call it up in jenkins
[12:01 PM PT]  Andrew Hellershanks: Cuga, writing tests is also about isolating a function so it doesn't rely on the half dozen other ones it might want to call.
[12:02 PM PT]  Ubit Umarov: well making tests is a bit more complex...
[12:02 PM PT]  Cuga.Rajal ok, so not as simple as that
[12:02 PM PT]  Ubit Umarov: about high level specitication and validation of that specification
[12:02 PM PT]  Andrew Hellershanks: It comes down to whether specific functions are being tested or procedures are being tested. For my work I've been doing function specific tests (so far).
[12:02 PM PT]  Ubit Umarov: independent of the example implementation
[12:03 PM PT]  Vincent.Sylvester dotnet /home/buildserver/nunit3/bin/net6.0/nunit3-console.dll $WORKSPACE/bin/OpenSim.Region.CoreModules.Tests.dll --result=$WORKSPACE/test-results/OpenSim.Region.CoreModules.Tests.dll-Results.xml
[12:03 PM PT]  Vincent.Sylvester Course that blows up since it's not written for that
[12:03 PM PT]  Ubit Umarov: most our testes are a fail   because they only test the current implementation
[12:04 PM PT]  Ubit Umarov: ie some looked to the exact code we had and made sure it kept giving some results..
[12:04 PM PT]  Ubit Umarov: a few are ofc done from specification and what ocde should do, not how it is doing..
[12:04 PM PT]  Andrew Hellershanks: That's the other part of writing tests. You need to know what is the correct expected behaviour. If a test fails you need to know if the test is faulty, the code being tested, or both.
[12:05 PM PT]  Ubit Umarov: we had a fun script compile test
[12:05 PM PT]  Vincent.Sylvester 50/50 with the current tests on that front
[12:05 PM PT]  Ubit Umarov: we did had a string with the source code
[12:06 PM PT]  Andrew Hellershanks: One recent test I wrote failed. At first I thought the test code was faulty. I eventually realized it was revealing a flaw in the code being testd.
[12:06 PM PT]  Ubit Umarov: the test was to compare the output of the lsl -> c# compiles, with exact strinb g match to what the compile did at the time of the test creation
[12:06 PM PT]  Ubit Umarov: see how bad that is?
[12:06 PM PT]  Andrew Hellershanks: eek
[12:07 PM PT]  Ubit Umarov: as soon i made a fix to the lsl-> c# that did blow up
[12:07 PM PT]  Andrew Hellershanks: Testing compilation of code is complicated. You need to compile the code and test that the generated code does what it is supposed to do.
[12:07 PM PT]  Ubit Umarov: ofc.. even a change on a space did blow that up
[12:08 PM PT]  Ubit Umarov: yeah.. not a easy thing
[12:08 PM PT]  Ubit Umarov: but that solution was not it!!
[12:08 PM PT]  Ubit Umarov: ( ofc i killed that test ages ago.. )
[12:09 PM PT]  Ubit Umarov: so do it right, should not be jsut convert current ones to new ones..
[12:09 PM PT]  Andrew Hellershanks: good. We don't care what the generated code looks like as long as it works correctly.
[12:10 PM PT]  Vincent.Sylvester I did add a few old ones back in and adjusted things a little bit to have more testing done, normally have very few tests actually doing useful stuff. Problem is I can't really write them from scratch just yet. Modifying existing ones is much less complex
[12:10 PM PT]  Ubit Umarov: but rethink them, and make them by high level spec.. as i said
[12:10 PM PT]  Vincent.Sylvester Some are pretty abstract almost esoteric and others are... questionable
[12:10 PM PT]  Ubit Umarov: some are ofc already like that..
[12:10 PM PT]  Andrew Hellershanks: BTW, do we need to differentiate between X and Y Scripting engines in Mantis?
[12:10 PM PT]  Ubit Umarov: the permissions tests
[12:11 PM PT]  Andrew Hellershanks: oh, no. Permissions test is a can of worms.
[12:11 PM PT]  Ubit Umarov: to try to test "by spec" for example
[12:11 PM PT]  Ubit Umarov: do
[12:11 PM PT]  Vincent.Sylvester Some are "place avatar make it walk" for crossing, others test directly on the functions, which of course work, but there might be more to it
[12:12 PM PT]  Vincent.Sylvester Permissions in regards to region and estate access are a mess, needs a bit of simplification in there, too many functions doing the same thing
[12:12 PM PT]  Vincent.Sylvester Someone also messed the order up
[12:12 PM PT]  Ubit Umarov: ofc to have a proper tests do take more menhours than to only do usefull code !
[12:12 PM PT]  Ubit Umarov: man hours?
[12:14 PM PT]  Vincent.Sylvester I have no problem investing time into it, but starting from scratch on something as abstract as tests...
[12:14 PM PT]  Vincent.Sylvester ouch my brain
[12:14 PM PT]  Cuga.Rajal hard enough to fix bugs, add features, keep updated to SL specs etc
[12:14 PM PT]  Cuga.Rajal testing always comes after that
[12:15 PM PT]  Andrew Hellershanks: The book I've been reading suggests writing the tests first. In our case the code already exists so we mostly need to test what we have.
[12:15 PM PT]  Ubit Umarov: actually some tests should me made before implemetation code ( and by dif ppl )
[12:15 PM PT]  Andrew Hellershanks: The idea in the book is that the tests are part of the documention of how some functions should work.
[12:15 PM PT]  Orbert.Tatham Code the tests to specifications, then test the functional code to that
[12:15 PM PT]  Ubit Umarov: yeap
[12:16 PM PT]  Ubit Umarov: well im no expert on that..  unit testing is a "science on its own"
[12:16 PM PT]  Ubit Umarov kicks Lyr
[12:16 PM PT]  Ubit Umarov: :(
[12:16 PM PT]  Ubit Umarov: oops :)
[12:16 PM PT]  Lyr Lobo laughs
[12:16 PM PT]  Vincent.Sylvester Not sure what's harder to read, OpenSim code or LL protocol documentation
[12:17 PM PT]  Andrew Hellershanks: Having different people do the tests and the implementation can be good and bad. It is good to split the work load but means you need good specs written so that implementors and testers know the expected behaviour in the same way.
[12:17 PM PT]  Vincent.Sylvester At least some OpenSim code is written by people with a functioning brain
[12:17 PM PT]  Andrew Hellershanks: :)
[12:18 PM PT]  Orbert.Tatham In my background, getting good specs was the hardest part
[12:18 PM PT]  Misterblue Waves: or that "OpenSim code" exists and "LL protocol documentation" doesn't ?
[12:18 PM PT]  Gavin.Hird is there code written by someone with a non-functioning brain?
[12:18 PM PT]  Andrew Hellershanks: Orbert, true. Often people don't know what they want until you give it to them.
[12:18 PM PT]  Vincent.Sylvester They did write some new stuff for some new inventory protocol Ubit showed me the other day and I nearly had my brain fall out of my head
[12:18 PM PT]  Misterblue Waves: is that code written by ChatGBT?
[12:18 PM PT]  Ubit Umarov: nahh ll did doc the protocol a lot
[12:19 PM PT]  Vincent.Sylvester Doesn't mean it makes sense
[12:19 PM PT]  Ubit Umarov: ll doc is really good.. well at least was
[12:19 PM PT]  Ubit Umarov: okj need go rl for 15min
[12:19 PM PT]  Ubit Umarov: so any fast question?
[12:20 PM PT]  Andrew Hellershanks: Ubit, It is almost 20 past. Let's just wrap it up for today.
[12:20 PM PT]  Lyr Lobo: have a great week!
[12:20 PM PT]  Misterblue Waves: take care all!
[12:20 PM PT]  Ubit Umarov: ok cya later :)
[12:20 PM PT]  Orbert.Tatham Peace, guys
[12:20 PM PT]  Andrew Hellershanks: Thank you all for coming. See you again next week.
[12:20 PM PT]  Motoko.Karu take care everyone...
[12:20 PM PT]  Ubit Umarov: (afk)
[12:20 PM PT]  Gavin.Hird yeah, I should head back to reality. Have a good Easter
[12:20 PM PT]  Jamie.Jordan great meeting have a great week yall
[12:20 PM PT]  Andrew Hellershanks: Yes, I have some 3D modelling to get on to.
[12:20 PM PT]  Cuga.Rajal *throws dollar bill to MrBlue
[12:21 PM PT]  Misterblue Waves: meeting over! time to dance on the tables!
Personal tools
About This Wiki