[DiscordArchive] Shouldn't a mock player be enough for that kind of testing?
[DiscordArchive] Shouldn't a mock player be enough for that kind of testing?
Archived author: Kaev • Posted: 2023-09-04T12:51:44.697000+00:00
Original source
Shouldn't a mock player be enough for that kind of testing?
Archived author: Kaev • Posted: 2023-09-04T12:52:08.635000+00:00
Original source
assert received and sent packets etc
Archived author: stoneharry • Posted: 2023-09-04T12:53:05.849000+00:00
Original source
It's really hard to unit test unless you write your software with unit testing in mind
Archived author: stoneharry • Posted: 2023-09-04T12:53:30.105000+00:00
Original source
Static calls, singletons, and other patterns that can't easily be mocked
Archived author: Kaev • Posted: 2023-09-04T12:53:32.825000+00:00
Original source
yeah, TDD definetly helps with that
Archived author: stoneharry • Posted: 2023-09-04T12:54:47.121000+00:00
Original source
I honestly have mixed feelings about the usefulness of unit tests. Certainly it's great when you want to test all the permutations of a method, like calculating a time from another time. But I feel server tests are much better at catching functional issues even if they are more expensive to run
Archived author: Eatos_AE • Posted: 2023-09-04T13:13:37.689000+00:00
Original source
There is also one "unpopular" testing method that companies invoke..Even though it is not called testing per say, it serves the purpose of that.
Alpha/Beta sign-ups. You get with a little bit of luck massive amounts of testers that pays you to do the testing, instead other way around.
<@160672495266037760> On some MOBA games we successfuly implemented bot testing and used them for either to find environment issues or performance issues.
Such as you can fall through the ground at specific angles incase if navmesh hasn't been properly generated, or there are missing invisible walls.
For performances usually tracking fps/memory/load times and on some different os/spec configuration.
Now it always comes down which software to use for testing, like <@271350028985958408> said do you design whole architecture with testing in mind or not.
If not you have implications later including it and decoupling overall code but there are other neat ways we managed to go around yet to serve the purpose.
Using of external visual tools such as OpenCV where we would write a simple movement code for instance and it goes takes ss and compares with needed result to be in that time.
Another usage we did some success with is using https://www.autoitscript.com/site/autoit...downloads/
It was a bit easier to teach QA's how to use it , unlike how to do overall scripting in UE.
Later we integrate QA from the start of the project, and made a list of needed tools for testing and spend 6 months in their education in UE so they can do at least cheat methods for faster progression.
[Embed: AutoIt Script Editor Downloads - AutoIt]
Overview This is the main download page for the AutoIt Script Editor and related files. Current Versions File Date updated Notes SciTE4AutoIt3.exe (5439Kb) 16-3-2021 Installer containing SciTE and all configuration files plus utilities.Update History. Definition files included: AutoIt v3.3.14.5 and BETA v3.3.15.3 For those who want to run their ...
https://www.autoitscript.com/site/autoit...downloads/
Archived author: Eatos_AE • Posted: 2023-09-04T13:13:39.509000+00:00
Original source
In Mobile: most of the problems comes from android as there are tons of devices that you need to support, and you don't have them all :/
So best bet is try to integrate testing capabilities of testing farms such as Firebase Test Lab.
And with Firebase Test Lab it really does help a lot to get results if you integrate testing properly.
Like we had a game atm it is not yet globally released but it is in selected countries: We needed to test 10 language integrations for whole game.
Whole game to pass to takes hell a lot of time on single device, single aspect combo. We utilies Firebase Test Lab and hand simultaneously game run on all 10 languages with different aspect categories wide, ultra wide, squared and ofc with some extra speed.. The output value we received was outmatched the value we spent developing a "Auto play" like pluggable feature to pass on, plus using test farms gave you ability to do other things while you wait for the results. ( This saddly people didnt like as they couldn't slack.. but it is what it is )
And now with each project on mobile, we think at start with this auto play feature to be developed. As with CI/CD system you can schedule at night or after working hours, tests/bots to do testing on the farms and when you comeback in the morning on the work, to check for some new issues.