Jump to content

After patch 313.9 the Argent is bugged


Lortam

Recommended Posts

  • Replies 75
  • Created
  • Last Reply

 

On 9/8/2020 at 7:45 AM, SuaveHyena said:

Automated testing of a 3D game at the front end isn't as common place or "easy" as you think.

 

This should have been tested and found but it's more feasible to do these tests manually.

 

I've worked with 3D games in my off time and; yes, it is entirely possible to write unit and integration tests for that type of software. Granted that full E2E testing with 3D simulation is entirely hit or miss, but I honestly couldn't give a hoot if the proper animation does not fire just so long as the argy picks something up.

If the software is designed appropriately, then core business functions can be isolated into objects that can be placed under various test conditions free of the animations themselves. Additionally, depending on the framework the team is using, it is also possible to verify that events and triggers are working as expected. It doesn't seem far fetched that the team would be capable of writing a test that makes sure the argy pick-up action is firing when the appropriate key is pressed.

 

On 9/8/2020 at 7:54 AM, LEDminer said:

Also, this bug with the argy is pretty random and supposedly the latest updates have nothing to do with the creature. They'd have to go through every nook and cranny in the game each update to make sure everything is working, and that would delay the patch.

Or at least I assume they didn't touch the argy...

 

That's the point of automated testing. It has it benefits during development, but the point is to make sure that developers don't regress functionality in completed areas of the software in future iterations. If the devs break something in a patch, then the test is going to catch it so long as someone runs it. Leaving regression testing to people simply isn't feasible.

Link to comment
Share on other sites

  • Volunteer Moderator
1 minute ago, LordMightyMax said:

 

 

I've worked with 3D games in my off time and; yes, it is entirely possible to write unit and integration tests for that type of software. Granted that full E2E testing with 3D simulation is entirely hit or miss, but I honestly couldn't give a hoot if the proper animation does not fire just so long as the argy picks something up.

If the software is designed appropriately, then core business functions can be isolated into objects that can be placed under various test conditions free of the animations themselves. Additionally, depending on the framework the team is using, it is also possible to verify that events and triggers are working as expected. It doesn't seem far fetched that the team would be capable of writing a test that makes sure the argy pick-up action is firing when the appropriate key is pressed.

 

 

That's the point of automated testing. It has it benefits during development, but the point is to make sure that developers don't regress functionality in completed areas of the software in future iterations. If the devs break something in a patch, then the test is going to catch it so long as someone runs it. Leaving regression testing to people simply isn't feasible.

Clearly you've never worked on a game like this at this scale.

What 3D games have you done regression testing on during your "free time"?

Asking for a friend.

Again.

Link to comment
Share on other sites

13 minutes ago, Joebl0w13 said:

Clearly you've never worked on a game like this at this scale.

What 3D games have you done regression testing on during your "free time"?

Asking for a friend.

Again.

You'll find my most visible work over on the Subnautica multiplayer mod, Nitrox. It is an open source project and can be reviewed on GitHub. It has been a while since I've contributed to the project, but I was a core contributor for around 8 months a few years back. 

https://github.com/SubnauticaNitrox/Nitrox

My handle was MadMaxOfYore, you can find me as a contributor under the insights tab.

This wasn't even native Unity development, and we still have unit and integration tests for components that we inject into the Subnautica runtime using IL injection. If we had access to the actual Unity project, then our testing capabilities would have been much more extensive.

My accomplishments include a complete overhaul of multiplayer session state management, the authentication protocol, and some core components of the net layer. I also developed the user login and server selection modules, as well as implemented a software based HSV filter that was capable of changing regions of any piece of player equipment to match a color that the players selecting when entering a game world. I had to remap the pixels of the equipment's mesh programmatically before sending everything to the GPU for rendering to get that working.

Outside of that, I was responsible for revamping their practices and general coding procedures so that core functionality of the mod could be vetted with a junit test suite. Our releases were mostly stable, an achievement considering that we had to hack our way into a game that wasn't designed with modders in mind. The only time things go crazy is when the original game devs pushed an update, which often results in outage for the mod itself due to the tight coupling that has to exist for the mod to function at all.

Do let me know if your friend has anymore questions. They seem very curious about my credentials.

Link to comment
Share on other sites

  • Volunteer Moderator
2 minutes ago, LordMightyMax said:

You'll find my most visible work over on the Subnautica multiplayer mod, Nitrox. It is an open source project and can be reviewed on GitHub. It has been a while since I've contributed to the project, but I was a core contributor for around 8 months a few years back. 

https://github.com/SubnauticaNitrox/Nitrox

My handle was MadMaxOfYore, you can find me as a contributor under the insights tab.

This wasn't even native Unity development, and we still have unit and integration tests for components that we inject into the Subnautica runtime using IL injection. If we had access to the actual Unity project, then our testing capabilities would have been much more extensive.

My accomplishments include a complete overhaul of multiplayer session state management, the authentication protocol, and some core components of the net layer. I also developed the user login and server selection modules, as well as implemented a software based HSV filter that was capable of changing regions of any piece of player equipment to match a color that the players selecting when entering a game world.

Outside of that, I was responsible for revamping their practices and general coding procedures so that core functionality of the mod could be vetted with a junit test suite. Our releases were mostly stable, an achievement considering that we had to hack our way into a game that wasn't designed with modders in mind. The only time things go crazy is when the original game devs pushed an update, which often results in outage for the mod itself due to the tight coupling that has to exist for the mod to function at all.

Do let me know if your friend has anymore questions. They seem very curious about my credentials.

So you did some regression testing on a mod. Not the whole of SubNautica  correct?

Surely you know the difference between the two.

Link to comment
Share on other sites

  • Volunteer Moderator

Program

Just now, LordMightyMax said:

As an aside; I find it hilarious that the spellcheck for this forum would think that the word "programmatically" was misspelled.

Coincidence, or telling? You decide. 

The forum doesn't have a spell check. That's on your browser.

 

Link to comment
Share on other sites

1 minute ago, Joebl0w13 said:

So you did some regression testing on a mod. Not the whole of SubNautica  correct?

Surely you know the difference between the two.

The fact that you don't know why this is a silly statement proves that you're asking questions that are out of your depth.

We had to develop within the Unity framework just like the original Subantica devs do. We just didn't do it with the benefit of the Unity editor. All of our components and features that we added to the game are built within the same constraints that they have. The fact is that it is fundamentally possible. It is a decision of the developers on the project not to practice that discipline. 

Link to comment
Share on other sites

  • Volunteer Moderator
1 minute ago, LordMightyMax said:

The fact that you don't know why this is a silly statement proves that you're asking questions that are out of your depth.

We had to develop within the Unity framework just like the original Subantica devs do. We just didn't do it with the benefit of the Unity editor. All of our components and features that we added to the game are built within the same constraints that they have. The fact is that it is fundamentally possible. It is a decision of the developers on the project not to practice that discipline. 

That doesn't mean you are testing the whole of the game, just the small part your mod plays in it. 

Link to comment
Share on other sites

1 minute ago, Joebl0w13 said:

That doesn't mean you are testing the whole of the game, just the small part your mod plays in it. 

Why does the volume matter? Yes, it would be more test code. That is a given. Omitting tests gets features out faster, but it doesn't matter how fast features are flying out the door when you're constantly dealing with production issues after-the-fact. Those issues have an impact on your bottom line. You act as-if the problems that ARK has had does not negatively effected it's popularity. The exact opposite is true.

The only people sticking around are the die-hard fans. I am one of them. Server populations are dwindling and people are leaving the game because of rampant bugs, constant instability from irresponsible development practices, and a framework that has been so badly cracked that people are able to cheat at-will. 

The fact of the matter is that WC and their developers have prioritized speed of delivery over stability. It is a cultural problem, not a technical one. I've been doing this long enough to know that shops that have a blase attitude towards automated testing are the ones that are not willing to go the extra mile to ensure that the job is done right. It is a question of work ethic, pride, and a commitment to excellence; not one of practicality.

If the software that manages the operations of a NICU can be broken down and validated with 100% test coverage, then this game can certainly be developed with coverage in the 60%-70% range. It is not nearly as complex as you're making it out to be.

Link to comment
Share on other sites

  • Volunteer Moderator
Just now, Noa said:

they are supposed to solve problems with bugs

Not at all.

They are there to offer support to the customer caused by bugs in the software. Just like any other support team does for any company.

And help with PEBKAC and 8th layer issues as well.

Link to comment
Share on other sites

3 minutes ago, Noa said:

Dont know, they are supposed to solve problems with bugs that  prevent  customers to enjoy game that development produced maybe?

The support staff are there to help you out. They don't participate in the development of the product.

The role I think you mean is that of the QA engineer. I'd be interested to know how WC actually uses them if they even have any employed. Are they really just button clickers?

Link to comment
Share on other sites

Just now, Joebl0w13 said:

Not at all.

They are there to offer support to the customer caused by bugs in the software. Just like any other support team does for any company.

And help with PEBKAC and 8th layer issues as well.

They are supposed to offer support to the customers caused by bugs in the software - that i said too.

Problem is that unsolved / fixed old and new produced bugs are too many. So support cant help customers in many cases because cases are too many.

Even if they decide to deal with customer and his problem they give a date month or later  to deal with and in that time everything is despawned.

So if development would be less productive with a bugs maybe support could have more time  to help in time.

Instead of that, they are unable to find  uploaded avatars, dinos as part of code that so8uld save uploaded tings probably never existed.

IM asking myself why we have lag every 15 min (like server is saving) if it never save anything?

So support need to use "stipulation of compensation" and "internal policy" or other way to not help.

 

Link to comment
Share on other sites

5 hours ago, Noa said:

Argentavis work for me now, im surprised that this thread is stil going.

About automatic testing,

Any idea how to test why WC didnt solve avatar losing upon server transfer and why upload is not saved.

Maybe to test also why support guys get paid.

Weird. When did the Argentavis fix roll out? I saw no update on my Ark and it still wasn't working for me yesterday.

Link to comment
Share on other sites

7 hours ago, LordMightyMax said:

 

I've worked with 3D games in my off time and; yes, it is entirely possible to write unit and integration tests for that type of software. Granted that full E2E testing with 3D simulation is entirely hit or miss, but I honestly couldn't give a hoot if the proper animation does not fire just so long as the argy picks something up.

If the software is designed appropriately, then core business functions can be isolated into objects that can be placed under various test conditions free of the animations themselves. Additionally, depending on the framework the team is using, it is also possible to verify that events and triggers are working as expected. It doesn't seem far fetched that the team would be capable of writing a test that makes sure the argy pick-up action is firing when the appropriate key is pressed.

Worked with them professionally in the context of QA.

First of all, you're assuming the software is "designed appropriately". Devs themselves admit it is messy.
Secondly, you don't know whether you'd even find the issue by throwing out the animation. The animation blueprint may actually be a core part of the cause even if the "argy pick up action fires". The function consists of a sequence of triggers, any which one may be preventing execution of the actual pick up code. Especially if the animation firing is a prerequisite before it even gets to the actual logic of connecting joints.

Thirdly, you don't throw out E2E testing just because you have something covered on unit, component or (system)integration level. So you may not care whether an animation fires, the end user will. Sounds like a poor mentality for someone who supposedly values QA.

 

Link to comment
Share on other sites

On 9/8/2020 at 2:54 PM, LEDminer said:

Also, this bug with the argy is pretty random and supposedly the latest updates have nothing to do with the creature. They'd have to go through every nook and cranny in the game each update to make sure everything is working, and that would delay the patch.

Or at least I assume they didn't touch the argy...

That's why they should have a regression set of tests. In games where you have many characters/classes, the same way we have dinos, it's risky and challenging to just "check the core functionality of the feature that was affected by the update" so they tend to have a set of functional tests to verify each character still behaves the way it's supposed to, has access to all of its abilities and that these abilities are displayed correctly when executed. It's some of the most boring test execution you can do.

And since it's a 3D game, we're not really at a stage where we can comfortably automate this. The RoI just isn't there since both development and maintenance costs will be huge compared to the return. It's cheaper and faster to have people check it manually. A human eyes (combined with brains) will always see more than a machine (when it comes to entertainment).

Link to comment
Share on other sites

Guys, may I kindly ask you to move the conversation about game testing into a respective section on the forum? I had to go through 22 posts, following my last post that I made, just to find out in the end that noone actually provided any useful data about the current situation with argies and the supposed fix for it, which (if I understand correctly) is still in progress. With all due respect, you got a little sidetracked. Please, refrain from further discussion about game testing in this particular topic.

 

Thank you in advance

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...