I urge the Blizzard Classic team to review their code and data.
Since the latest issue with raid xp, mob dungeon difficulty and layering abuse, it is hard to trust in your reference data.
How can such elemental concepts of the game be so wrong at release? It makes your reference look not believable anymore.
We need you to look into these and other issues and probably once resolved a fresh start, with proper data, code and without layering (this release has proven what impact it has in terms of exploits).
This doesn’t make sense at all. They’re able to tell it was broken because they can look at Classic and compare it to the reference client to notice the discrepancy.
You do not need that to know that the xp gained should not be the same and they had the reference client to compare prior to release. It is not their comparison that showed there is something wrong but people who pointed this very obvious flaw out.
So you are saying they did not compare xp gain when testing their implementation? That doesn’t make sense as its such a fundamental part of the game and easy to automate.
Then what other not that important issues must they have missed? This would be even worse…
And no where did it say they confirmed this by comparing with that particular reference.
They could come to the conculsion even by comparing to the retail reference as this is the same for all versions of the game. (more reason they should have catched it)
If you know anything about test automation…seriously.
Okay? So, as they said, it was a bug that it wasn’t working as it was intended. Which means they didn’t check.
As I said… it was oversight.
Do you think all software is the same? There’s no universal way to test things.
Even so, this issue slipped past. Do you think that’s because Blizzard does’t know how to set up automatic testing, or do you think it’s because automatic testing wouldn’t have identified this issue?
Automated testing would have catched the issue if setup properly against requirment in case the reference is correct.
The software does not matter in that case. Its a simple algorithm with variables, input and output. There is no difficulty writing an automated test for this. Can be unit or end to end depending on their logs and actual implementation. You would definitly not run all that manually all the time.
I’m thinking you don’t know a lot about software development.
Okay, so who sets up the requirements, exactly? Use your head for a moment. It can check the client to its reference automatically, but what it’s checking must be set up in the first place. Even if automatic testing would’ve identified this issue, it’s entirely possible that what needed to be checked wasn’t being checked.
In otherwords… oversight.
Why do you think they have a beta? They can’t just run automatictesting.exe and fix the game completely.
You seem really naive making such claims, especially with the rest of your comment.
You need requirements for any type of testing. Manual or automated.
Automation has nothing to do with this. Not sure what you want to test against or report if you have no clear requirements (other than “weak” common sense) but please enlighten me.
The beta is not replacing automated testing. It is also not ensuring the software is working as should be. It is naive to believe a beta is enough for a release.
Well, that is automation…
And you think missing a test case in such fundamental level is a good thing or somehow an excuse?
I don’t think you understand the point I’m making.
Do you think WoW made itself? Do you think whatever automatic testing they use is written automatically, as well? No.
People made it. People make mistakes. Whatever was supposed to be checked automatically was either not put in, or couldn’t be put in.
I didn’t say that.
I never said there wasn’t automation. But what is being automatically checked must be manually determined.
Oh, no, of course not. It’s a pretty big oversight, but the point of your OP is to tell Blizzard their reference client is somehow discredited because of this. It isn’t.
The validity of the reference client isn’t in question, because this issue wasn’t caused by the reference client being inaccurate.
If you again mean requirements to be determined, not if you/they claim to have a/the reference. Those become the requirement then. Everything the reference is/does has to be replicated in the their implementation. You still need to write test cases but the requirements are given and if not met then either the reference or their implementation is wrong.
I agree that there is two options, either their QA team has made a big mistake or their reference is flawed. Not really sure which is “better”.
Conclusion remains in both cases though, to not simply trust in “we have the reference”.
