Ranger 1
Ranger Corp
572
   |
Posted - 2011.12.21 22:30:00 -
[1] - Quote
Ai Shun wrote:Hung TuLo wrote:I do understand your point.
Depending on the severity and criticalness of the problem. You don't even thing of placing something into production. And yes you do have those meetings between QA and dev teams.
In the end QA sometimes does not have the "power or control" to sto something from going into production.
It mainly is the 80/20 law or the 90/10 law.
As long as 80% - 90% of the code is working as expected then a push to production is waranted. Once again that depends on the critical nature of the problem. In your example of the UI problem. If the UI was found to be working 95% of the time and the problem was minor. Should the entire push be stopped?
Most of the time. The answer to these things are not your regular QA folks or you dev folks. Its their managers that are answering to the CEO or other bigwigs that will make the ultimate decision. By and large the QA staff is doing their job.
Remember QA just reports the facts. The test the software and report problems, possible causes and possible ramifications.
QA does not cause the coding problems they find them.
Its those in the nosebleed seats that make the decisions. If you can, read that book. It is mainly for developers, but it changed my mind about how the process should work. I know we can't always deal with ideal scenarios, but a lot of the up-front should come from development; not QA. Seriously. I strongly recommend that book.
I think what it basically boils down to is that if a release was held up until ALL bugs (regardless of severity) were squshed, it would never be released... especially when discussing software as unique and intricate as EVE is.
Much of the code in EVE is very unique... completely uncharted ground even for companies that have tried for years to come up with something similar, and failed.
Given the persistent nature of the environment and the countless variations of end user interactions with the software, plus the fact that (although they have made strides in this direction) there really is no effective way to stress test their code under normal use conditions prior to release, you end up with a lot of patches after the fact. It is a reality of the situation, not a failure in processes.
You can have most other types of software completely isolated on test platforms (word processors, music players, cooking programs, most other multi player games, what have you), fully test all possible uses with the maximum number of simultaneous users possible, and still have a good chance of missing something.
You can't do that with EVE.
Many of the things that make EVE unique and special to us have this inevitable downside. I think they've done a pretty fair job of minimizing that downside and implementing needed bug fixes in a timely manner.
Perfect code is a worthy goal, just (in this case) not a very realistic expectation.
Revenge should not stop at the ship!
It's not so much a mission statement,-áit's more like a family motto. |