Subject Re: IB_Date (testing, QA, etc.)
Author mmenaz
Replied each sentence below :)

--- In IBObjects@y..., "Jason Wharton" <jwharton@i...> wrote:
> I plan to avoid the "release in order to test" mentality. The only time I
> will release in order to test is when I specifically indicate a release as a
> test release. I assume that the majority of my customers don't want to feel
> like their efforts are doubled as my QA efforts.

Yes, you are definetly right. And consider that if you flag a release as a "test" one, usually you have to introduce significant (for the programmer) advantages to make them test it...
When I submit some improvements to you it's something I've tested in the situation I'm using it, so I count on your test too. But what I meant is that we two can test only a (small) finite number of situations, so we will never be sure about the testing.
Beyond this specific situation (TIB_Date), do you think would be good having a "tester" release and a "production worthy" one (like Linux)? A developer release tested by selected guys that want to help you, but also open for everyone "brave". As a IBO contributor, I would be much sure about modification I make too.

>
> IBO is something I ask money for and as a result I feel it is my
> responsibility to make sure my releases are production worthy upon release.

Yes, Borland is asking money for VCL too but D6 ActionManager component is not that great example of QA <g>. Nothing I've seen compares with IBO in quality and details.

> I do a significant amount of testing and take great caution to make sure
> this remains the case.

Thank you, I benefit from it too.

>
> Your changes were all supposed to be "rolled back" because of the other
> problem that other code introduced.

I really use them, so I will have to inherit from your code a specialized component of apply the changes every release you submit (I will see the best solution). That is not a big issue, but since I would like to keep being in syncro with IBO standard and have other people benefit from my work (I've spent a lot of time upon it), can we collaborate in fixing it for your customers too? Or do you think that it does nothing that worths the trouble or that can not be fixed for all the platforms you support?

Thanks
Marco Menardi