> Thanks, I'll update and try again.. Don't think I don't appreciate the
> hard
> work you are all doing, because I do.. I just want to see the quality
and
> confidence go up, which will ensure that the project will be more
often
> considered for use in various shops. For instance, I co-founded a
Java-

I don't complain. Bugs are part of my life :-)
I know XDoclet has the potential to become as popular as Ant. And we're
just getting started, we have great plans for v1.2.

> One method of doing this may be to use expect or do file comparisons.
That

Not a good idea IMO. A simple change in one of the sources and you have
to update ALL comparison targets. For example let's say Dmitri adds
support for something in jBoss, he then updates 4 source files, now we
might have to change 4*6 files and update all expected test results,
even a simple javadoc change (we copy javadoc to generated remote
interface/home/etc). It's too much volatile. I think a better approach
is to actually deploy it and *call* all those methods. So if something
goes wrong (create methods disappears, etc) we'll get a compile time
error. It's still a bit hard to test struts/etc form, should be accessed
through web. So far we all relied on our personal tests, I test it on
WebSphere on my project, Dim on jBoss, Aslak on WL, and so on. Anyway
this method is good for generated code (hard to test TX_REQUIRED is
indeed set for method x in generated xml), but for generated xml we
check against DTD, but we could add an XPath test to make sure it there
in generated xml file.
We haven't reached the level to test the above scenarios, I think
because we planned to reorganize samples to become smaller and more
testable but we couldn't find enough time for v1.1 to do that.

You know it's hard or unwilling to add tests to a huge already existing
source tree :-)

> Another method was one that was mentioned earlier, one of testing
> deployment. This would involve deploying generated beans into JBoss
and
> Weblogic, and running unit tests against them to check logic, CMP/BMP,
> etc.

Yup.

> These would involve more work but ensure that deployment occurs
properly
> (no
> exceptions upon deployment) and that the behavior is expected (data in
DB
> by
> testng with a straight SQL connection outside the bean or requerying
the
> bean and comparing fields for CMP and Castor JDO). My main concern on
this
> topic is that if you or others are only testing with one application
> server,
> issues could arise with another and not be caught. Its fairly
inexpensive
> (time wise) to download an eval of WebLogic or talk with them about a
2-6
> month dev license for testing your project as you do with JBoss. And
of

No fear of licenses, but time is money and we're all busy ppl :-)
I would do that for sure if I was paid for. I can't tell my girlfriend
to deadlock till I test it on WL too ;-)

> Sorry for the long-windedness, I just wanted to ensure that I convey
my
> thoughts adequately and provide a good background and vision for what
I
> would expect to see come from XDoclet. If this doesn't align with your
> goals, please let me know so that I may adjust my vision and
technology
> decisions.

Appreciate.

Ara.


_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com


_______________________________________________
Xdoclet-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/xdoclet-user

Reply via email to