Re: [odf-adoption] ODF Adoption TC Call - 17 July 2007 - Agenda

[robert_weir@us.ibm.com]

I think I missed the start of this thread,
but here's my thoughts:

We need what I call atomic, single feature
tests, that drill down and in an easy-to-interpret way give a pass/fail
indication of how any application deals with a particular part of the standard.
 And we need 1,000's of these.  This is just good basic testing,
akin to a unit test.

And we also need something more like
a system test, where we attempt real-world scenarios that have greater
complexity and use a number of features in typical patterns of use.

I don't think we can avoid doing it
both ways.  The atomic tests alone will not find all problems, since
some application problems will only show themselves via interactions with
other features.  But the atomic tests can give us very broad and detailed
test coverage, and that is important as well.

When I was proposing subjects for the
interoperability workshop I was thinking of focusing more on trying to
uncover and resolve problems by using some complex, real-world test cases.
 But I'm certainly open to having a portion of the event be dedicated
to discussion of the atomic tests.

What do you think?

-Rob

___________________________

Rob Weir
Software Architect
Workplace, Portal and Collaboration Software
IBM Software Group

email: robert_weir@us.ibm.com
phone: 1-978-399-7122
blog: http://www.robweir.com/blog/

Thomas Zander <zander@kde.org> wrote on 07/17/2007
04:44:39 PM:

> On Sunday 15 July 2007 15:14:36 Charles-H. Schulz wrote:
> > I should point out that the test suite is by no means a complete
tool
> > and that development efforts have stopped a while ago on this
project.
> 
> Naturally, nobody was saying otherwise :)
> I have to ask, Is that a reason to not use the results already obtained
in 
> the test suite?  The spec is still the same as when the tests
were 
> written.
> 
> > Let's pay attention to the documents and the results we are going
to
> > get with this...
> 
> There are quite a lot of places where the testsuite points out problems
in 
> loading of a example document; I found a really funny one where 
> openoffice reads the following line;
>  <style:paragraph-properties fo:line-height="2000"/>
> as being 0.79 inch.
> 
> I have no idea how it reasons that 2000 equals that value; other 
> implementations say that 2000 is in postscript points and a huge value
> (some 705 mm).  Using different units has different effects.
> When I create a document that loads so majorly different in two 
> applications, I think that is a good interoperability bug for this
> conference. Anyone agree?
> 
> These little problems are thus really easy to find and solve by going
> through the testsuite and finding the problems and fixing them in
the 
> office applications.
> 
> Can you guys give me a good reason why for this conference we should
> invent new test-documents instead of reusing the ones already created?
> 
> Bottom line is; I naturally don't object to creating new test documents
at 
> all. I just want to see OOo actually fix bugs. After all; the testsuite
> has been there for over a year and we found only 1 bug that has been
> fixed in OOo's ODF-loading code in that time.
> If this is due to problems with the testsuite; lets tackle *that*
problem 
> instead of duplicating the effort.
> -- 
> Thomas Zander
> [attachment "atthtvhp.dat" deleted by Robert Weir/Cambridge/IBM]