Hi Joerg,
On Tue, 2006-11-07 at 12:40 +0100, Joerg Sievers wrote:
> > + specifications are critical for (at a minimum):
> > + file formats
> > + complex / unfamiliar behaviours
>
> + behaviour changes affecting other's work (e.g. the automated
> gui testing is extremely dependent to the basics of OOo)
hokay - well, this can easily be amortised by partitioning the tests
such that it is easy/fast to run the tests on the piece of GUI code you
changed to verify they are still perfect before marking the CWS 'Ready
for QA', and/or writing guidelines on how best to avoid breaking the GUI
test tool, and/or improving the GUI test tool & code so it is less
fragile under test :-) [ string names for key widgets eg. ]. Clearly a
situation where the regression tests are fragile over small UI changes,
require lots of maintenance and produce lots of false positives is in
no-one's interest. Presumably also getting lots of developers to run the
tests themselves & try to analyse the output may result in more interest
in improving the test framework itself.
> > + we need to be able to execute these way more
> > quickly: < ~2 hours, to get yes/no answers on
> > individual CWS' faster.
>
> On it's way, You can directly contact me for an update.
Great; that's good news indeed; thank you.
> > + Wiki
> > + using a wiki for specs allows easier spec editing
> > and construction and maintenance
>
> Don't think so, but there is now one. Try to design UI in it and you
> will love .odt :-)
As I've said before, I am certain that the process of designing a UI is
best done either in a UI Engineers head, or on some paper, or even
better with several iterative prototype models and filmed & analysed
studies of test subjects using each etc. The spec. document should not
be used as part of a workflow, but -only- to communicate relevant
information about the finished result to interested parties; hence my
desire to remove the IMHO unhelpful iTeaming aspect.
HTH,
Michael.
--
[EMAIL PROTECTED] <><, Pseudo Engineer, itinerant idiot
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
