On Mon, 24 Jan 2005, Peter W. Draper wrote:
> > And/or we could institute a policy that a normal test only does things
> > which can work without requiring network connections, or maybe a system
> > property along the lines of java.awt.headless which test programs ought
> > to check before attempting something networky.
>
> > Opinions?
>
> Yes, let's do that. I've added a note to AntDevelopmentSystem requiring it
> for the test target.
I've modified this note to reflect the current situation vis-a-vis
headless builds:
--
< in the absence of a network connection and display console.
--
> in the absence of a network connection. If the system property
> =java.awt.headless= is set it should run without error in the absence
> of a display console.
--
If you feel strongly then change it back to your first version and
we can do like you said, I'm not fussed (but note the first thing
that fails with no java.awt.headless but no DISPLAY is util.XMLUtilTest).
I'll get going on fixing the tests to work netless.
Mark
--
Mark Taylor Starlink Programmer Physics, Bristol University, UK
[log in to unmask] +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/
|