On Oct 21, 2006, at 3:19 PM, kee nethery wrote:
Rather than callbacks, I'd probably go for things being called
from the openstack handler with results being written to a text
file with the same name as the test stack. That way you just
launch each stack and don't have to worry too much about
communication.
But isn't writing to a file a form of communication? I would
think that 'return' or 'send' are simpler, however there is always
a chance those are broken, but unlikely at the same time. Saving
to a file does sound like it could fit into a collection of
communication methods.
The advantage of a file is that you can have the master stack look
in all the files sometime after it launches all the tests (and
after they really should be finished) to see which tests suites
failed and report on those. If a test suite fails, it won't hang
all the other tests.
Couldn't a test manager write a file as easily as the test itself?
Presumably a test manager could also take the stack and save it with
some properties set before executing each test in the the stack.
That could provide similar information. Of course, one might assume
that file I/O is more primitive than a stack save.
Also, I think this ...
*****
on test
return 2+2=4
end test
*****
...is simpler and more directly tests than...
*****
on test testPath
open file testPath for write
write 2+2=4 to file testPath
close testPath
end test
******
... but the latter is not too bad.
Now about "launches all the tests". That looks interesting.
Presumably only one collection is interactive, but maybe others could
run in the background. What do you have in mind?
There was once one Revolution bug in which Rev not only crashed but
logged me out! So there might be some interaction among processes
running tests.
Dar
_______________________________________________
use-revolution mailing list
use-revolution@lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription
preferences:
http://lists.runrev.com/mailman/listinfo/use-revolution