Sam,
Kewl will give it a go.Instument-Manager now compiles and makes a jar even if AltRMI barfed. Clearly one could not deploy like that, but the tinderbox concept is about early warnings now making deployables. Thus Instrument-Manager's reflection-loading of AltRMi enabled versions of its core classes, is a good place to halt the cascade of "gump broken" messages. Altrmi's own gump warning should be enough.
Cool!I'm coding blind on GUMP to be honest as I only have windows here. I have two Shuttle SS40's and a Sony VAIO. None of which (yet) accept a version of Linux that will work without major fiddling.
Heck, Gump was developed on windows. But just so I can eliminate your fear of updating gump descriptors, let me talk you through this (I assume that you have a working ant?):
cvs co jakarta-gump
cd jakarta-gump
notepad project\jakarta-avalon-excalibur
[on line 631, column 10, change "depend" to "option". Save and Exit]
ant gen
cvs commit
That's it! Note that the ant gen (which takes only a minute) doesn't actually build anything, it only verifies that the descriptors are syntatically correct. After you commit, merely watch the next night's build...
What does option mean? It means that on the rare occasion that altrmi actually builds (just kidding!), it will be included in the classpath. When it doesn't, it won't. Simple enough, eh?
(this layman has visions of eight hours of CVS checkouts, an three zigawatts of gump time for it to come up with the answer "42", on a single "ant gen")
- Paul
--
To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
