On Tue, May 27, 2008 at 10:51 PM, Matthieu Riou <[EMAIL PROTECTED]> wrote:
> On Tue, May 27, 2008 at 2:41 PM, Ciaran <[EMAIL PROTECTED]> wrote: > > > Hi All, > > Am I doing something badly wrong, the ODE build times are prohibitively > bad > > for me at the minute, I'm doing: > > > > 'rake TEST=no package' > > > > And this can take ~45 minutes to complete! Sometimes I can get away > with > > making a small change and re-running the package command and its quite > > quick > > ~10 minutes, but most of the time need to rake clean, then rake package. > > > > My CPU is fully max'd out on the ruby process (not the javac process!). > Is > > this normal? (Never really used rake/buildr before) > > > > I often have problems with Javadoc, doing a full package builds the distro > that includes it and for some reason that takes forever. So usually I > either > do: > > rake build TEST=no (when I don't need wars/jars) > cd axis2-war; rake package TEST=no (when I need to package, it only builds > what's necessary for the war) > rake ode:package TEST=no (when I want all packages but not the distro) > > That should speed up your build considerably (slightly more than a mn on my > machine for a clean build). Thanks that should help a *lot* I've now written a patch for the document() support we discussed earlier, its a rough draft but should work, if I'm honest I'd prefer to refactor some stuff rather than what I've done (which is block copy some code from the BPELCompiler into the runtime resolver, and added a processURI to the OProcess object, without attaching the compiler-context to the OProcess I can't see what else to do, to do it neatly) but I don't know the code-base well enough to do that, I'll raise a bug and attach a patch for comment shortly, but it works for my test-case here.. I've also tidied up the existing resolution code a bit to stop some null-pointer possibilities :) Take Care - Ciaran
