Thinking about these benchmarks. Does it make sense to benchmark against HTML unit?
http://gwt-code-reviews.appspot.com/547801/diff/1/4 File /user/build.xml (right): http://gwt-code-reviews.appspot.com/547801/diff/1/4#newcode598 /user/build.xml:598: <property name="compile.tests.complete" value="true"/> On 2010/05/24 20:54:55, Lex wrote:
This looks too tricky. Is it here so that the benchmark targets can
depend on
various targets, and yet prevent those targets from actually building?
If so,
why not simply depend on more specifically what the targets require to
be built?
If the existing ant targets are too coarse, then they can certainly be
factored
into more specific ones.
Refactored into "-benchmark.web.xyz" tasks that do not depend on compile and compile.tests. http://gwt-code-reviews.appspot.com/547801/diff/1/4#newcode608 /user/build.xml:608: --> On 2010/05/24 20:54:55, Lex wrote:
I don't see anything running in parallel here. Drop the parallel and
sequential
tags and make it simply three antcalls ?
Done. http://gwt-code-reviews.appspot.com/547801/show -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
