Le 13 avr. 05, à 16:47, Daniel Fagerstrom a écrit :
...There are certainly things left to do in the refactoring but everything is supposed to work as it is right now...
Before voting, what worries me slightly is the "supposed to work" bit.
I didn't look into the new stuff yet, are there tests (automated or not) to help make a more precise assessment?
Of course :) http://svn.apache.org/viewcvs.cgi/cocoon/blocks/unsupported/template/trunk/test/org/apache/cocoon/
I wouldn't dare trying to refactor such a complicated beast as JXTG whithout automated testing. Two of the testcases that use the expression #{$cocoon/request/protocol} fails, but they fails for the original JXTG as well, this has to do, IIRC, with Sylvain's and Carsten's refactoring of environment handling in flow.
With supposed to work, I meant that its impossible to grant that we didn't introduce bugs in functionality we didn't thought of testing. Also its quite some while since I tested it outside the testing framework, maybe Leszek has more recent experiences, of running it out in the wild.
Although I love automatic testing and wouldn't develop any Java code without it, it is still limited by ones creativity in thinking out test cases. The only "certain" test method is to put it into real use whith more users, which we ask for being allowed to do now.
We are of courses dedicated to fix or help fixing any problem that might occur.
/Daniel
