On 2002.02.19 01:06:04 -0500 Dmitri Colebatch wrote: > > > It sounds very attractive in some ways, because we can look quickly > at > > > our changes. otoh, we would also need to have a policy on > > > when we put new versions in cvs.... I dont know - it sounds good to > me, > > > as a test.... what do other ppl think? > > > > + It's way better than nothing (probably) > > + It's simple > > - it breaks on many many irrelevant changes > > > > Can we run diff from ant to get a report? > > there is an ant cvs task - I've never used it, but I'm guessing we could > use diff from there. > > > What I have been thinking of for a testsuite base involves > > > > an xml comparator (insensitive to formatting, just sensitive to > content). > > I wrote one of these for a different testing purpose, I think it would > be > > easy to use here. > > > > a class examiner that compares the properties of 2 classes to make sure > > they are the same. > > jenesis (http://www.inxar.org/jenesis/) would probably come in handy for > that. I suppose our XJavadoc engine might well also help > (o: > > > These would be somewhat more complex than running diff on text files, > > however I think it might be worthwhile to avoid sensitivity to > formatting > > changes. > > yeah - the main problem being that at some stage someone would need to > sit down and do a sizable chunk of up-front work. how about, > use jenesis to parse the cvs version of an generated file, then parse the > new generated file, and do a content comparison - that > would avoid the whitespace problem, and avoid someone having to do the > upfront programming, and the maintainence. on the bad side, > I imagine it'd be dog slow as a test. so have that for source code, your > xml comparator for xml output, and we should have a pretty > solid "change observer" > > thoughts? > I was thinking more in terms of putting the binary class files from compiling the "known good" generated source in cvs, and doing "comparative introspection" on them. I haven't tried it, but I imagine it would be pretty easy - get all the reflective stuff from each class and compare.
I guess it would be better to put "known good" java source in cvs, and compile both "known good" and "newly generated" each time to compare. david jencks > cheesr > dim > > > _______________________________________________ > Xdoclet-devel mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/xdoclet-devel > > _______________________________________________ Xdoclet-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/xdoclet-devel
