On Wed, 30 Oct 2002, Marcus Boerger wrote: > At 11:38 30.10.2002, Melvyn Sopacua wrote: > >At 22:57 29-10-2002, Marcus Börger wrote: > > > >> --use var_dump() > > > >Why? > > > >Please Marcus - discuss with the person who developed the test before > >altering it's behavior. > >What's the positive side of using var_dump? > > I saw no need in first comparing results and then compare the result of the > compare > again in run-test.php. The far better way seems to present the output and > compare the > expected output in run-test.php. The reason i did was that i was wondering > what i would > do when i got any .diff .out from a user for whom the test faild? I would > receive somthing > like "unexpected result". Wow that would be a good help in finding the bug.
Ok - good point. > > >Here are the negatives: > >-- iconv could be unsupported (which was on my list of things to fix for > >this and other tests) > > If the entity fails, it fails in both cases and so the meta test > > succeeds but, just because > > a character misses the test fails for no good reason. > > We can rewrite the test so it becomes independant from iconv. I'm looking into some stuff, to get this done in the best way. I think I'll change sablot.c to reflect if --with-iconv-dir was given and provide the information xslt_sab_libs() (same for js). The main reason, is not this test by the way - but in a hosting environment, it may also be very convenient, to know if you have iconv recoding available in Sablotron, and js support and what the expat version is. It may also be possible to get this via the Sablotron library. Anyhow - it depends what can be done in which timeframe. > >-- indenting is something being discussed on the Sablist and so may change > >and can be > > platform specific. However indenting is handled, it will be handled > > consistently for a > > platform. > > If above is reason enough to display the full result we could use > --EXPECTREGEX-- > und skip over any whitespace. But your original compare would be much easier. > Perhaps use your compare and var_dump on failure? That's the best compromise I think. -- With kind regards, Melvyn Sopacua <?php include("not_reflecting_employers_views.txt"); ?> -- PHP CVS Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php