On Thu, Feb 15, 2001 at 05:05:47PM -0500, [EMAIL PROTECTED] wrote:
> On Thu, Feb 15, 2001 at 01:38:22PM -0500, barries wrote:
> > What do folks think of adding something like the following to Test.pm:
> <snicker-snatch>
> >
> > This would make for very succinct easy to maintain test suites, if your
> > test suite is simple enough:
> >
> > use Test qw( do_tests ) ;
> >
> > do_all_tests(
> > frobnitz => sub {
> > ok( 1 ) ;
> > },
> > skip => sub {
> > ok( 1 ) ;
> > },
> > todo => sub {
> > ok( 1 ) ;
> > },
> > ) ;
>
> What does this gain you over the more flexible and easier to learn and write:
It's no more difficult to learn except for having to deal with closures.
Which is a pain when dealing with tests that need to pass state from one
to the next (say to check multiple results, or in situations where you
need to sample a starting condition that figures in to the expected
results some how).
Indeed, parts are simpler: both the current skip() and the proposed
todo() are a little misguided: they do not skip the test, that's up to
you.
To take your example, it would need to be more complex than you seem to
think, often a fair amount, to get equivalent behavior, especially in
situations where test are non-trivial and must not actually be run when
being skip()ed or todo()ed:
noplan ;
>
### incorrect:> ok(frobnitz() eq 'HGLAG', 'frobnitz ok');
ok( frobnitz() eq "HGLAG", 1, "frobnitz" );
>
if ( $i_dont_feel_like_it ) {
### incorrect:> skip("I don't feel like it");
skip( 1, 0, 0 ) ;
}
else {
ok( blather(), "blather result" ) ;
}
# ok( not_ready_yet(), "planned result" ) ;
> todo("I'll do it later"); # This line might work, but disagrees with
# the current behavior of ok() for "todo"
# tests, which whines on a true arg.
> See Pod::Tests and pod2test
heh, I think you forget that I prototyped a bunch of very similar
features months ago and concocted the =alsofor syntax ;-).
> Its advantage (aside from existing) is that you don't slow compilation
> with lots of extra test code that Perl has to compile on every run.
You seemed to have missed the -nobloat option on the filter w/ subs
proposal, FWIW.
I think the filtering approach combined with the Pod::Tests is cool,
though. Running a module that does a
use selftest ;
like so:
perl lib/Foo/Bar.pm
could ignore all normal code lines and run the tests, while normal use
like "use Foo::Bar" would leave it untouched, not even slowing
compilation other than to load the filter and execute a few quite perl
ops to check caller()'s result (the filter would not install itself).
Best of both worlds, perhaps. I'll prototype that soon, tuits
permitting.
In other news, a simple test.pl could crawl lib and test all modules
there.
I've been using a scheme much like that for 5 months or so now on a
project and it makes maintaining the test suite with the code very
conventient. I also let test.pl take a command line parameter to test
named files using files or package name syntax:
test.pl lib/Foo/Bar.pm
test.pl Foo::Bar # tests lib/Foo/Bar.pm and lib/Foo/Bar/**.pm
I just map an editor key to the command line
./test.pl %
where the % is replaced by the filename of the current buffer. That's
about as convenient as you can get: one click testing. Hey, I should
patent that!
Anyway, with a filter, test.pl might be unnecessary I think.
perl -w %
would do the trick. I'll try to scrape a prototype together for this
soon.
- Barrie
