Re: [PATCH t/op/misc.t] cleanup

2001-09-02 Thread Jarkko Hietaniemi

 @@ -739,3 +783,9 @@
  # keep this last - doesn't seem to work otherwise?

This requirement magically went away while I was away?

  eval a.b.c.d.e.f;sub
  EXPECT
 +
 + perlbug ID 20010831.001
 +($a, b) = (1, 2);
 +EXPECT
 +Can't modify constant item in list assignment at - line 1, near );
 +Execution of - aborted due to compilation errors.

-- 
$jhi++; # http://www.iki.fi/jhi/
# There is this special biologist word we use for 'stable'.
# It is 'dead'. -- Jack Cohen



Re: [PATCH t/op/misc.t] cleanup

2001-09-02 Thread Jarkko Hietaniemi

I think the wholesale renaming of t/op/misc as t/run/kill_perl is
really wrong.

(I think you are reading too much into the leading comments, and other
people have been reading too little into them.)

t/op/misc has NOT consistently been the place for core-dumping tests;
it has some yes, but not all, and some of its tests are _not_
core-dumping, they are, ta-dah, misc, tests that over the years have
had no other appropriate place.

Also, I do not see any particular advantage in having a separate 
test script *just* for core-dumping tests: the test harnesses
(TEST and harness) should handle core dumps (as well as they can)
in any test script.

What should be done instead of the simple renaming is that each of the
tests in the run/kill_perl aka op/misc should be siphoned off to
other, as appropriate as possible, test scripts (possibly creating new
test scripts if appropriate), and leave just the bare minimum set of
tests that really doesn't fit elsewhere, into t/something/misc.

-- 
$jhi++; # http://www.iki.fi/jhi/
# There is this special biologist word we use for 'stable'.
# It is 'dead'. -- Jack Cohen



Re: [PATCH t/op/misc.t] cleanup

2001-09-02 Thread Michael G Schwern

On Sun, Sep 02, 2001 at 05:05:43PM +0300, Jarkko Hietaniemi wrote:
 I think the wholesale renaming of t/op/misc as t/run/kill_perl is
 really wrong.
 
 (I think you are reading too much into the leading comments, and other
 people have been reading too little into them.)
 
 t/op/misc has NOT consistently been the place for core-dumping tests;
 it has some yes, but not all, and some of its tests are _not_
 core-dumping, they are, ta-dah, misc, tests that over the years have
 had no other appropriate place.

Here's the Master Plan for t/op/misc.t, and I think it'll jive with
what you're thinking, just in a different order.

First order of business was to figure out what the hell misc.t does,
clean it up and explain it a little better.  That's done.

Then we want to STOP people from using t/op/misc.t as a dumping
ground.  Thus, the rename and better instructions.  It's obviously
*not* intended for miscellaneous tests, it's for tests that kill the
interpretor.  The bad name as lead to it being used as a dumping
ground.

With the dumping stopped, THEN we can start doing the archeological
digging necessary to figure out what in misc.t was put in there
because it was a killer and what's in there just because they didn't
know where else it should go.  This might just involve running the
test against older versions of Perl and see what blows up, or it might
involve digging back through the p5p archives.

Most of what we dig out of it should be able to fit into existing
tests.  If we truly still have a need for t/op/misc.t, it'll be
reinstated.


So I want to rename it first, then pull it apart to avoid people
continuing to dump miscellaneous tests in while we're trying to get
rid of them.  Rather than pull it apart then rename it.


It's not a huge deal, just so long as it eventually gets pulled apart
and renamed, whatever the order.



 Also, I do not see any particular advantage in having a separate 
 test script *just* for core-dumping tests: the test harnesses
 (TEST and harness) should handle core dumps (as well as they can)
 in any test script.

Problems with that:

Some of these tests require *very specific* conditions and
combinations to cause a failure.  Simply sticking them into an eval
STRING isn't going to trigger it (a lot of them are syntax errors).
You've got to run it as a seperate program.  Thus all the special
life-support in misc.t.

also...

If you try to lump these all together into one test, say 40 of them,
if #8 fails, the test explodes and you don't anything about 9 through
40.  The alternative is to have a seperate test script for each test
with it's own life support.


Finally, since somebody's already done the hard work of setting up a
test that can safely handle core-dumping tests in a cross-platform
manner, why throw it out?


-- 

Michael G. Schwern   [EMAIL PROTECTED]http://www.pobox.com/~schwern/
Perl6 Quality Assurance [EMAIL PROTECTED]   Kwalitee Is Job One
Skrewtape I've heard that semen tastes different depending on diet.  Is that
true?
Skrewtape Hello?
Schwern Skrewtape:  Hang on, I'm conducting research.



Re: [PATCH t/op/misc.t] cleanup

2001-09-01 Thread Michael G Schwern

On Sun, Sep 02, 2001 at 03:15:29AM +0300, Jarkko Hietaniemi wrote:
  @@ -739,3 +783,9 @@
   # keep this last - doesn't seem to work otherwise?
 
 This requirement magically went away while I was away?

Oh, I fixed that but forgot to remove the comment.


-- 

Michael G. Schwern   [EMAIL PROTECTED]http://www.pobox.com/~schwern/
Perl6 Quality Assurance [EMAIL PROTECTED]   Kwalitee Is Job One
GuRuThuG make a channel called #Perl, and infest it with joking and 
   fun it doesnt make alot of sense.