From @moritz

Attached patch, mostly courtesy Vasily "bacek" Chekalkin, adds the
eval_lives_ok and eval_dies_ok methods to Test.pm, and adds three more
passing tests to 'make spectest_regression'.

--
Moritz Lenz
http://moritz.faui2k3.org/ | http://perl-6.de/

From @moritz

Oops, forgot to attach patch. Now it's really there.

--
Moritz Lenz
http://moritz.faui2k3.org/ | http://perl-6.de/

From @moritz

Index: languages/perl6/t/spectest_regression.data
===================================================================
--- languages/perl6/t/spectest_regression.data	(revision 28065)
+++ languages/perl6/t/spectest_regression.data	(working copy)
@@ -6,6 +6,8 @@
 S02-literals/autoref.t
 S02-literals/hex_chars.t
 S02-literals/radix.t
+S02-polymorphic_types/subset-code.t             # pure
+S02-polymorphic_types/subset-range.t
 S03-operators/autoincrement.t                   # pure
 S03-operators/comparison.t
 S03-operators/cross-metaop.t
@@ -39,6 +41,7 @@
 S29-list/map_empty_list.t
 S29-list/map_flattening.t
 S29-list/map_function_return_values.t
+S29-list/map.t
 S29-list/mutating_listops.t                     # pure
 S29-list/sort.t
 S29-num/abs.t
Index: languages/perl6/Test.pm
===================================================================
--- languages/perl6/Test.pm	(revision 28065)
+++ languages/perl6/Test.pm	(working copy)
@@ -122,7 +122,23 @@
     lives_ok($closure, '');
 }
 
+multi sub eval_dies_ok($code, $reason) {
+    eval ( $code );
+    proclaim((defined $!), $reason);
+}
+multi sub eval_dies_ok($code) {
+    eval_dies_ok($code, '');
+}
 
+multi sub eval_lives_ok($code, $reason) {
+    try { eval ($code) }
+    proclaim((not defined $!), $reason);
+}
+multi sub eval_lives_ok($code) {
+    eval_lives_ok($code, '');
+}
+
+
 ## 'private' subs
 
 sub proclaim($cond, $desc) {

From @particle

On Wed, Jun 4, 2008 at 2​:46 PM, Moritz Lenz <moritz@​casella.verplant.org> wrote​:

Oops, forgot to attach patch. Now it's really there.

thanks, applied as of r28074.
~jerry

The RT System itself - Status changed from 'new' to 'open'

From @ronaldxs

Moritz Lenz wrote​:

Oops, forgot to attach patch. Now it's really there.

Your implementations of eval_lives_ok and eval_dies_ok seem
inconsistent. eval_lives_ok uses try and eval_dies_ok does not. The
two implementations may catch different types of exceptions. I am
proposing the patch below​:

$ svn diff Test.pm
Index​: Test.pm

--- Test.pm (revision 28117)
+++ Test.pm (working copy)
@​@​ -122,17 +122,20 @​@​
  lives_ok($closure, '');
}

+sub eval_exception ($code) {
+ my $eval_exception;
+ try { eval ( $code ); $eval_exception = $! }
+ $eval_exception // $!;
+}
multi sub eval_dies_ok($code, $reason) {
- eval ( $code );
- proclaim((defined $!), $reason);
+ proclaim((defined eval_exception($code)), $reason);
}
multi sub eval_dies_ok($code) {
  eval_dies_ok($code, '');
}

multi sub eval_lives_ok($code, $reason) {
- try { eval ($code) }
- proclaim((not defined $!), $reason);
+ proclaim((not defined eval_exception($code)), $reason);
}
multi sub eval_lives_ok($code) {
  eval_lives_ok($code, '');

From @moritz

Ronald Schmidt wrote​:

Moritz Lenz wrote​:

Oops, forgot to attach patch. Now it's really there.

Your implementations of eval_lives_ok and eval_dies_ok seem
inconsistent. eval_lives_ok uses try and eval_dies_ok does not. The
two implementations may catch different types of exceptions. I am
proposing the patch below​:

Good catch. Actually I already encountered I problem with these two
subs, but haven't got around to fix it so far.

Are there any exceptions that are not caught by eval?
Currently rakudo seems to catch every run time exception, but parse
failures still throw an exception. I'm pretty sure that's wrong, because
it's contrary to Perl 6's exception philosophy.

So eval() needs more work, but in the mean time your patch works just
fine (and it will continue to work when eval is fixed), so I vote +1 here.

Cheers,
Moritz

$ svn diff Test.pm
Index​: Test.pm

--- Test.pm (revision 28117)
+++ Test.pm (working copy)
@​@​ -122,17 +122,20 @​@​
lives_ok($closure, '');
}

+sub eval_exception ($code) {
+ my $eval_exception;
+ try { eval ( $code ); $eval_exception = $! }
+ $eval_exception // $!;
+}
multi sub eval_dies_ok($code, $reason) {
- eval ( $code );
- proclaim((defined $!), $reason);
+ proclaim((defined eval_exception($code)), $reason);
}
multi sub eval_dies_ok($code) {
eval_dies_ok($code, '');
}

multi sub eval_lives_ok($code, $reason) {
- try { eval ($code) }
- proclaim((not defined $!), $reason);
+ proclaim((not defined eval_exception($code)), $reason);
}
multi sub eval_lives_ok($code) {
eval_lives_ok($code, '');

--
Moritz Lenz
http://moritz.faui2k3.org/ | http://perl-6.de/

From @particle

On Thu, Jun 5, 2008 at 3​:05 PM, Moritz Lenz <moritz@​casella.verplant.org> wrote​:

Ronald Schmidt wrote​:

Moritz Lenz wrote​:

Oops, forgot to attach patch. Now it's really there.

Your implementations of eval_lives_ok and eval_dies_ok seem
inconsistent. eval_lives_ok uses try and eval_dies_ok does not. The
two implementations may catch different types of exceptions. I am
proposing the patch below​:

Good catch. Actually I already encountered I problem with these two
subs, but haven't got around to fix it so far.

Are there any exceptions that are not caught by eval?
Currently rakudo seems to catch every run time exception, but parse
failures still throw an exception. I'm pretty sure that's wrong, because
it's contrary to Perl 6's exception philosophy.

So eval() needs more work, but in the mean time your patch works just
fine (and it will continue to work when eval is fixed), so I vote +1 here.

agreed, good catch. applied with some formatting tweaks (eg. putting
eval_exception() in the 'private' subs area) as r28124.
~jerry

From @particle

On Fri, Jun 6, 2008 at 7​:34 PM, Vasily Chekalkin <bacek@​bacek.com> wrote​:

jerry gay wrote​:

On Wed, Jun 4, 2008 at 2​:46 PM, Moritz Lenz <moritz@​casella.verplant.org>
wrote​:

Oops, forgot to attach patch. Now it's really there.

thanks, applied as of r28074.
~jerry

This is wrong patch. eval() shouldn't throws any exceptions.

Correct patch for 'eval()' and Test.pm attached.

would you reformat this in universal diff format please? my patch
program doesn't speak git.
~jerry

From @bacek

On Fri Jun 06 20​:11​:36 2008, particle wrote​:

(Sorry for spam if any. Looks like my message was rejected by mailman.)

eval() shouldn't throws any exceptions.

Correct patch for 'eval()' and Test.pm attached.

--
Bacek.

From @bacek

===================================================
--- Test.pm (orig)
+++ Test.pm (modified)
@@ -123,14 +123,16 @@ multi sub lives_ok($closure) {
 }
 
 multi sub eval_dies_ok($code, $reason) {
-    proclaim((defined eval_exception($code)), $reason);
+	eval($code);
+    proclaim(defined($!), $reason);
 }
 multi sub eval_dies_ok($code) {
     eval_dies_ok($code, '');
 }
 
 multi sub eval_lives_ok($code, $reason) {
-    proclaim((not defined eval_exception($code)), $reason);
+	eval($code);
+    proclaim((not defined $!), $reason);
 }
 multi sub eval_lives_ok($code) {
     eval_lives_ok($code, '');
@@ -139,12 +141,6 @@ multi sub eval_lives_ok($code) {
 
 ## 'private' subs
 
-sub eval_exception($code) {
-    my $eval_exception;
-    try { eval ($code); $eval_exception = $! }
-    $eval_exception // $!;
-}
-
 sub proclaim($cond, $desc) {
     $testing_started  = 1;
     $num_of_tests_run = $num_of_tests_run + 1;
===================================================
--- src/builtins/control.pir (orig)
+++ src/builtins/control.pir (modified)
@@ -113,13 +113,12 @@ on error.
     unless have_lang goto no_lang
     'die'('Lanuage parameter to eval unimplemented.')
   no_lang:
+    push_eh catch
 
     .local pmc compiler, invokable
     .local pmc res, exception
     compiler = compreg 'Perl6'
     invokable = compiler.'compile'(code)
-
-    push_eh catch
     res = invokable()
     pop_eh
     exception = new 'Failure'

From @bacek

jerry gay wrote​:

On Wed, Jun 4, 2008 at 2​:46 PM, Moritz Lenz <moritz@​casella.verplant.org> wrote​:

Oops, forgot to attach patch. Now it's really there.

thanks, applied as of r28074.
~jerry

This is wrong patch. eval() shouldn't throws any exceptions.

Correct patch for 'eval()' and Test.pm attached.

--
Bacek.

From @bacek

diff --git a/languages/perl6/Test.pm b/languages/perl6/Test.pm
index 5cb04ae..ccc34a9 100644
--- a/languages/perl6/Test.pm
+++ b/languages/perl6/Test.pm
@@ -123,14 +123,16 @@ multi sub lives_ok($closure) {
 }
 
 multi sub eval_dies_ok($code, $reason) {
-    proclaim((defined eval_exception($code)), $reason);
+	eval(code);
+    proclaim(defined($!), $reason);
 }
 multi sub eval_dies_ok($code) {
     eval_dies_ok($code, '');
 }
 
 multi sub eval_lives_ok($code, $reason) {
-    proclaim((not defined eval_exception($code)), $reason);
+	eval($code);
+    proclaim((not defined $!), $reason);
 }
 multi sub eval_lives_ok($code) {
     eval_lives_ok($code, '');
@@ -139,12 +141,6 @@ multi sub eval_lives_ok($code) {
 
 ## 'private' subs
 
-sub eval_exception($code) {
-    my $eval_exception;
-    try { eval ($code); $eval_exception = $! }
-    $eval_exception // $!;
-}
-
 sub proclaim($cond, $desc) {
     $testing_started  = 1;
     $num_of_tests_run = $num_of_tests_run + 1;
diff --git a/languages/perl6/src/builtins/control.pir b/languages/perl6/src/builtins/control.pir
index 479b1b6..558a4c4 100644
--- a/languages/perl6/src/builtins/control.pir
+++ b/languages/perl6/src/builtins/control.pir
@@ -113,13 +113,12 @@ on error.
     unless have_lang goto no_lang
     'die'('Lanuage parameter to eval unimplemented.')
   no_lang:
+    push_eh catch
 
     .local pmc compiler, invokable
     .local pmc res, exception
     compiler = compreg 'Perl6'
     invokable = compiler.'compile'(code)
-
-    push_eh catch
     res = invokable()
     pop_eh
     exception = new 'Failure'

From @bacek

jerry gay wrote​:

would you reformat this in universal diff format please? my patch
program doesn't speak git.

Strange... It is 'universal diff' format. Can be applied with
'patch -p1 < eval.diff' in top-level parrot directory. Or with
'patch -p3 < eval.diff' in languages/perl6.

Anyway, attached patch can be applied within 'patch -p0' in
languages/perl6 directory.

--
Bacek.

From @bacek

===================================================
--- Test.pm (orig)
+++ Test.pm (modified)
@@ -123,14 +123,16 @@ multi sub lives_ok($closure) {
 }
 
 multi sub eval_dies_ok($code, $reason) {
-    proclaim((defined eval_exception($code)), $reason);
+	eval(code);
+    proclaim(defined($!), $reason);
 }
 multi sub eval_dies_ok($code) {
     eval_dies_ok($code, '');
 }
 
 multi sub eval_lives_ok($code, $reason) {
-    proclaim((not defined eval_exception($code)), $reason);
+	eval($code);
+    proclaim((not defined $!), $reason);
 }
 multi sub eval_lives_ok($code) {
     eval_lives_ok($code, '');
@@ -139,12 +141,6 @@ multi sub eval_lives_ok($code) {
 
 ## 'private' subs
 
-sub eval_exception($code) {
-    my $eval_exception;
-    try { eval ($code); $eval_exception = $! }
-    $eval_exception // $!;
-}
-
 sub proclaim($cond, $desc) {
     $testing_started  = 1;
     $num_of_tests_run = $num_of_tests_run + 1;
===================================================
--- src/builtins/control.pir (orig)
+++ src/builtins/control.pir (modified)
@@ -113,13 +113,12 @@ on error.
     unless have_lang goto no_lang
     'die'('Lanuage parameter to eval unimplemented.')
   no_lang:
+    push_eh catch
 
     .local pmc compiler, invokable
     .local pmc res, exception
     compiler = compreg 'Perl6'
     invokable = compiler.'compile'(code)
-
-    push_eh catch
     res = invokable()
     pop_eh
     exception = new 'Failure'

From @ronaldxs

Vasily Chekalkin wrote​:

This is wrong patch. eval() shouldn't throws any exceptions.

Correct patch for 'eval()' and Test.pm attached.

As Mr Lenz wrote with respect to eval exceptions and this patch​:

Are there any exceptions that are not caught by eval?
Currently rakudo seems to catch every run time exception,
/***** BUT PARSE FAILURES STILL THROW AN EXCEPTION. *****/
I'm pretty sure that's wrong, because
it's contrary to Perl 6's exception philosophy.

So eval() needs more work, but in the mean time your patch works just
fine (and it will continue to work when eval is fixed), so I vote +1 here.

For example​:

eval 'my $x = 2; $x = $x / 0; say $x;'; say 'eval exception ', $!;
eval exception Divide by zero

eval 'my $x; my $y; ($x = $y) = 3; say $x;'; say 'eval exception ', $!;
Method 'lvalue' not found for invocant of class 'PAST​::Stmts'

In the second case eval throws an exception and it would make sense for
eval_dies_ok to catch it and report it as a case that died. The second
case is from the pugs lvalue.t suite. I am currently doing work in that
area and would like to use eval_(dies|lives)_ok for some tests including
tests from lvalue.t.

Ron

From @bacek

On Sat Jun 07 07​:13​:07 2008, ronaldxs wrote​:

Probably I don't understand correctly your question. This is output from
my version of patched rakudo​:

$ ../../parrot perl6.pbc

eval 'my $x = 2; $x = $x / 0; say $x;'; say 'eval exception ', $!;
eval exception Divide by zero

Is it what you want?

--
Bacek

From allbery@ece.cmu.edu

On 2008 Jun 6, at 23​:53, Vasily Chekalkin wrote​:

jerry gay wrote​:

would you reformat this in universal diff format please? my patch
program doesn't speak git.

Strange... It is 'universal diff' format. Can be applied with
'patch -p1 < eval.diff' in top-level parrot directory. Or with
'patch -p3 < eval.diff' in languages/perl6.

Anyway, attached patch can be applied within 'patch -p0' in
languages/perl6 directory.
--- Test.pm (orig)
+++ Test.pm (modified)
@​@​ -123,14 +123,16 @​@​ multi sub lives_ok($closure) {

Your mail program is mangling line endings (note last line of above
quote). Try it as an attachment.

--
brandon s. allbery [solaris,freebsd,perl,pugs,haskell] allbery@​kf8nh.com
system administrator [openafs,heimdal,too many hats] allbery@​ece.cmu.edu
electrical and computer engineering, carnegie mellon university KF8NH

From @moritz

I'm resolving this ticket, because now (aka r28926) eval_lives_ok and
eval_dies_ok are implemented (and implemented in a consistent way).

If eval() still throws an exception in some cases that should be a
separate ticket.

Cheers,
Moritz

@moritz - Status changed from 'open' to 'resolved'