LTA error messages on literals with too many decimals in Rakudo

[p6rt]

Migrated from rt.perl.org#73236 (status was 'resolved')

Searchable as RT73236$

[p6rt]

From @masak

<masak> rakudo​: say 0.00000000000000000000 # also by diakopterbot
<p6eval> rakudo 6867a5​: OUTPUT«No applicable candidates found to
dispatch to for 'infix​:<div>' [...]
<masak> rakudo​: say 0.0000000000000000000
<p6eval> rakudo 6867a5​: OUTPUT«Nominal type check failed for parameter
'$b'; expected Int but got Num instead [...]
* masak submits rakudobug

Here it's just a question of catching the error sooner and emitting a
nice error message. At least until we can actually handle arbitrarily
long decimal expansions.

[p6rt]

From @masak

<lue> rakudo​: say 0.00000000000000000000000000000000​
0000000000000000000000000000000000
<p6eval> rakudo 5e5969​: OUTPUT«Divide by zero␤current instr.​:
'infix​:<div>' [...]
<iakopter> nice! 3 different bugs
<lue> the exact error changes with the number of extra 0's you have!
<lue> rakudo​: say 0.0000000000000000000000000000000​
00000000001000000000000000000000 # hee hee
<p6eval> rakudo 5e5969​: OUTPUT«Nominal type check failed for
parameter '$a'; expected Int but got Num instead [...]
<diakopter> omg
<lue> first $b, now $a!
<diakopter> 4 different bugs

[p6rt]

From [Unknown Contact. See original ticket]

<lue> rakudo​: say 0.00000000000000000000000000000000​
0000000000000000000000000000000000
<p6eval> rakudo 5e5969​: OUTPUT«Divide by zero␤current instr.​:
'infix​:<div>' [...]
<iakopter> nice! 3 different bugs
<lue> the exact error changes with the number of extra 0's you have!
<lue> rakudo​: say 0.0000000000000000000000000000000​
00000000001000000000000000000000 # hee hee
<p6eval> rakudo 5e5969​: OUTPUT«Nominal type check failed for
parameter '$a'; expected Int but got Num instead [...]
<diakopter> omg
<lue> first $b, now $a!
<diakopter> 4 different bugs

[p6rt]

@masak - Status changed from 'new' to 'open'

[p6rt]

From @bbkr

all cases work on Kiev build, taking for tests.

[p6rt]

From @bbkr

It looks like it's not resolved yet for very long decimal part​:

$ perl6 -e 'eval "say 0." ~ "0" x 1024; say $!'
Divide by zero

Leaving open.

[p6rt]

From @bbkr

last case works on * release

perl6 -e 'eval "say 0." ~ "0" x 1024; say $!'
0
Any()

[p6rt]

From @bbkr

also did tests using following code​:

use Test; my $e = "0."; for 1..1024 { $e ~= "0"; eval_lives_ok $e }

and all cases passed

[p6rt]

From @bbkr

tests added in t/spec/S02-literals/numeric.t

[p6rt]

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

[p6rt]

From @kyleha

This is an automatically generated mail to inform you that tests are now available in t/spec/S02-literals/numeric.t

commit 806c7d7a2efa5235bb9ee53d0d0a60a7ebe94a74
Author​: bbkr <bbkr@​c213334d-75ef-0310-aa23-eaa082d1ae64>
Date​: Tue Aug 3 11​:59​:02 2010 +0000

  [t/spec] tests for RT #​73236 LTA error messages on literals with too many decimals
 
  git-svn-id​: http://svn.pugscode.org/pugs@&#8203;31892 c213334d-75ef-0310-aa23-eaa082d1ae64

Inline Patch

diff --git a/t/spec/S02-literals/numeric.t b/t/spec/S02-literals/numeric.t
index 135fb87..921358f 100644
--- a/t/spec/S02-literals/numeric.t
+++ b/t/spec/S02-literals/numeric.t
@@ -31,6 +31,24 @@ nok 1+1i ~~ Real, '1+1i is not Real';
 is_approx 3.14159265358979323846264338327950288419716939937510e0,
           3.141592, 'very long Num literals';
 
+# RT #73236
+{
+    eval_lives_ok '0.' ~ '0' x 19,
+        'parsing 0.000... with 19 decimal places lives';
+
+    eval_lives_ok '0.' ~ '0' x 20,
+        'parsing 0.000... with 20 decimal places lives';
+
+    eval_lives_ok '0.' ~ '0' x 63,
+        'parsing 0.000... with 63 decimal places lives';
+
+    eval_lives_ok '0.' ~ '0' x 66,
+        'parsing 0.000... with 66 decimal places lives';
+
+    eval_lives_ok '0.' ~ '0' x 1024,
+        'parsing 0.000... with 1024 decimal places lives';
+}
+
 done_testing;
 
 # vim: ft=perl6 sw=4 ts=4 expandtab