Bleadperl v5.19.3-296-gffdb8b1 breaks BAREFOOT/Test-File-1.34.tar.gz #13224
Comments
From @andkgit bisect commit ffdb8b1 Fix two line numbers bugs involving quote-like ops diagnostics Also affected: FERRENCY/Text-MicroMason-2.13.tar.gz Sample fails not yet arrived at cpantesters. This is from Text::MicroMason: # Failed test 'first line of $@ describes the error location' perl -V Summary of my perl5 (revision 5 version 19 subversion 4) configuration: Characteristics of this binary (from libperl): -- |
From @cpansproutOn Tue Sep 03 19:58:33 2013, andreas.koenig.7os6VVqR@franz.ak.mind.de wrote:
The Test::File failure can be reduced to: foo( The line number returned by caller() (and used by warn) is a single "@{[]}" used to reset the line number for the surrounding statement to The reason this happens is that PL_copline is now localised and Test::File’s test suite is sensitive to that, as it uses As long as we are limited to one line number per statement, which line If necessary, though, I could allow changes to PL_copline inside the (For those who have no idea what I am talking about, PL_copline is set
So now perl reports 9 instead of 8.
And the syntax error is on line 9. So that looks like a bug fix, too. -- Father Chrysostomos |
|
The RT System itself - Status changed from 'new' to 'open' |
From @cpansproutOn Sat Sep 07 01:20:26 2013, sprout wrote:
Maybe the best solution here is to allow entersub ops to store a line Maybe all ops should store a line number.... (I’m not the first to -- Father Chrysostomos |
From @cpansproutOn Sat Sep 07 18:47:34 2013, sprout wrote:
Could someone else familiar with ops review my suggestions? Should I -- Father Chrysostomos |
From @bulk88On Sat Dec 21 18:07:08 2013, sprout wrote:
I suggest won't fix. IMO line numbers are on a statement, not expression/substatement basis. None of the suggestions sound like the increase in memory or CPU will justify a bug about undocumented and subject to change behavior. -- |
From zefram@fysh.orgFather Chrysostomos via RT wrote:
None of the above. The current behaviour is not a bug; there is not -zefram |
From @bulk88On Sat Dec 21 20:25:13 2013, bulk88 wrote:
I did think of something, what if a single bit in the op struct indicated if this op is on a new line, or previous line. The line number of the current op * is determined by walking from the COP along the the linked list to the current op * and counting the number of newline bits. The algorithm for computing the line number is very overhead intensive/slow, but takes very very little memory, just 1 bit per op struct. -- |
From zefram@fysh.orgbulk88 via RT wrote:
Unworkable. With conditionals, multiple paths can be taken through the -zefram |
From perl5-porters@perl.orgZefram wrote:
When does *that* occur? |
From zefram@fysh.orgFather Chrysostomos wrote:
#line can trivially make a statement notionally split between files. -zefram |
From @jkeenanBuddy Burden has released dev version 1.35_01 of Test-File, which appears to address the problem. Once he promotes that to a full version, we'll be able to close this ticket. Thank you very much. |
From @rurbanOn Mon, Dec 23, 2013 at 6:59 AM, Zefram <zefram@fysh.org> wrote:
So you have no idea about my oplines branch from 2008, which moves the It it not ready, as I did just to test the optimization idea. -- |
From @rjbsOn Wed Dec 25 18:25:51 2013, jkeenan wrote:
Jim: Test-File 1.36 is out: https://metacpan.org/release/Test-File Are we done? -- |
From @jkeenanOn Mon Jan 20 19:47:17 2014, rjbs wrote:
I will take the ticket for the purpose of investigating and confirming. |
From @jkeenanOn Tue Jan 21 04:55:37 2014, jkeenan wrote:
Test-File v1.36 does pass all its tests at: However, reviewing the ticket, I see that it was originally filed against CPAN distro Text-MicroMason. Some of this distro's tests are still failing: ##### At the moment, I'm not able to diagnose this further. Suggestions? Thank you very much. |
From @tonycozOn Tue Jan 21 10:22:35 2014, jkeenan wrote:
Bisected down to: commit ffdb8b1 Fix two line numbers bugs involving quote-like ops I was going to try and fix #line directives in quote-like operators, Before this commit, the #line directive would last until the end of qq{${ with: perl ../bisect.pl --start v5.12.0 -e '$code = `cat /home/tonyc/foo.compiled.pl`; eval where foo.compiled.pl contains the code generated from the template. This appears to be the result of a bug fix, if I expand the Hello world string in the by a few more lines the error message on 5.14.2 stays on line 8, but increases correctly in blead. I've reported this to the Text-MicroMason queue as https://rt.cpan.org/Ticket/Display.html?id=92771 Tony |
From @jkeenanOn Wed Feb 05 16:34:26 2014, tonyc wrote:
Tony, if I understand you correctly, Text-MicroMason was inadvertently relying on buggy behavior in Perl 5 -- behavior which we have now corrected. If so, given that you have notified Text-MicroMason of this problem, we can close this ticket and remove a blocker for 5.20. I will close this ticket within 7 days unless someone convinces me otherwise. Thank you very much. |
From @tonycozOn Wed Feb 05 19:31:15 2014, jkeenan wrote:
Exactly that.
I've closed it. Tony |
|
@tonycoz - Status changed from 'open' to 'resolved' |
From @andk"Father Chrysostomos via RT" <perlbug-followup@perl.org> writes:
Also affected: IKEGAMI/Syntax-Feature-QwComments-v1.12.0.tar.gz -- |
From @ikegamiOn Sun Dec 22 04:08:19 2013, zefram@fysh.org wrote:
What do you mean by "current behaviour"? Currently, blead is issuing error messages with no line numbers, and I'd consider that a bug. |
From @ikegamiOn Fri May 02 06:28:45 2014, ikegami@adaelis.com wrote:
Or maybe that wasn't mentioned? The commit also removes " at -e line 1" from the following: $ ./perl -e'qw' |
|
@ikegami - Status changed from 'resolved' to 'open' |
From @tonycozOn Fri May 02 06:41:33 2014, ikegami@adaelis.com wrote:
Please try the attached patch. Tony |
From @tonycoz0001-perl-119593-ensure-qw-q-and-report-line-on-a-missing.patchFrom f1e03a55e00886d916008377fb3adc2913e35288 Mon Sep 17 00:00:00 2001
From: Tony Cook <tony@develop-help.com>
Date: Tue, 6 May 2014 16:27:14 +1000
Subject: [perl #119593] ensure qw//, q'' and '' report line on a missing
terminator
---
t/lib/croak/toke | 15 +++++++++++++++
toke.c | 8 ++++----
2 files changed, 19 insertions(+), 4 deletions(-)
diff --git a/t/lib/croak/toke b/t/lib/croak/toke
index d6026d2..0572094 100644
--- a/t/lib/croak/toke
+++ b/t/lib/croak/toke
@@ -134,3 +134,18 @@ Execution of - aborted due to compilation errors.
<<"foo
EXPECT
Unterminated delimiter for here document at - line 1.
+########
+# NAME Unterminated qw//
+qw/
+EXPECT
+Can't find string terminator "/" anywhere before EOF at - line 1.
+########
+# NAME Unterminated q//
+qw/
+EXPECT
+Can't find string terminator "/" anywhere before EOF at - line 1.
+########
+# NAME Unterminated ''
+'
+EXPECT
+Can't find string terminator "'" anywhere before EOF at - line 1.
diff --git a/toke.c b/toke.c
index 3d992f6..ea88183 100644
--- a/toke.c
+++ b/toke.c
@@ -6848,6 +6848,8 @@ Perl_yylex(pTHX)
case '\'':
s = scan_str(s,!!PL_madskills,FALSE,FALSE,FALSE,NULL);
+ if (!s)
+ missingterm(NULL);
COPLINE_SET_FROM_MULTI_END;
DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
if (PL_expect == XOPERATOR) {
@@ -6857,8 +6859,6 @@ Perl_yylex(pTHX)
else
no_op("String",s);
}
- if (!s)
- missingterm(NULL);
pl_yylval.ival = OP_CONST;
TERM(sublex_start());
@@ -8385,9 +8385,9 @@ Perl_yylex(pTHX)
case KEY_q:
s = scan_str(s,!!PL_madskills,FALSE,FALSE,FALSE,NULL);
- COPLINE_SET_FROM_MULTI_END;
if (!s)
missingterm(NULL);
+ COPLINE_SET_FROM_MULTI_END;
pl_yylval.ival = OP_CONST;
TERM(sublex_start());
@@ -8397,9 +8397,9 @@ Perl_yylex(pTHX)
case KEY_qw: {
OP *words = NULL;
s = scan_str(s,!!PL_madskills,FALSE,FALSE,FALSE,NULL);
- COPLINE_SET_FROM_MULTI_END;
if (!s)
missingterm(NULL);
+ COPLINE_SET_FROM_MULTI_END;
PL_expect = XOPERATOR;
if (SvCUR(PL_lex_stuff)) {
int warned_comma = !ckWARN(WARN_QW);
--
1.7.10.4
|
From @tonycozOn Mon May 05 23:28:08 2014, tonyc wrote:
New patch with explanation added. Tony |
From @tonycoz0001-perl-119593-ensure-qw-q-and-report-line-on-a-missing.patchFrom bf65bd5366e6c89587d7f8f75c8ec3cdc37ac7ab Mon Sep 17 00:00:00 2001
From: Tony Cook <tony@develop-help.com>
Date: Wed, 7 May 2014 09:33:03 +1000
Subject: [perl #119593] ensure qw//, q'' and '' report line on a missing
terminator
scan_str() only sets PL_multi_end on success, but the qw, q amd '' cases
were calling the COPLINE_SET_FROM_MULTI_END; macro before reporting
failure, overwriting the line used for reporting errors.
For the simplest case, as in the ticket and the tests, PL_multi_end is
zero, so the error is reported without a line number.
---
t/lib/croak/toke | 15 +++++++++++++++
toke.c | 8 ++++----
2 files changed, 19 insertions(+), 4 deletions(-)
diff --git a/t/lib/croak/toke b/t/lib/croak/toke
index d6026d2..0572094 100644
--- a/t/lib/croak/toke
+++ b/t/lib/croak/toke
@@ -134,3 +134,18 @@ Execution of - aborted due to compilation errors.
<<"foo
EXPECT
Unterminated delimiter for here document at - line 1.
+########
+# NAME Unterminated qw//
+qw/
+EXPECT
+Can't find string terminator "/" anywhere before EOF at - line 1.
+########
+# NAME Unterminated q//
+qw/
+EXPECT
+Can't find string terminator "/" anywhere before EOF at - line 1.
+########
+# NAME Unterminated ''
+'
+EXPECT
+Can't find string terminator "'" anywhere before EOF at - line 1.
diff --git a/toke.c b/toke.c
index 3d992f6..ea88183 100644
--- a/toke.c
+++ b/toke.c
@@ -6848,6 +6848,8 @@ Perl_yylex(pTHX)
case '\'':
s = scan_str(s,!!PL_madskills,FALSE,FALSE,FALSE,NULL);
+ if (!s)
+ missingterm(NULL);
COPLINE_SET_FROM_MULTI_END;
DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
if (PL_expect == XOPERATOR) {
@@ -6857,8 +6859,6 @@ Perl_yylex(pTHX)
else
no_op("String",s);
}
- if (!s)
- missingterm(NULL);
pl_yylval.ival = OP_CONST;
TERM(sublex_start());
@@ -8385,9 +8385,9 @@ Perl_yylex(pTHX)
case KEY_q:
s = scan_str(s,!!PL_madskills,FALSE,FALSE,FALSE,NULL);
- COPLINE_SET_FROM_MULTI_END;
if (!s)
missingterm(NULL);
+ COPLINE_SET_FROM_MULTI_END;
pl_yylval.ival = OP_CONST;
TERM(sublex_start());
@@ -8397,9 +8397,9 @@ Perl_yylex(pTHX)
case KEY_qw: {
OP *words = NULL;
s = scan_str(s,!!PL_madskills,FALSE,FALSE,FALSE,NULL);
- COPLINE_SET_FROM_MULTI_END;
if (!s)
missingterm(NULL);
+ COPLINE_SET_FROM_MULTI_END;
PL_expect = XOPERATOR;
if (SvCUR(PL_lex_stuff)) {
int warned_comma = !ckWARN(WARN_QW);
--
1.7.10.4
|
From @khwilliamsonI spent an hour or so familiarizing myself with the code and the patch, and it looks good to me. So with rjbs' approval, I've pushed it as commit 9612f77 |
|
@khwilliamson - Status changed from 'open' to 'resolved' |
From @ikegamiI had problems with my linux box that kept me from testing it until today. On Mon, May 12, 2014 at 3:07 PM, Karl Williamson via RT <
|
Migrated from rt.perl.org#119593 (status was 'resolved')
Searchable as RT119593$
The text was updated successfully, but these errors were encountered: