Red Hat Bugzilla – Bug 78919
The split function has a loop error
Last modified: 2007-04-18 12:48:49 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0; LM-MS; T312461) Description of problem: The split function produces a loop error when attempting to split a string. This problem exists in the 5.8.0 version (RH 8.0) of Perl but not the 5.6.1 version (RH 7.3). Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Execute the splitTest.pl script 2. 3. Actual Results: The error occurs at line 34 of the script with Perl 5.8.0, prints: Start problem split Works with ver 5.6.1 not 5.8.0 Split loop. Expected Results: Start problem split Works with ver 5.6.1 not 5.8.0 Number of elements = 4 [ junker.pl ] [ junk.pl ] [ splitTest.pl ] [ test.pl ] end of program Additional info:
Created attachment 87230 [details] Perl test script
If you backslash the space in the split function does it work? Alternatively, you can use '\s' which should encompass all whitespace.
(@fieldList) = split /[ \t\n]+/, $fileList; # original produces loop error #(@fieldList) = split /[' '\t\n]+/, $fileList; # loop error #(@fieldList) = split (/[' '\t\n]+/, $fileList); # loop error #(@fieldList) = split /[\ \t\n]+/, $fileList; # loop error #(@fieldList) = split (/[\ \t\n]+/, $fileList); # loop error #(@fieldList) = split /[\s\t\n]+/, $fileList; # loop error #(@fieldList) = split (/[\s\t\n]+/, $fileList); # loop error #(@fieldList) = split ' ', $fileList; # does work with 5.8 All produce the same loop error with Perl 5.8, all work with Perl 5.6.1.
Created attachment 87265 [details] Output of your script on my machine
WORKSFORME. I see no error when running your script. I even went so far to uncomment the lines which you said caused more loop errors. My original suggestion to you works without a problem: @fieldList = split /\s+/, $fileList; $count2 = @fieldList; print "Number of elements = ", $count2 ,"\n"; for $count2 ( @fieldList ) { print "\t [ $count2 ] \n"; } I received the following as output (as expected): Number of elements = 3 [ junker.pl ] [ junk.pl ] [ test.pl ] Did you try running this under a Red Hat Linux 8.0 box, or under Cygwin session (judging from the path of your attachment)?
I ran the script under Cygwin (no problem), Red Hat 7.3 (no problem), Red Hat 8.0 (problem). I guess I will have to take the script home and run on my RH 8.0 system, since I only have 3 Intel boxes to play with.
I'm just wondering because the following: @fieldList = split /\s+/, $fileList; @fieldList = split /[\ \t\n]+/, $fileList; are both Perl clean. I verified it on two other Red Hat Linux 8 based boxes without having a problem. $ rpm -q perl perl-5.8.0-55
I don't have any problem with: @fieldList = split /\s+/, $fileList; either, works just like for my data set: (@fieldList) = split ' ', $fileList;
' ' == simply a space. '\s' == [\ \t\n\r] (or the plethora of whitespace that's out there). Either way - this should not be a bug since I can get your script working fine on a Red Hat Linux 8 box.
I have executed the script on another RH 8 installation and I have the same problem. @fieldList = split /\s+/, $fileList; #works @fieldList = split /[\ \t\n]+/, $fileList; #loop error
$ perl $fileList = "test.pl\n Test.pl\ string.pl\n"; @fieldList = split /[\ \t\n]+/, $fileList; foreach(@fieldList) { print "$_\n"; } << CTRL+D >> test.pl Test.pl string.pl Still WORKSFORME on two different Red Hat platforms. Could you please provide the output of what you are doing since I can't generate your error message? It might help (or may not).
Created attachment 87384 [details] Test script and output in a tar ball
I have added an attachment with the current version of the test script and the output. I sent the script out to another organization they were able to execute it correctly on all but one of there RH 8.0 installs. The, now 3, systems that have a problem with the script have one thing in common all three were total new installations. The other RH 8.0 systems that work were upgraded from 7.2 to 8.0. Do you know how your machines got to be 8.0? If all of your systems are upgrades then this problem may only apply to new 8.0 installations verses upgrades to 8.0. Thoughts? All tests were run on Red Hat 8.0 Linux systems. My administrative system is a win2k system, I use Cygwin on the win2k system so I don't have to use DOS.
I see the "Split loop" error in your text file (but have yet to see it on my system). I have three Red Hat Linux 8.0 workstations - one fresh install, one upgraded from Red Hat Linux 7.2->7.3.94 (null)->8.0, one upgraded from Red Hat Linux 7.1->8.0. All three have no problems. Since you've been able to reproduce this problem - run your perl script as follows (go into debugging mode): perl -d splitTest.pl Quick reference: Enter "s" for each line until you come to the culprit and submit all the relavent output information when it script fails.
The trace from line 36 until the script terminated. main::(splitTest.pl:36): print "Start problem split\nWorks with ver 5.6.1 not 5.8.0\n"; DB<1> s Start problem split Works with ver 5.6.1 not 5.8.0 main::(splitTest.pl:38): @fieldList = split /\s+/, $fileList; # does work with 5.8 DB<1> s main::(splitTest.pl:39): print "Split 2\n\n"; DB<1> s Split 2 main::(splitTest.pl:41): (@fieldList) = split /[ \t\n]+/, $fileList; # original produces loop error DB<1> s main::CODE(0x8209ef8)(splitTest.pl:41): 41: (@fieldList) = split /[ \t\n]+/, $fileList; # original produces loop error DB<1> s utf8::CODE(0x8209ef8)(/usr/lib/perl5/5.8.0/utf8.pm:3): 3: $utf8::hint_bits = 0x00800000; DB<1> s utf8::CODE(0x8209ef8)(/usr/lib/perl5/5.8.0/utf8.pm:5): 5: our $VERSION = '1.00'; DB<1> s utf8::CODE(0x8209ef8)(/usr/lib/perl5/5.8.0/utf8.pm:22): 22: 1; 23: __END__ DB<1> s main::CODE(0x8209ef8)(splitTest.pl:41): 41: (@fieldList) = split /[ \t\n]+/, $fileList; # original produces loop error DB<1> s main::CODE(0x8209ef8)(splitTest.pl:41): 41: (@fieldList) = split /[ \t\n]+/, $fileList; # original produces loop error DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:18): 18: my ($class, $type, $list, $minbits, $none) = @_; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:19): 19: local $^D = 0 if $^D; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:21): 21: print STDERR "SWASHNEW @_\n" if DEBUG; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:35): 35: my $file; ## file to load data from, and also part of the %Cache key. DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:36): 36: my $ListSorted = 0; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:38): 38: if ($type) 39: { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:143): 143: my $extras; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:144): 144: my $bits; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:146): 146: my $ORIG = $list; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:147): 147: if ($list) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:148): 148: my @tmp = split(/^/m, $list); DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:149): 149: my %seen; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:151): 151: $extras = join '', grep /^[^0-9a-fA-F]/, @tmp; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:154): 154: grep {/^([0-9a-fA-F]+)/ and not $seen{$1}++} @tmp; # XXX doesn't do ranges right DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:153): 153: sort { hex $a <=> hex $b } DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:157): 157: if ($none) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:162): 162: if ($minbits < 32) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:163): 163: my $top = 0; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:164): 164: while ($list =~ /^([0-9a-fA-F]+)(?:\t([0-9a-fA-F]+)?)(?:\t([0- 9a-fA-F]+))?/mg) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:172): 172: $top > 0xffff ? 32 : 173: $top > 0xff ? 16 : 174: $top > 1 ? 8 : 1 DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:176): 176: $bits = $minbits if $bits < $minbits; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:178): 178: my @extras; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:179): 179: for my $x ($extras) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:180): 180: pos $x = 0; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:181): 181: while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:182): 182: my $char = $1; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:183): 183: my $name = $2; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:184): 184: print STDERR "$1 => $2\n" if DEBUG; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:185): 185: if ($char =~ /[-+!]/) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:181): 181: while ($x =~ /^([^0-9a-fA-F\n])(.*)/mg) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:179): 179: for my $x ($extras) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:201): 201: print STDERR "CLASS = $class, TYPE => $type, BITS => $bits, NONE => $none\nEXTRAS =>\n$extras\nLIST =>\n$list\n" if DEBUG; DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:203): 203: my $SWASH = bless { 204: TYPE => $type, 205: BITS => $bits, 206: EXTRAS => $extras, 207: LIST => $list, 208: NONE => $none, 209: @extras, DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:212): 212: if ($file) { DB<1> s utf8::SWASHNEW(/usr/lib/perl5/5.8.0/utf8_heavy.pl:216): 216: return $SWASH; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:223): 223: my ($self, $start, $len) = @_; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:224): 224: local $^D = 0 if $^D; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:225): 225: my $type = $self->{TYPE}; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:226): 226: my $bits = $self->{BITS}; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:227): 227: my $none = $self->{NONE}; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:228): 228: print STDERR "SWASHGET @_ [$type/$bits/$none]\n" if DEBUG; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:229): 229: my $end = $start + $len; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:230): 230: my $swatch = ""; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:231): 231: my $key; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:232): 232: vec($swatch, $len - 1, $bits) = 0; # Extend to correct length. DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:233): 233: if ($none) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:237): 237: for ($self->{LIST}) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:238): 238: pos $_ = 0; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:239): 239: if ($bits > 1) { 240: LINE: DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:274): 274: while (/^([0-9a-fA-F]+)(?:[ \t]+([0-9a-fA-F]+))?/mg) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:237): 237: for ($self->{LIST}) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:289): 289: for my $x ($self->{EXTRAS}) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:290): 290: pos $x = 0; DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:291): 291: while ($x =~ /^([-+!])(.*)/mg) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:289): 289: for my $x ($self->{EXTRAS}) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:334): 334: if (DEBUG) { DB<1> s utf8::SWASHGET(/usr/lib/perl5/5.8.0/utf8_heavy.pl:341): 341: $swatch; DB<1> s Split loop. Debugged program terminated. Use q to quit or R to restart, use O inhibit_exit to avoid stopping after program termination, h q, h R or h O to get additional info. DB<1>
Yee-haw - now we are getting somewhere. I do not see the utf8_heavy.pl references because my language is not set to "en_US.utf8". It is instead set to "en_US". There is our difference. And you are accurate in assessing that it is a bug since Perl does not seem to be handling the split properly when addressing "en_US.utf8" - I can now duplicate your problem with the "Split loop" bug (boy, it would have been handy if I had recommended you to run the Perl debugger in the first place!) You can work around this problem temporarily by executing your script as: LANG=en_US ./splitTest.pl _OR_ Adding "export LANG=en_US" to your .bashrc script in your home directory.
Added "export LANG=en_US" to my .bashrc script in my home directory now all works well. from an environment dump before LANG=en_US.UTF-8 after LANG=en_US Thanks for your help. I'll follow the resolution of the problem.
Chip, is this problem resolved now? Just curious as I have bugs reported about this same problem with procinfo's lsdev script. I'm unable to reproduce it at all on any OS version on any system, and other bug reports seem to indicate this problem is fixed. bug #82432 is the one I'm currently investigating