failure when line length approaches 2^30 characters #10212
Comments
From mbmiller+m@gmail.comThis bug is highly replicable. See the attached perlbug -d output, but we The problem is that perl fails when line length approaches 2^30 characters $ perl -pi -e 'BEGIN{undef $/} ; s/\A/ID /' foo.txt ...with this result: $ ls -l foo.txt That is, zero bytes remain from a 1.5 GB file. (I know that I should have $ head -c 1073741818 foo.txt | perl -pe 's/./AB/' > bar.txt That did not fail, so I tried this: $ head -c 1073741819 foo.txt | perl -pe 's/./AB/' > bar.txt But that failed (with the same "Substitution ... chunk 1." message as Of course, 2^30 = 1073741824, which happens to be exactly 5 bytes greater cc: three people who helped me figure this out Mike -- |
From mbmiller+m@gmail.comFlags: This perlbug was built using Perl v5.8.3 - Fri Jul 11 14:54:19 UTC 2008 Site configuration information for perl v5.8.3: Configured by abuild at Fri Jul 11 14:47:00 UTC 2008. Summary of my perl5 (revision 5.0 version 8 subversion 3) configuration: Locally applied patches: @INC for perl v5.8.3: Environment for perl v5.8.3: |
From mbmiller+m@gmail.comHello Perl Debuggers-- One of our supercomputing staff looked into this and the origin of the Mike Date: Fri, 5 Mar 2010 11:05:35 -0600 Hi Mike, I took a look into this and basically perl thinks you have an infinite From pp_hot.c in the perl src: I32 iters = 0; maxiters = 2 * slen + 10; /* We can match twice at each if (iters++ > maxiters) So assuming signed ints, slen can be (2^31 - 10) / 2 which is exactly what To do what you want to do, something like this should be a (faster) work perl -pi -e 'print "ID " if $. == 1' foo.txt Mark On Thu, 4 Mar 2010, perlbug-followup@perl.org wrote:
|
From mbmiller+m@gmail.comHello PerlBug People-- We never received a reply to this message from a month ago. We are Thanks. Mike On Fri, 5 Mar 2010, Mike Miller wrote:
Please note that I am using a new email address: mbmiller+m@gmail.com. |
From @iabynOn Mon, Apr 05, 2010 at 03:05:39PM -0500, Mike Miller wrote:
Sorry that no-one replied. We tend to have a lot of bugs reports and not -- |
|
The RT System itself - Status changed from 'new' to 'open' |
From @ikegamiOn Tue, Apr 6, 2010 at 7:32 AM, Dave Mitchell <davem@iabyn.com> wrote:
Seems the OP actually talks about an infinite loop when maxiters = 2 * slen + 10; overflows. It's hard to tell because he didn't include any code or any |
From @ikegamiOn Tue, Apr 6, 2010 at 11:48 AM, Mike Miller
My apologies, I was looking at a reply thinking it was the OP. |
From mbmiller+m@gmail.comOn Tue, 6 Apr 2010, Eric Brine wrote:
What sort of code are you missing? My original post is below. It Thanks. Mike Date: Thu, 4 Mar 2010 15:28:47 -0600 (CST) This bug is highly replicable. See the attached perlbug -d output, but we The problem is that perl fails when line length approaches 2^30 characters $ perl -pi -e 'BEGIN{undef $/} ; s/\A/ID /' foo.txt ...with this result: $ ls -l foo.txt That is, zero bytes remain from a 1.5 GB file. (I know that I should have $ head -c 1073741818 foo.txt | perl -pe 's/./AB/' > bar.txt That did not fail, so I tried this: $ head -c 1073741819 foo.txt | perl -pe 's/./AB/' > bar.txt But that failed (with the same "Substitution ... chunk 1." message as Of course, 2^30 = 1073741824, which happens to be exactly 5 bytes greater cc: three people who helped me figure this out Mike -- |
From mbmiller+m@gmail.comFlags: This perlbug was built using Perl v5.8.3 - Fri Jul 11 14:54:19 UTC 2008 Site configuration information for perl v5.8.3: Configured by abuild at Fri Jul 11 14:47:00 UTC 2008. Summary of my perl5 (revision 5.0 version 8 subversion 3) configuration: Locally applied patches: @INC for perl v5.8.3: Environment for perl v5.8.3: |
Migrated from rt.perl.org#73306 (status was 'open')
Searchable as RT73306$
The text was updated successfully, but these errors were encountered: