summaryrefslogtreecommitdiff
path: root/Bugs/lcsort
diff options
context:
space:
mode:
Diffstat (limited to 'Bugs/lcsort')
-rw-r--r--Bugs/lcsort37
1 files changed, 37 insertions, 0 deletions
diff --git a/Bugs/lcsort b/Bugs/lcsort
new file mode 100644
index 0000000000..2cbe8613ad
--- /dev/null
+++ b/Bugs/lcsort
@@ -0,0 +1,37 @@
+Article 20341 of comp.lang.perl:
+Path: netlabs!news.cerf.net!mvb.saic.com!MathWorks.Com!news.kei.com!hookup!swrinde!gatech!taco.cc.ncsu.edu!taco.cc.ncsu.edu!setzer
+From: setzer@math.ncsu.edu (William Setzer)
+Newsgroups: comp.lang.perl
+Subject: perl5a6 bug: sort sub @ary
+Date: 30 Mar 94 13:49:20
+Organization: The Chebychev Church of Probability
+Lines: 23
+Distribution: world
+Message-ID: <SETZER.94Mar30134920@math.math.ncsu.edu>
+NNTP-Posting-Host: math.ncsu.edu
+
+Ok, here's a weird one. Using the precompiled perl5a6 binary for the
+sun, the following dumps core:
+----
+@S = (0, 1, 2, 3);
+for ("A", "D", "b", "c") { $foo{$i++} = $_; }
+# for (sort { lc($foo{$a}) cmp lc($foo{$b}) } @S) { print $foo{$_}, "\n"; }
+for (sort Alpha @S) { print $foo{$_}, "\n"; }
+
+sub Alpha { lc($foo{$a}) cmp lc($foo{$b}); }
+----
+[Output]
+Segmentation fault (core dumped)
+----
+However, if you comment out the "sort Alpha" and uncomment the "sort
+{lc(...) cmp lc(...) }" line, everything works as expected. Also note
+that if you uncomment both "sort"s, everything also works as expected.
+Do-do-do-do [twilight zone music :-].
+
+William
+
+PS: Does anyone actually care about these bug reports? I thought I
+ remember Larry saying that he preferred them in c.l.p., but it
+ wouldn't be the first time my memory has fritzed.
+
+