Re^5: Dangerous Names

Ah, I see what you mean. ~~You're wrong about it dying, though. It's only warning.~~

I don't find it unusual since I've previously heard that { $a cmp $b }, { $a <=> $b }, { $b cmp $a } and { $b <=> $a } are handled specially to avoid having to call a Perl sub repeatedly.

>perl -MO=Concise -e"@a = sort { $a <=> $b } @a"
8  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter ->2
2     <;> nextstate(main 2 -e:1) v ->3
-     <1> ex-aassign vKS/64 ->8
-        <1> ex-list lK ->-
3           <0> pushmark s ->4
7           <@> sort lK/INPLACE,NUM ->8
4              <0> pushmark s ->5
6              <1> rv2av[t7] lKRM*/1 ->7
5                 <#> gv[*a] s ->6
-        <1> ex-list lK ->-
-           <0> ex-pushmark s ->-
-           <1> ex-rv2av lKRM*/1 ->-
-              <0> ex-gv s ->-
-e syntax OK

>perl -MO=Concise -e"@a = sort { 1; $a <=> $b } @a"
8  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter ->2
2     <;> nextstate(main 2 -e:1) v ->3
-     <1> ex-aassign vKS/64 ->8
-        <1> ex-list lK ->-
3           <0> pushmark s ->4
7           <@> sort lKS*/INPLACE ->8
4              <0> pushmark s ->5
-              <1> null sK/1 ->5
-                 <1> ex-leave sKP ->-
-                    <0> enter s ->-
-                    <;> nextstate(main 1 -e:1) v ->-
-                    <0> ex-const v ->-
-                    <;> nextstate(main 1 -e:1) v ->-
-                    <2> ncmp[t5] sK/2 ->(end)
-                       <1> ex-rv2sv sK/1 ->-
-                          <#> gvsv[*a] s ->-
-                       <1> ex-rv2sv sK/1 ->-
-                          <#> gvsv[*b] s ->-
6              <1> rv2av[t7] lKRM*/1 ->7
5                 <#> gv[*a] s ->6
-        <1> ex-list lK ->-
-           <0> ex-pushmark s ->-
-           <1> ex-rv2av lKRM*/1 ->-
-              <0> ex-gv s ->-
-e syntax OK
[download]

While I don't find it unusual, I do think the difference in behaviour is a bug.

Comment on Re^5: Dangerous Names Select or Download Code

Replies are listed 'Best First'.
Re^6: Dangerous Names by samtregar (Abbot) on Dec 17, 2008 at 22:13 UTC
Oh, right, I have die on warnings running in this context I think. UPDATE: No, actually it does die. This is not a warning. Observe: `$ perl -e '@names = (["Nan Tregar"], ["Sam Tregar"]); eval { @names + = sort { $a->[0] <=> $b->[0] } @names; }; print "ERROR WAS: $@"' ERROR WAS: Sort subroutine didn't return a numeric value at -e line + 1.` [download] -sam	[reply] [d/l]
Re^7: Dangerous Names by ikegami (Patriarch) on Dec 17, 2008 at 22:42 UTC
Oops, I guess I flubbed my test code. I submitted a bug report.	[reply]


Don't ask to ask, just ask
	PerlMonks