7

I can't figure out what's going on here. Where does the 8 below come from?

Time::HiRes provides an overload of stat which expands the times to have high-resolution (which is supported on my system).

$ perl  -MTime::HiRes -e 'print +(stat("foo"))[8], "\n"'              # V1
1322915623
$ perl  -MTime::HiRes=stat -e 'print +(stat("foo"))[8], "\n"'         # V2
8
$ perl  -MTime::HiRes=stat -e '@a = stat("foo"); print $a[8], "\n"'   # V3
1322915623

That particular file doesn't have a high-resolution timestamp, but that's not the mystery: the mystery is V2, which prints 8. In fact, it always prints the number in square brackets.

The obvious answer, it parses differently, does not seem correct:

$ perl -MO=Deparse -MTime::HiRes -e 'print +(stat("foo"))[8], "\n"'         # V1
use Time::HiRes;
print((stat 'foo')[8], "\n");
-e syntax OK
$ perl -MO=Deparse -MTime::HiRes=stat -e 'print +(stat("foo"))[8], "\n"'    # V2
use Time::HiRes (split(/,/, 'stat', 0));
print((stat 'foo')[8], "\n");
-e syntax OK

They deparse the same (other than the different option to use Time::HiRes).

It works fine if I use my own function in similar syntax, and I can't get the "wrong" answer even if I return something silly from my function:

$ perl -e 'sub bar() { return qw(a b c d e f g h i j) }; print +(bar)[8], "\n"'
i
$ perl -e 'sub bar() { return undef }; print +(bar)[8], "\n"'

$

This is Debian's perl package, version 5.14.2-5. I get the same results with 5.10.1-17squeeze2.

How does V2, above, produce 8? Am I misunderstanding Perl syntax in some way, or do I just need to file a bug report?

edit: As @cjm says, this is a bug. It has been fixed in Time-HiRes-1.9725 according to the report.

derobert
  • 49,731
  • 15
  • 94
  • 124

2 Answers2

3

It's definitely a bug, although I'm not sure whether it's in core Perl or in Time::HiRes. I get the same results with Perl 5.14.2 on Gentoo (and also with 5.8.9 and 5.10.0). Have you noticed that it doesn't matter what you put in the subscript?

$ perl -MTime::HiRes=stat -e 'print +(stat("foo"))[215.4], "\n"' 
215.4
$ perl -MTime::HiRes=stat -e 'print +(stat("foo"))["bar"], "\n"'
bar

I'd probably report it in Time::HiRes first.

Note: While they deparse the same, they do generate different opcodes (due to the difference between calling a built-in and a user-defined sub):

$ perl -MO=Concise -MTime::HiRes -e 'print +(stat("foo"))[8], "\n"' 
c  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter ->2
2     <;> nextstate(main 271 -e:1) v:{ ->3
b     <@> print vK ->c
3        <0> pushmark s ->4
9        <2> lslice lK/2 ->a
-           <1> ex-list lK ->6
4              <0> pushmark s ->5
5              <$> const(IV 8) s ->6
-           <1> ex-list lK ->9
6              <0> pushmark s ->7
8              <1> stat lK/1 ->9
7                 <$> const(PV "foo") s ->8
a        <$> const(PV "\n") s ->b
-e syntax OK

$ perl -MO=Concise -MTime::HiRes=stat -e 'print +(stat("foo"))[8], "\n"' 
e  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter ->2
2     <;> nextstate(main 271 -e:1) v:{ ->3
d     <@> print vK ->e
3        <0> pushmark s ->4
b        <2> lslice lK/2 ->c
-           <1> ex-list lK ->6
4              <0> pushmark s ->5
5              <$> const(IV 8) s ->6
-           <1> ex-list lK ->b
6              <0> pushmark s ->7
a              <1> entersub[t1] lKS/TARG,1 ->b
-                 <1> ex-list lK ->a
7                    <0> pushmark s ->8
8                    <$> const(PV "foo") sM ->9
-                    <1> ex-rv2cv sK ->-
9                       <$> gv(*stat) s ->a
c        <$> const(PV "\n") s ->d
-e syntax OK
cjm
  • 61,471
  • 9
  • 126
  • 175
  • Thank you, bug reported https://rt.cpan.org/Public/Bug/Display.html?id=72926 ... you may want to add your `$Time::HiRes::VERSION`s, I don't have them. – derobert Dec 04 '11 at 02:40
0

I have never used the command line execution so I speak to oddities in its workings.

I have seen unexpected results when using a subscript on a function that should return an array.

$y = localtime()[5];    # failed for me (I forget just how)

but

$y = (localtime())[5];  # worked fine

This suggests to me a bug in (my implementation of) Perl. A better test might be to try it in an actual script:

use Time::HiRes qw(stat);

my @x = stat("foo");

print $x[8],"\n";

I am using ActiveState's Perl on WinXP so my results may not be the same. Still, I think it may be useful to try it in very simple code to see what it does.

jon
  • 11
  • 1
  • 2
    This is not a bug: Perl's [list indexing](http://perldoc.perl.org/perldata.html#List-value-constructors) syntax is `(LIST)[INDEX]`. The parentheses are required. (The behavior observed by derobert and cjm, however, _is_ a bug.) – Ilmari Karonen Feb 13 '12 at 22:03