BrowserUk has asked for the wisdom of the Perl Monks concerning the following question:

Ignore this. The demo code is broken and the description describes the wrong thing.

If you run this one-liner and monitor the process memory usage after each of the 3 stages, it remains constant (as you might expect):

perl -e"$x=chr(0);$x x=512*1024**2; <>;$r=\substr($x,256x1024**2); <>;
+substr($$r,256x1024**2-1,1)='x'; <>"

[download]

However, if you do the same with this one-liner, which creates the same size buffer, takes the same lvalue ref and modifies the same byte of the original string via that lvalue ref, this time the memory usage temporarily triples, before falling back to double:

perl -e"$x=chr(0);$x x=512*1024**2; <>;$r=\substr($x, 256x1024**2); <>
+;vec($$r,2**31-1,1)=1; <>"

[download]

The question is why? Is it a bug that should be reported? Can the demo be simplified?

With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.

The start of some sanity?

Replies are listed 'Best First'.
Re: Tracking down an Lvalue bug?
by Anonymous Monk on Mar 17, 2012 at 21:42 UTC

    Hmm, are you sure you want 256x1024**2 in that substr? I think this is increasing the number beyond what you intended , which doesn't matter with substr because you're replacing all that trailing string with one char, but does matter with vec because nothing is cut-off

    Adjusted for my machine I run

    A perl -e"$x=chr(0);$x x=5*1024**2; <>;$r=\substr($x, 2*1024**2); <>;vec($$r,2*1024**2-1,1)=1; <>"

    B perl -e"$x=chr(0);$x x=5*1024**2; <>;$r=\substr($x, 2*1024**2); <>;substr($$r,2*1024**2-1,1)='x'; <>"

    And I consistently get results like
    AB
    7,760 k
7,776 k
10,876 k

    7,760 k
7,776 k
10,880 k

[reply]
[d/l]
[select]

      Grr. That'll teach me to try reduce my testcase to a one-liner :(

      This 256x1024**2 was intended to be 256*1024**2. Surprisingly (or not) neither strict nor warnings catch the typo.

      And when corrected the dynamics of the bug change substantially.

      Thanks for calling me on it. Back to the drawing board!

      With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
      Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
      "Science is about questioning the status quo. Questioning authority".
      In the absence of evidence, opinion is indistinguishable from prejudice.

      The start of some sanity?
[reply]
[d/l]
[select]

        Back to the drawing board!

        What are you trying to do?

[reply]

Back to Seekers of Perl Wisdom