APR::PerlIO -- Perl IO layer for APR

  # under mod_perl
  use APR::PerlIO ();

  sub handler {
      my $r = shift;

      die "This Perl build doesn't support PerlIO layers"
          unless APR::PerlIO::PERLIO_LAYERS_ARE_ENABLED;

      open my $fh, ">:APR", $filename, $r->pool or die $!;
      # work with $fh as normal $fh
      close $fh;

      return Apache2::Const::OK;
  }

  # outside mod_perl
  % perl -MAPR -MAPR::PerlIO -MAPR::Pool -le \
  'open my $fh, ">:APR", "/tmp/apr", APR::Pool->new or die "$!"; \
   print $fh "whoah!"; \
   close $fh;'

CWAPR::PerlIO implements a Perl IO layer using APR's file manipulation API internally.

Why do you want to use this? Normally you shouldn't, probably it won't be faster than Perl's default layer. It's only useful when you need to manipulate a filehandle opened at the APR side, while using Perl.

Normally you won't call open() with APR layer attribute, but some mod_perl functions will return a filehandle which is internally hooked to APR. But you can use APR Perl IO directly if you want.

Not every Perl will have full CWAPR::PerlIO functionality available.

Before using the Perl IO APR layer one has to check whether it's supported by the used APR/Perl build. Perl 5.8.x or higher with perlio enabled is required. You can check whether your Perl fits the bill by running:

  % perl -V:useperlio
  useperlio='define';

It should say define.

If you need to do the checking in the code, there is a special constant provided by CWAPR::PerlIO, which can be used as follows:

  use APR::PerlIO ();
  die "This Perl build doesn't support PerlIO layers"
      unless APR::PerlIO::PERLIO_LAYERS_ARE_ENABLED;

Notice that loading CWAPR::PerlIO won't fail when Perl IO layers aren't available since CWAPR::PerlIO provides functionality for Perl builds not supporting Perl IO layers.

See Prerequisites.

Most of the API is as in normal perl IO with a few nuances listed in the following sections.

META: need to rework the exception mechanism here. Current success in using errno ($!) being set (e.g. on open()) is purely accidental and not guaranteed across all platforms and functions. So don't rely on $!. Will use CWAPR::Error for that purpose. Open a file via APR Perl IO layer.

  open my $fh, ">:APR", $filename, $r->pool or die $!;

ret: ( integer )

success or failure value (boolean).

since: 2.0.00

Sets CW$fh's position, just like the CWseek() Perl call:

  seek($fh, $offset, $whence);

If CW$offset is zero, CWseek() works normally.

However if CW$offset is non-zero and Perl has been compiled with with large files support (CW-Duselargefiles), whereas APR wasn't, this function will croak. This is because largefile size CWOff_t simply cannot fit into a non-largefile size CWapr_off_t.

To solve the problem, rebuild Perl with CW-Uuselargefiles. Currently there is no way to force APR to build with large files support.

since: 2.0.00

The C API provides functions to convert between Perl IO and APR Perl IO filehandles.

META: document these

mod_perl 2.0 documentation. The perliol(1), perlapio(1) and perl(1) manpages.

mod_perl 2.0 and its core modules are copyrighted under The Apache Software License, Version 2.0.

The mod_perl development team and numerous contributors.