Permalink
Commits on Aug 10, 2016
  1. v0.11

    avar committed Aug 10, 2016
        - Ever since 0.09 we haven't been using the XS version at all.
    
          The reason for this was that due to our intaction with
          Dist::Zilla we wouldn't define a $VERSION at BEGIN time, and
          would default to 0.01 as a placeholder, DynaLoader would
          silently fail to load it, and we'd always use the pure-perl
          version.
    
        - We now warn on compile-time if we can't load the XS version.
    
        - Document how to use the pure-perl version directly, and make it
          use the same API, for anyone who's interested in that, e.g. for
          testing.
    
        - Update the benchmark v.s. Linux::Smaps for its latest version &
          ours. We're proportionally faster now than when I initially
          released this.
Commits on Dec 1, 2014
  1. v0.10

    avar committed Dec 1, 2014
        - Fix test failures under non-C locales, I was matching against
          strerror(), just don't do that. See RT #95936.
Commits on Jul 1, 2012
  1. v0.09

    avar committed Jul 1, 2012
        - CPAN RT #78029: fix a memory leak reported by Norbert
          Buchmüller.
  2. Fix a memory leak created by not calling sv_2mortal()

    avar committed Jul 1, 2012
    See perlxs(1) for why this needs to be done, I wasn't able to come up
    with a reliable test case for this though, but this should fix the
    leak.
Commits on Jul 20, 2011
  1. v0.08

    avar committed Jul 20, 2011
        - The 0.07 releas had a bugfix in the XS API. It now actually
          works as documented when given an argument.
    
        - Minor documentation nits.
  2. Changes for 0.08

    avar committed Jul 20, 2011
  3. v0.07

    avar committed Jul 20, 2011
        - Check errno right after fopen()
  4. Open /proc/$ARG/smaps under the C interface, not just $ARG

    avar committed Jul 20, 2011
    We didn't prepend /proc/ and append /smaps to the argument you were
    giving under the C interface.
    
    There's a potential buffer overflow here. But process names will never
    be this long, and hopefully nobody's stupid enough to expose this
    module to hostile remote users.
    
    Reported-by: David Schmidt <davewood@gmx.at>
  5. Tiny.xs: (void)hv_store, not hv_store

    avar committed Jul 20, 2011
    This eliminates some warnings.
  6. Tiny.xs: explain odd code

    avar committed Jul 20, 2011
  7. Check errno right after fopen()

    avar committed Jul 20, 2011
Commits on Apr 3, 2011
  1. v0.06

    avar committed Apr 3, 2011
        - We now parse smaps with XS code, and fall back to PP if we can't
          can_cc().
  2. benchmark modified for new code

    avar committed Apr 3, 2011
  3. t/all.t: test that we had some memory

    avar committed Apr 3, 2011
    If I screw up some of the C code this'll catch it.
  4. An XS version of this

    avar committed Mar 31, 2011
  5. a better gitignore file

    avar committed Apr 3, 2011
Commits on Mar 30, 2011
  1. v0.05

    avar committed Mar 30, 2011
        - Add a =cut before the 1; in Tiny.pm
  2. Add a =cut before the 1; in Tiny.pm

    avar committed Mar 30, 2011
    Reported-by: Torsten Förtsch <torsten.foertsch@gmx.net>
  3. v0.04

    avar committed Mar 30, 2011
        - Update documentation
Commits on Mar 29, 2011
  1. v0.03

    avar committed Mar 29, 2011
        - Add more information to our die() messages
  2. v0.02

    avar committed Mar 29, 2011
        - Default to /proc/self/smaps, not /proc/$$/smaps
  3. get rid of a scope

    avar committed Mar 29, 2011
    This should be slightly faster than creating the scope.
  4. Default to /proc/self/smaps, not /proc/$$/smaps

    avar committed Mar 29, 2011
    This is presumably faster, or at least more intuative.
  5. v0.01

    avar committed Mar 29, 2011
        - Initial buggy version
  6. add gitignore

    avar committed Mar 29, 2011
  7. get_smaps_summary: optimize the routine

    avar committed Mar 29, 2011
    See the "benchmarks" branch, but basically we used to be "read" here,
    and are now "substr_etc":
    
        $ perl -Ilib bench.pl
                     Rate      slurp       read    split_n substr_etc
        slurp       867/s         --        -7%       -13%       -17%
        read        934/s         8%         --        -6%       -11%
        split_n     993/s        14%         6%         --        -5%
        substr_etc 1044/s        20%        12%         5%         --