Revision history for Devel::Size::Report

2008-04-06 v0.13 Tels 147 tests
  * require at least Perl 5.006 (Devel::Size doesn't compile on 5.005)
  * fix failing tests  on v5.10 and version

2008-03-28 v0.12 Tels 148 tests
  * fix failing tests (due to bugfix in Devel::Size, thanx Wolfgang Schemmel!)
  * add POD tests and POD coverage tests
  * fix POD errors revealed by coverage testing
  * require at least Perl 5.005002

2006-09-09 v0.11 Tels 146 tests
  * clean up Makefile.PL
  * add a SF_MAGIC() flag (not yet used)
  * fix failing tests under 64 bit (need to add more than 7 bytes)
  * fix undef warning from blessed()
  * "use constant" instead of "sub () ..."

Older revision history:

2004-01-06 v0.01 Tels 42 tests (not released)
  * first release with:
     report_size(), track_size(), element_type(), entries_per_element() 

2004-01-09 v0.02 Tels 42 tests
  * fixed bug in calculating overhead with contained hashes
  * added example/psize

2004-01-10 v0.03 Tels 49 tests
  * psize: don't print out added "use" statements
  * report_size(): print overhead in %, too (Thanx nadim!)
  * knows about CODE, REGEXP, LVALUE and Scalar refs and prints them correctly
  * stop at cyclic references (e.g. don't loop forever)
  * disable Devel::Size warnings (about incomplete CV etc, thanx nadim!)
  * drop the "Key " from the output, it is rendundand (thanx nadim!)
  * report_size: bytes => '' in $options works correctly now (thanx nadim!)

2004-01-10 v0.04 Tels 55 tests
  * fixed the bug that sometimes looking at a scalar again after looking at
    a longer scalar reported the wrong (bigger) size. $pre = shift vs. $_[0]
    usage in total_size() was the culprit (slot reusing).  Thanx Sadahiro!
  * Size report outputs version of Devel::Size::Report, too
  * refactored tests into three files, added some more tests
  * use Scalar::Util to get reftype and refaddr
  * added option 'addr' to report_size() to print elem address
  * example/psize takes options as second string argument

2004-01-13 v0.05 Tels 76 tests
  * references to ARRAY, HASH are handled properly
  * references to SCALARs are now reported likewise than ARRAY/HASH refs
  * fixed indendation bug with hash keys pointing to SCALARs
  * more tests, especially for the addr => 1 feature
  * psize takes options as single-letter abbrev., too
  * added option "class" to optionally show classnames

2004-10-18 v0.06 Tels 76 tests
  * fixed example/ - duh!
  * some small cleanups in comments in
  * amended TODO
  * scalar references (\0) are now reported with their correct size

2004-10-23 v0.07 Tels 100 tests
  * rename S_KEY and S_REF to SF_KEY and SF_REF
  * document exportable methods
  * small comments added
  * The "Total" line now includes the number of elements
  * added options 'terse' (supresss gory details) and 'summary' (include
    per-class overview at end of report) to report_size()
  * added S_VSTRING
  * added SF_WEAK, SF_RO, SF_DUAL
  * simplify test code by using like()
  * added 'te' and 's' option to psize
  * added more examples to example/

2004-11-10 v0.08 Tels 116 tests (released 2005-02-27)
  * confess() if options to report_size() is not a hash
  * option debug: print time and memory used to make stats
  * streamline element_type()
  * report_size(): inline element_type() for a tad more speed
  * use undef instead of '' for unblessed things to save a bit of memory
  * more tests (element_type, type, options, double ref, cycles)
  * return blessed() from _type() and use it to avoid second call to blessed()
  * fix bug with double scalars in a hash (f.i. via Array-RefElem)
  * use Array-RefElem to save about 50% of the %SEEN hash (which represents
    about 3-6% of the memory when generating reports on huge structures)

2005-02-27 v0.09 Tels 122 tests
  * Don't need to track simple scalars with a REFCNT of 1 in the SEEN hash,
    this reduces the memory overhead quite a bit 
  * add option "track_doubles", that only tracks double/shared hash keys and
    array elemts if requested. Default is off, this saves memory.
  * Rewrote _track_size() to modify a global array instead of passing the
    results back via the stack. This avoids that the entire results array
    exists temporarily twice when returning the results and speeds up anything
    by a factor of two to three :o)
  * add a test for "Glob"
  * streamline array and hash member tracking (for simple elements)
  * streamline type()
  * add track_sizes() and hide_tracks()

2005-03-04 v0.10 Tels 145 tests
  * amend README, SYNOPSIS, example/
  * psize: support for option "d", print help if no arguments are given
  * indend => indent (Thanx Jim Cromie!)
  * sizes for array/hash refs were missing the ref itself (16 bytes on 32 bit
    systems, more on 64 bit). This also makes "[]" and "[ 1, [] ]" report
    consistent sizes for the inner array.
  * compute correct size for circular references
  * rewrote the size-gathering code to use size() instead of total_size(),
    this makes it much faster, especially for deep structures and ones with
  * track_size() and track_sizes() also need to honor option "doubles"
  * does no longer warn about "deep recursion"