CHANGES

2.037 22 June 2011

• IO::Uncompress
  • get globmapper tests working on VMS [RT# 68926]
• IO::Uncompress::Unzip
  • Fixed limitation where Streamed Stored content was not supported.

2.036 18 June 2011

• IO::Compress::Zip & IO::Uncompress::Unzip
  • Added support for LZMA (method 14) compression/uncompresion.
• IO::Compress::Unzip
  • Fixed CRC issue when compression is Store or Bzip2 and Strict option is set.
• IO::Compress::Zip
  • Fixed Zip64 issue where the content size is exactly 0xFFFFFFFF

2.035 6 May 2011

• RT #67931: Test failure on Windows

2.034 2 May 2011

• Compress::Zlib
  • Silence pod warnings. [RT# 64876]
  • Removed duplicate words in pod.
• IO::Compress::Base
  • RT #56942: Testsuite fails when being run in parallel
  • Reduce symbol import - patch from J. Nick Koston
  • If the output buffer parameter passed to read has a value of undef, and Append mode was specified when the file was opened, and eof is reached, then the buffer paramer was left as undef. This is different from when Append isn't specified - the buffer parameter is set to an empty string.
  • There area couple of issues with reading a file that contains an empty file that is compressed. Create with -- touch /tmp/empty; gzip /tmp/empty. Issue 1 - eof is not true immediately. Have to read from the file

    to trigger eof. Issue 2 - readline incorrectly returns an empty string the first

    time it is called, and (correctly) undef thereafter. [RT #67554]

2.033 11 Jan 2011

• Fixed typos & spelling errors. [perl# 81816]

2.032 4 Jan 2011

• IO::Uncompress::Base
  • An input file that had a valid header, and so would allow creation of the uncompression object, but was then followed by corrupt data would trigger an infinite loop when using the input line oprator. [RT #61915]
• IO::Compress::Gzip
  • XFL default settings for max compression & fastest algorithm were the wrong way around. Thanks to Andrey Zholos for spotting this.
• IO::Compress::Base::Common
  • Fixed precedence problem in parameter parsing code.

2.030 22 July 2010

• IO::Compress::Zip
  • Updates to documentation.
  • Changes default value for ExtAttr on Unix to 0100644
• IO::Uncompress::Unzip Reworked the "Name" option and examples in the pod.
• IO::Uncompress::Base Fixed problem with nextStream not returning 0 when there is no next stream and Transparent is false.

2.027 24 April 2010

• Compress::Zlib Remove autoload code from Zlib.pm. [perl #74088]

2.026 7 April 2010

• IO::Uncompress::Zip
  • Some updates to IO::Compress::Zip documentation.
  • Fixed default setting for ExtAttr.

2.025 27 March 2010

• IO::Uncompress::Unzip The "Name" option wasn't documented.
• Allow zlib version check to be disabled by setting TEST_SKIP_VERSION_CHECK environment variable. [RT #54510]

2.024 7 January 2010

• Compress::Zlib Get memGunzip & memGzip to set $gzerrno [RT# 47283]
• Compress::Zlib Export memGunzip, memGzip and zlib_version on demand [RT# 52992]
• examples/io/anycat This sample was using IO::Uncompress::AnyInflate. Much better to use IO::Uncompress::AnyUncompress.

2.023 9 November 2009

• IO::Compress::AnyUncompress Added support for lzma_alone & xz.

2.022 9 October 2009

• IO::Compress - Makefile.PL Fix for core.

2.021 30 August 2009

• IO::Compress::Base.pm
  • Less warnnings when reading from a closed filehandle. [RT# 48350]
  • Fixed minor typo in an error message. [RT# 39719]
• Makefile.PL The PREREQ_PM dependency on Scalar::Util got dropped when IO-Compress was created in 2.017. [RT# 47509]
• IO::Compress::Zip.pm
  • Removed restriction that zip64 is only supported in streaming mode.
  • The "version made by" and "extract" fields in the zip64 end central record were swapped.
  • In the End Central Header record the "offset to the start of the central directory" will now always be set to 0xFFFFFFFF when zip64 is enabled.
  • In the End Central Header record the "total entries in the central directory" field will be set to 0xFFFF if zip64 is enabled AND there are more than 0xFFFF entries present.
• IO::Uncompress::Unzip.pm
  • Don't consume lots of memory when walking a zip file. This makes life more bearable when dealing with zip64.
• Compress::Zlib.pm
  • documented that memGunzip cannot cope with concatenated gzip data streams.
• Changed test harness so that it can cope with PERL5OPT=-MCarp=verbose [RT# 47225]
• IO::Compress::Gzip::Constants.pm
  • GZIP_FEXTRA_MAX_SIZE was set to 0xFF. Should be 0xFFFF. This issue came up when attempting to unzip a file created by MS Office 2007.

2.020 3 June 2009

• IO::Uncompress::Base.pm
  • Fixed problem with LimitOutput where a call to uncompress created more uncompressed output, but didn't consume any of the input buffer. The symptom is the underlying compression library (zlib or bzip2) thinks the input stream is corrupt. [RT #46582]

2.019 4 May 2009

• IO::Uncompress::Adapter::Bunzip2
  • Fixed problem with EOF check.

2.018 3 May 2009

• IO::Uncompress::Bunzip2
  • The interface to Compress-Raw-Bzip2 now uses the new LimitOutput feature. This will make all of the bzip2-related IO-Compress modules less greedy in their memory consumption.
• IO::Compress::Zip
  • Fixed exTime & exUnix2
  • Fixed 'Use of uninitialized value in pack' warning when using ZIP_CM_STORE.

2.017 30 March 2009

• Merged IO-Compress-Base, IO-Compress-Bzip2, IO-Compress-Zlib & Compress-Zlib into IO-Compress.
• The interface to Compress-Raw-Zlib now uses the new LimitOutput feature. This will make all of the zlib-related IO-Compress modules less greedy in their memory consumption.
• Removed MAN3PODS from Makefile.PL
• A few changes to get the test harness to work on VMS courtesy of Craig. A. Berry.
• IO::Compress::Base & IO::Uncompress::Base Downgraded some croaks in the constructors to just set $! (by letting the code attempt to open a file and fail). This makes the behavior more consistent to a standard open. [RT #42657]
• IO::Uncompress::Base Doing a seek with MultiStream could drop some of the uncompressed data. Fixed.
• IO::Compress::Zip
  • Fixed problem with the uncompressed & uncompressed fields when zip64 is enabled. They were set to 0x0000FFFF instead of 0xFFFFFFFF. Also the ZIP64 extra field was 4 bytes short. Problem spotted by Dino Chiesa.
• IO::Uncompress::Unzip
  • use POSIX::mktime instead of Time::Local::timelocal to convert the zip DOS time field into Unix time.
• Compress::Zlib
  • Documented Compress::Zlib::zlib_version()

2.015 3 September 2008

• Makefile.PL Backout changes made in 2.014

2.014 2 September 2008

• Makefile.PL Updated to check for indirect dependencies.

2.013 18 July 2008

• IO::Compress::Base
  • Allow IO::Compress::Base::Parameters::parse to accept an IO::Compress::Base::Parameters object.

2.012 15 July 2008

• IO::Compress::Base
  • Silenced an uninitialised value warning when reading a line at a time from a zip file where the content uses ZIP_CM_STORE. [Problem spotted & fixed by Jeff Holt]
• IO::Compress::Base & IO::Uncompress::Base
  • local-ise $!, $? et al in the DESTROY methods.

2.011 17 May 2008

• IO::Compress::Base
  • Fixed problem that prevented the creation of a zip file that contained more than one compression method.
• IO::Compress::Base::Common
  • The private Validator class in this module clashes with another CPAN module. Moved Validator into the IO::Compress::Base::Common namespace. [RT #35954]
• IO::Uncompress::Unzip
  • Print an error message if the zip file contains a member compressed with bzip2 and IO::Uncompress::Bunzip2 is not available.
  • Could not cope with mixed compression zip files. For example a zip file that contains both STORED and DEFLATED content. [RT #35573]

2.010 5 May 2008

• Fixed problem that meant Perl 5.10 could not upgrade this module. [RT #35342 & 35341]

2.009 20 April 2008

• Removed the alpha status from File::GlobMapper
• IO::Compress::Base When writing output never output a zero length buffer. Done to improve interoperability with other tied filenandle modules.
• Changed IO::Uncompress::Base to not use the offset parameter of the read method when reading from a filehandle.

  The object returned from Net::FTP::retr implements a non-standard read method. The third parameter is used for a timeout value rather than an offset. [rt.cpan#33231]

• Changed IO::Uncompress::Base to not use the eof method when reading from a filehandle.

  The object returned from Net::FTP::retr implements both the read method and the eof method. Unfortunately the implementation of the read method uses non-buffered IO (by using sysread) while the eof method uses buffered IO. Mixing buffered and non-buffered IO results in data corruption.

• IO::Compress::Zip
  • Added exUnix2 option to allow storing of UID & GID.
  • When running on a Unix derivative the ExtAttr option now defaults to the equivalent of 0666. For all other systems the default remains 0.
• Compress::Zlib
  • Minor documentation issue with flush. [rt.cpan.org #31446]

2.008 2 November 2007

• Minor documentation changes in README
• t/compress/truncate.pl EBCDIC Cleanup.
• IO::Compress::Gzip::Constants.pm Tidied up the character classes used to defined invalid FNAME & FCOMMENT fields for EBCDIC.
• Compress::Zlib lib/Compress/Zlib.pm -- 1.x Backward Compatibility issues gzclose - documented return value was wrong, should be 0 for ok. gzflush - return value didn't match 1.x, should return 0 if ok. [rt.cpan.org #29215] and Debian bug #440943 http://bugs.debian.org/440943

2.006 1 September 20007

• Makefile.PL Added INSTALLDIRS directive to install as a core module when built on a perl >= 5.9.
• IO::Uncompress::RawDeflate
  • Fixed export problem - "$RawDeflateError" and "rawdeflate" were not being exported with ":all".
• Compress::Zlib
  • t/03zlib-v1.t Fixed crc32 and adler32 tests in to remove ascii assumption.
  • lib/Compress/Zlib.pm Make gzreadline not support $/, just like in Compress::Zlib 1.x Folk who want $/ with readline support can get it in IO::Uncompress::Gunzip. [rt.cpan.org #28663] and Debian bug #435656 http://bugs.debian.org/435656

2.005 18 June 2007

• Stephen Turner reported a problem when using IO::Uncompress::Gunzip with XML::Parser. Turns out there were two issues.

  Firstly an IO::Uncompress object isn't an IO::Handle. It is now.

  Secondly the implementation of "read" wasn't honouring this

SCALAR will be grown or shrunk to the length actually read.

        In particular it didn't do the right thing on EOF.
        This has been fixed.

      * IO::Compress::Gzip & IO::Uncompress::Gunzip

        - RFC1952 says that the FNAME & FCOMMENT header fields must be ISO
          8859-1 (LATIN-1) characters. The code can optionally police this.
          Added a fix for this logic when running on EBCDIC.

      * Compress::Zlib 
        Added info about removing Compress::Zlib version 1, before
        installing version 2. 

2.004 3 March 2007

• Made seek less wasteful of memory.
• IO::Compress::Zip
  • Added Zip64 documentation.
  • Fixed extended timestamp. Creation time isn't available in Unix so only store the modification time and the last access time in the extended field.
  • Fixed file mode.
  • Added ExtAttr option to control the value of the "external file attributes" field in the central directory.
  • Added Unix2 extended attribute ("Ux"). This stores the UID & GID.
• IO::Compress::Gzip
  • Fixed 050interop-gzip.t for Windows
• IO::Compress::Bzip2
  • Fixed 050interop-bzip2.t for Windows
• Compress::Zlib
  • rewrote memGzip using IO::Compress::Gzip::gzip

2.003 2 January 2007

• Added explicit version checking

2.002 29 December 2006

• Documentation updates.
• Added IO::Handle to the ISA test in isaFilehandle
• Add an explicit use_ok test for Scalar::Util in the test harness. The error message reported by 01misc implied the problem was somewhere else. Also explicitly check that 'dualvar' is available.
• Compress::Zlib
  • Fix append mode with gzopen. rt-cpan.org 24041
  • Allow gzopen to read from and write to a scalar reference.

2.001 1 November 2006

• Remove beta status.

2.000_14 26 October 2006

• IO::Uncompress::Base Added support for $/ in record mode
• IO::Uncompress::Base The readline interface was substantially slower than the 1.x equivalent. This has now been sorted. Thanks to Andreas J. Koenig for spotting the problem.
• IO::Uncompress::AnyUncompress Added IO::Uncompress::Lzf to the list of supported uncompressors.
• IO::Uncompress::Base Added TrailingData to one-shot interface.
• IO::Uncompress::AnyUncompress Remove raw-deflate (RFC1951) from the default list of compressors to check. It can still be included if the new RawInflate parameter is supplied. This change was made because the only way to tell if content is raw-deflate is to attempt to uncompress it - a few false positives have popped up recently, which suggests that auto-detecting raw deflate is far from perfect. The equivalent change has been made to IO::Uncompress::AnyInflate. [Core patch #28445]
• Don't check that filehandles are writable. It would seem that "-w *STDOUT" on windows returns false. [Core Patch #28415]
• IO::Uncompress::Deflate Beefed up the magic signature check. Means less false positives when auto-detecting the compression type.
• IO::Uncompress::UnZip Tighten up the zip64 extra field processing to cope with the case wheere only some of the local header fields are superseded.
• IO::Uncompress::AnyInflate Remove raw-deflate (RFC 1951) from the default list of compressors to check. It can still be included if the new RawInflate parameter is supplied. This change was made because the only way to tell if content is raw-deflate is to attempt to uncompress it - a few false positives have popped up recently, which suggests that auto-detecting raw deflate is far from perfect. The equivalent change has been made to IO::Uncompress::AnyUncompress. [Core patch #28445]

2.000_13 20 June 2006

• Store compress & uncompressed sizes as 64-bit.
• For one-shot uncompression, like this

unzip "some.zip" => \@a, MultiStream => 1;

Push each uncompressed stream from "some.zip" onto @a.

• Added IO::Compress::Base::FilterEnvelope
• Added IO::Uncompress::Base::nextStream
• The '-' filehandle now maps to either *STDIN or *STDOUT. This keeps mod_perl happier. Was using these before

            new IO::File("<-")
            new IO::File(">-")

      * Preliminary support for reading zip files with zip64 members.

2.000_12 3 May 2006

• Moved the code for creating and parsing the gzip extra field into IO::Compress::Zlib::Extra.pm so that IO::Compress::Zip & IO::Uncompress::Unzip can use it as well.
• Added ExtraFieldLocal & ExtraFieldCentral options to IO::Compress::Zip. These allow the creation of user-defined extra fields in the local and central headers, just like the ExtraField option in IO::Compress::Gzip.
• Moved the zip constants into IO::Compress::Zip::Constants
• Added exTime option to IO::Compress::Zip. This allows creation of the extended timestamp extra field.
• Added Minimal option to IO::Compress::Zip. This disables the creation of all extended fields.
• Added TextFlag option to IO::Compress::Zip.
• Documented Comment and ZipComment options in IO::Compress::Zip.
• Compress::Zlib Fixed gzread to zap the output buffer to an empty string when zero bytes are requested. This matches the behaviour of C::Z 1.x

2.000_11 10 April 2006

• Transparent + InputLength made more robust where input data is not compressed.
• Updated Documentation for zip modules.
• Changed IO::Compress::Zip 'Store' option to 'Method' and added symbolic constants ZIP_CM_STORE, ZIP_CM_DEFLATE and ZIP_CM_BZIP2 to allow the compression method to be picked by the user.
• Added support to allow bzip2 compressed data to be written/read with IO::Compress::Zip and IO::Uncompress::Unzip.
• Beefed up 050interop-gzip.t to check that the external gzip command works as expected before starting the tests. This means that this test harness will just be skipped on problematic systems.
• Merged core patch 27565 from Steve Peters. This works around a problem with gzip on OpenBSD where it doesn't seem to like compressing files < 10 bytes long.
• Beefed up 050interop-bzip2.t to check that the external bzip2 command works as expected before starting the tests. This means that this test harness will just be skipped on problematic systems.

2.000_10 13 March 2006

• AnyUncompress doesn't assume that IO-Compress-Zlib is installed any more.
• Documentation updates.
• Compress::Zlib Changed gzread so that its behaviour matches C::Z::gzread 1.x if it is called after eof. In this case it will write an empty string into the output parameter. This change is solely for backward compatibility reasons.

2.000_09 3 March 2006

• Released to CPAN.

2.000_08 2 March 2006

• Split IO::Compress::Base into its own distribution.
• Split IO::Compress::Bzip2 into its own distribution.
• Added opened, autoflush and input_line_number.
• Beefed up support for $.
• Split IO::Compress::Zlib into its own distribution.
• Beefed up support for zip/unzip
• Breakout zlib specific code into separate modules.
• Limited support for reading/writing zip files

2.000_06 5 October 2005

• Added eof parameter to Compress::Zlib::inflate method.
• Fixed issue with 64-bit

2.000_05 4 October 2005

• Renamed IO:: to IO::Compress:: & IO::Uncompress::*

2.000_04 23 September 2005

• Fixed some more non-portable test that were failing on VMS.
• fixed problem where error messages in the oneshot interface were getting lost.

2.000_03 12 September 2005

• Fixed some non-portable test that were failing on VMS.
• Fixed export of zlib constants from the IO::* classes

2.000_02 6 September 2005

• Split Append mode into Append and Merge
• Fixed typos in the documentation.
• Added pod/FAQ.pod
• Added libscan to Makefile.PL
• Added InputLength for IO::Gunzip et al

2.000_01 22 August 2005

• Fixed VERSION in Compress::Gzip::Constants
• Removed Compress::Gzip::Info from the distribution.

2.000_00 21 August 2005

• First Beta relase of Compress::zlib rewrite.

Compress-Zlib version 1 Changes

1.38 - 6 September 2005

• Integrate core change 25304 -- Symbian Update
• Added libscan to Makefile.PL

1.37 - 12 August 2005

• Change to t/03examples.t for VMS from Abe Timmerman

1.36 - 3 August 2005

• Renamed zlib-src-1.2.3 to zlib-src to help VMS
• Fixed Makefile.PL for VMS
• Fixed t/03examples.t for VMS
• Added a couple of notes about incompatibility with Unix compress.

1.35 - 16 July 2005

• Updated zlib source to 1.2.3
• Fixed problem with where two calls to gzclose would hang the debugger. See https://rt.cpan.org/Ticket/Display.html?id=13789
• Added code from Alexey Tourbin to use XSLoader when available, and DynaLoader otherwise.
• Documented that the compress & uncompress functions were not the same as the Unix utilities of the same name.
• Fixed 05gzsetp -- it left a temp file hanging around.
• Integrate core change 24787 - SvUPGRADE returns void in blead
• Integrate core change 24788 - Makefile.PL adjustments for the core

1.34 - 30 January 2005

• Fixed typo in the README
• Fixed examples.t on Win32 where paths have embedded whitespace.
• Fix for Cygwin and core integration from Jos I. Boumans
• Upgrade zlib source to 1.2.2

1.33 - 14 January 2004

• Reworked Makefile.PL to avoid creating a private copy of zlib. This both simplifies the build, plus it makes life easier for VMS.
• Patches for Makefile.PL to get it to work on VMS supplied by Craig A. Berry.
• memGunzip has very slow on FreeBSD. Turns out to be down to the way realloc works on FreeBSD. Changed both inflate & deflate to use exponentially increasing buffer sizes when they need to realloc. Thanks to Peter Jeremy for the lowdown on FreeBSD memory allocation.

1.32 - 26 November 2003

• Steve Hay reported a problem on rt.cpan.org with Windows and MSCV++ 6.0 where the source from the zlib directory was getting installed with the rest of the module. https://rt.cpan.org/Ticket/Display.html?id=1741

  This has been fixed by renaming the "zlib" directory to "zlib-src" thus avoiding a conflict with the name of this Perl module.

• Fixed a bug in the inflate method where the input buffer is an lvalue (via substr). Problem & solution reported by Salvador Fandiqo.
• Tightened up the logic in Makefile.PL when BUILD_ZLIB is True. Issue spotted by Ralf S. Engelschall.

1.31 - 29 October 2003

• Reinstated the creation of .bak files - $^I seems to need a backup file on Windows. For OpenVMS, the extension _bak is used.

1.30 - 28 October 2003

• Bundled a sub-set of the zlib source with the module and changed the default make behaviour to build with the included zlib source. The previous behaviour of using a pre-built zlib library is still available for those that want it.
• Added prototypes to the subs in Zlib.pm that didn't already have them. Patch from Ed Avis.
• No .bak files are created by Makefile.PL any more - this keep distclean much happier. Patch suggested by Ed Avis. This also fixes a similar problem reported by Dr. Martin Zinser on OpenVMS.
• Documentation for some of the gz functions updated.
• Format strings modified in DispStream to keep OpenVMS happy. Problem reported by Dr. Martin Zinser.

1.22 - 17 June 2003

• Makefile.PL now displays a warning about not installing Compress::Zlib via the CPAN shell.
• Fix to allow intermingling of gzread & gzreadline - patch supplied by Doug Perham.
• memGunzip will silently now work if the gzip trailer is missing. Some HTTP Origin Servers seem to leave it out.

1.21 - 28 April 2003

• Tests 148 & 150 from t/02zlib.t were failing on redhat 9.
• Added a few words about the problems with Mac OS X to the README file.

1.20 - 4 April 2003

• Fixed bug in gzopen where $gzerrno wasn't being set correctly. The symptom was $gzerrno was set to Z_MEM_ERROR although the file was opened ok. If gzopen failed, $gzerrno was being set correctly. This problem wasn't spotted before because the typical test to determine whether gzopen passed or failed was to check its return value.

1.19 - 31 October 2002

• fixed a problem with t/02zlib.t that was failing with redhat 8.

1.18 - 24 October 2002

• fixed a Win32 problem in t/02zlib.t by changing sysread to read.
• zlib 1.0.5 & older doesn't have gzsetparams & gzeof. Added a new variable to config.in to flag an old version of zlib. Split out the tests for gzsetparams into t/05gzsetp.t

1.17 - 23 May 2002

• Moved the test to check the versions of libz & zlib.h into a separate file and added troubleshooting notes to README.
• In gzopen, only attempt to call "tell" for normal files.
• Fixed to work in taint mode.
• Broke changes out of README into Changes file.
• Replaced internal use of Z_PARTIAL_FLUSH symbol with Z_SYNC_FLUSH. zlib.h says /* will be removed, use Z_SYNC_FLUSH instead */

1.16 - 13 December 2001

• Fixed bug in Makefile.PL that stopped "perl Makefile.PL PREFIX=..." working.

1.15 - 4th December 2001

• Changes a few types to get the module to build on 64-bit Solaris
• Changed the up/downgrade logic to default to the older constructs, and to only call a downgrade if specifically requested. Some older versions of Perl were having problems with the in-place edit.
• added the new XS constant code.

1.14 - 27th August 2001

• Memory overwrite bug fixed in "inflate". Kudos to Rob Simons for reporting the bug and to Anton Berezin for fixing it for me.

1.13 - 31st June 2001

• Make sure config.in is consistent when released.

1.12 - 28th April 2001

• Modified Makefile.PL to only enable the warnings pragma if using perl 5.6.1 or better.

1.11 - 17th February 2001

• Added logic in Makefile.PL to toggle between using $^W and the warnings pragma in the module.
• The module, the examples & the test harness are now all strict & warnings clean.

1.10 - 5th February 2001

• fixed a bug in memGunzip. Used Z_ERR instead of Z_DATA_ERROR.

1.09 - 15th January 2001

• Silenced a few compiler warnings.
• Updated zlib home site in README & Zlib.pm to www.info-zip.org
• Minor typo in Zlib.pm - a link used AUTHORS instead of AUTHOR -- spotted by Frank Martini.
• Mention Archive::Zip
• added memGunzip. This is largely based on code provided by Jim Leonard.
• $deflate->flush can now take an optional parameter. Valid values are Z_NO_FLUSH, Z_PARTIAL_FLUSH, Z_SYNC_FLUSH, Z_FULL_FLUSH and Z_FINISH. The default is Z_FINISH.

1.08 - 6 Jan 2000

• uncompress was clobbering its input parameter. Now it doesn't. This bug was spotted by Deven T. Corzine.
• If a variable that only ever contained a number was given to compress or deflate, it would not be compressed properly. Now it will be coerced to a string and then compressed. This bug was spotted by Deven T. Corzine.

1.07 - 27 Nov 1999

• ANSI-ified the static functions in Zlib.xs
• Added the ability to build zlib along with the module. This feature is 90% based on a Makefile provided by Gurusamy Sarathy.

1.06 - 20 Sep 1999

• Fixed a nasty problem where inflate could truncate the data returned. Thanks to Douglas Thomson <dougt@mugc.cc.monash.edu.au> for both spotting the problem and fixing the bug.
• Added a note about the undocumented features in zlib that are required when accessing zip files.
• gzclose will now get called automatically when the gzip object is destroyed.

1.05 - 3 June 1999

• Previous release used newSVpvn, which doesn't exist in 5.004_04 or earlier. Changed to use newSVpv instead.
• The module needs Perl 5.004 or better, so updated the version checking in Zlib.pm and Makefile.PL

1.04 - 27 May 1999

• Bug 19990527.001: compress(undef) core dumps -- Fixed.

1.03 - 17 Mar 1999

• Updated to use the new PL_ symbols. Means the module can be built with Perl 5.005_5*

1.02 - 31 Jan 1999

• The return codes for gzread, gzreadline and gzwrite were documented incorrectly as returning a status code.
• The test harness was missing a "gzclose". This caused problem showed up on an amiga. Thanks to Erik van Roode for reporting this one.
• Patched zlib.t for OS/2. Thanks to Ilya Zakharevich for the patch.

1.01 - 23 Nov 1997

• A number of fixes to the test suite and the example scripts to allow them to work under win32. All courtesy of Gurusamy Sarathy.

1.00 - 14 Nov 1997

• Fixed crc32 & adler32. They were very broken.
• The following functions can now take a scalar reference in place of a scalar for their buffer parameters:

            compress
            uncompress
            deflate
            inflate
            crc32
            adler32
 
        This should mean applications that make use of the module don't
        have to copy large buffers around.

• Normally the inflate method consumes I<all> of the input buffer before returning. The exception to this is when inflate detects the end of the stream (Z_STREAM_END). In this case the input buffer need not be completely consumed. To allow processing of file formats that embed a deflation stream (e.g. zip, gzip), the inflate method now sets the buffer parameter to be what remains after inflation.

  When the return status is Z_STREAM_END, it will be what remains of the buffer (if any) after deflation. When the status is Z_OK it will be an empty string.

  This change means that the buffer parameter must be a lvalue.

• Fixed crc32 and adler32. They were both very broken.
• Added the Compress::Zlib::memGzip function.

0.5 - Confirmed that no changes were necessary for zlib 1.0.3, or 1.0.4.

        The optional parameters for deflateInit and inflateInit can now
        be specified as an associative array in addition to a reference
        to an associative array. They can also accept the -Name
        syntax.

        gzopen can now optionally take a reference to an open
        filehandle in place of a filename. In this case it will call
        gzdopen.

        Added gzstream example script.

0.4 - Upgrade to support zlib 0.99

Added dictionary interface.

        Fixed bug in gzreadline - previously it would keep returning
        the same buffer. This bug was reported by Helmut Jarausch

        Removed dependency to zutil.h and so dropped support for 
        
            DEF_MEM_LEVEL (use MAX_MEM_LEVEL instead)
            DEF_WBITS     (use MAX_WBITS instead)

0.3 - Added prototype specification.

0.2 - Fixed a minor allocation problem in Zlib.xs

0.1 - first alpha release. 2nd October 1995