Revision history for CGI-Buffer

0.85	Fri Oct 20 13:03:18 EDT 2023
	Don't try to build Brotli on platforms it doesn't support

0.84	Sat Aug 26 17:14:06 EDT 2023
	Fix uninitialised variable
	Try to track https://www.cpantesters.org/cpan/report/cbb57c6a-bf6b-11ed-b310-26e96d8775ea
	Ensure 403 statuses are printed

0.83	Fri Feb 24 15:45:21 EST 2023
	Bump minimum version because of IO::Compress::Brotli failures
	Fix problem where Last-Modified header not set when no cache
	Generate 304 on non-cached files
	Added HTTP/2.0 recognition
	Make debugging more readable
	Log correctly when data isn't retrieved from the cache
	Call logger instead of warn where possible

0.82	Tue Aug  7 09:45:13 EDT 2018
	Fix RT#123101
	Optmise <center>

0.81	Thu Jul 19 08:50:53 EDT 2018
	Get the return status from the CGI::Info object
	Optimise lists a bit more

0.80	Mon Sep 25 13:03:24 EDT 2017
	Remove false positive bug reports
	Work with the latest IO::Compress::Brotli

0.79	Fri Jan  6 11:17:01 EST 2017
	Fix Brotli tests which failed in some places
	Improved handling of HTTP status
	Put different languages in different caches

0.78	Mon Oct 10 09:35:18 EDT 2016
	Print stack trace when calling init too late
	Support Brotli
	Honour the order of HTTP_ACCEPT_ENCODING

0.77	Sat Feb 27 11:55:55 EST 2016
	Fix RT112029
	Call the right routine

0.76	Sun Feb 14 19:10:51 EST 2016
	generate_key - honour the vary header
	Fix problem were server side cached contents weren't compressed when
		loading the content from the cache
	If you give a cache to init() then later give cache => undef,
		the server side cache is no longer used.

0.75	Sun May 17 09:00:46 EDT 2015
	Added X-Cache and X-Cache-Lookup headers
	Fixed overly agressive optimisation
	Don't use the cache if the client sends max-age=0
	Added cache_age option to init()
	Fix set_options when given a ref
	Fix RT102226
	Improved handling of Wide characters

0.74	Sat Jan 24 10:47:50 EST 2015
	Added t/strict.t, t_unused.t and t/used.t
	When a script has been updated clear all the cache
	Added can_cache()

0.73	Fri Nov  7 16:48:40 EST 2014
	It's is_search_engine, not is_search!
	Fix missing encoding_utf8

0.72	Fri Nov  7 11:03:55 EST 2014
	Added the info option to init
	Removed some unused modules, and defer loading others
	Removed Exporter and use vars, use our instead
	Separate the web and mobile caches

0.71	Wed Apr  2 16:50:48 EDT 2014
	Speed optimisation with some output

0.70	Fri 13 Dec 10:24:45 EST 2013
	Loop optimisation until complete for even smaller output
	Use Test::Tempdir to create the temporary files when testing
	# Use RegExp::List to consolidate some scans of the output
	Ensure logger isn't called after it's been destroyed since that
		would cause Log4Perl to say it hasn't been initialised
	Use $cobject->value() rather than $cache->get() to see if the
		object data is there as well. There were some false
		negatives
	Accept HTTP_TE as well as HTTP_ACCEPT_ENCODING
	Output parsing improvements courtesy of Regexp::List
	Use Test::Most instead of Test::More
	Support HTTP_RANGE as well as Range
	Added more JavaScript optimisations when optimise_level >= 2

0.69	Tue Apr  2 14:59:59 EDT 2013
	Bumped minimum version of CGI::Info, needed for Cygwin
	t/output.t: try harder to ensure the "to be tested" version of
		CGI::Buffer is tested
	Minor speed improvements
	_check_modified_since, fix in 0.68 was broken

0.68	Sat Feb  2 12:13:03 EST 2013
	Handle situation where not modified but Etag is different would
		generate 304
	Added generate_304 option (defaults to 1)
	Handle situation where IF_MODIFIED_SINCE isn't a valid date
	On some platforms (e.g. Cygwin) t/is_cached is failing, so
		added debugging to find out why

0.67	Sun Oct 21 23:00:13 EDT 2012
	Remove spaces after <body>.
	Fix removal of spaces between </head> and <body>.
	Better handling of no body, e.g. when redirecting using Location

0.66	Thu Oct 11 12:11:11 EDT 2012
	Add sanity check.
	Only add Last-Modified when statis is 200
	Don't get the object from the cache twice

0.65	Sat Sep 29 14:05:17 EDT 2012
	General tidy up and minor speed improvements
	Improved handling of cache errors

0.64	Mon Sep 24 17:08:15 EDT 2012
	Only store in the cache when status is 200

0.63	Sat Sep 22 18:45:27 EDT 2012
	Fixed t/is_cached.t for systems without CHI
	Sometimes 304 was not generated for when giving IF_MODIFIED_SINCE in
		scenarios were it could have been
	Amalgameted that 3 separate caches into one cache.
	Minor speed improvements

0.62	Wed Sep 19 10:09:53 BST 2012
	Added resilience to is_cached in case the age of the script can't be
		determined.
	Removed logger calls from END, since it'll have gone by then
	is_cached now returns false unless both body and headers can be found

0.61	Sat Sep 15 17:57:03 BST 2012
	Fixed typo in subroutine name

0.60	Fri Sep 14 08:44:00 BST 2012
	Removed superflouous whitespace around the top of the document and
		in lists
	Added t/unused.t
	Remove spaces after <div>
	Remove spaces before </div>
	Simplified the cache key
	If the script has been modified since the value in IF_MODIFIED_SINCE
		don't return 304, since the output may have been changed
	If the script has been modified since the time it was stored in the cache
		don't return true from is_cached()
	Use the cookies to ensure that different states of the client are stored
		in different entries in the cache

0.59	Wed Aug  8 10:00:00 BST 2012
	t/is_cached: fix tests on systems without CHI

0.58	Tue Aug  7 12:40:03 BST 2012
	Improve the new-line handling between header fields
	Replaces stats_file option with logger
	Fix putting Content-Length in twice when cache is given
	Optimise <hr> tags
	Remove more redundant whitespace around hrefs
	Test::Kwalitee isn't needed for build, only for optional testing
	Only get the etag value from the cache when needed

0.57	Tue Jul 24 13:51:43 BST 2012
	Use get() rather than is_valid()
	If get on the body fails, remove the corresponding headers from the
		cache

0.56	Tue Jul 24 13:45:57 BST 2012
	Always store the unzipped body in the cache.  It uses more disc
		space, but makes the code much simpler and more robust

0.55	Mon Jul 23 12:14:34 BST 2012
	Only store a response in the cache when HTTP status is 200
	Ensure that the headers are in the cache as well as the body
		before deciding to send the server's cached response

0.54	Thu Jul 19 13:53:58 BST 2012
	lint_content is now off by default

0.53	Wed Jul 18 10:57:50 BST 2012
	Added support for lint_content to put output through HTML::Lint

0.52	Fri Jul 13 09:27:43 BST 2012
	Don't create cache entry for empty headers
	src/testit: check etag and last-modifed don't return 304 when the
		content has changed

0.51	Thu Jul 12 11:20:19 BST 2012
	Store the etag value in the cache to ensure correctness
	Added src/testit to check etag and last-modified HTTP headers

0.50	Tue Jul 10 19:28:36 BST 2012
	Fixed use of an uninitialised variable
	Don't set 304 if status != 200

0.49	Sun Jun 17 22:57:52 EDT 2012
	Made the stats_file more readable
	Fixed IF_MODIFIED_SINCE which was often not working

0.48	Sun Jun 17 08:48:25 EDT 2012
	Added stats_file option

0.47	Fri Jun 15 12:56:57 EDT 2012
	Fix bug where 307 wasn't sent if the first request set encoding to gzip
		and subsequent check didn't

0.46	Sat Jun  2 14:02:39 EDT 2012
	304 can be sent irrespective of the value of gzip - so check
		for both in the cache

0.45	Tue May 29 12:19:41 EDT 2012
	Honour Cache-Control: private
	Look at If-Modified-Since before If-None-Match to reduce etag
		calculations
	Unlikely possibility: if cached headers have an etag that matches our
		body but we somehow missed that, then send 304

0.44	Mon May 28 10:39:57 EDT 2012
	Corrected logic error

0.43	Mon May 28 00:31:40 EDT 2012
	Generate Last-Modified header when possible

0.42	Sat May 26 15:31:54 EDT 2012
	Tightned tests for Content-Length
	Send 304 if object in cache is newer than IF_MODIFIED_SINCE

0.41	Mon May 14 12:30:32 EDT 2012
	Allow options to be given in the 'use' call

0.40	Tue May  1 20:58:23 BST 2012
	Check gzip is valid
	Avoid 'Wide character in memGzip'
	Renamed set_options to init, but kept set_options as a synonym for
		legacy apps
	Allow a reference to be passed to init

0.39	Fri Apr 20 09:04:51 BST 2012
	Avoid "Wide character in subroutine entry" when generating ETags
	Remove warning on perl5.15.9
	Added tests that If-None-Match can generate 304 return status

0.38	Wed Apr 18 11:32:31 BST 2012
	Only do kwalitee test in author environment because it throws too
		many false positives
	Remove blank spaces between table cells

0.37
	Turned JavaScript optimisation off for optimise_content level 1.

0.36	Sat Mar 17 16:47:25 EDT 2012
	Drop use of HTML Clean until RT402 is fixed.

0.35	Thu Mar  8 10:07:35 EST 2012
	Include JavaScript::Packer as a prerequisite
	Bump minimum required HTML::Packer in an attempt to fix
		http://www.cpantesters.org/cpan/report/b077836c-688d-11e1-b4da-8ac6d9130bc2
	Added aggressive optimisation for document.write and t/js.t

0.34	Wed Mar  7 15:51:08 EST 2012
	Bump minimum required HTML::Clean in an attempt to fix
		http://www.cpantesters.org/cpan/report/40d9566e-6673-11e1-a50d-8482959b33a2
	Introduce optimise_content levels, 0 = OFF, 1 = normal, 2 = aggressive (may fail)

0.33	Mon Feb 27 11:42:45 EST 2012
	Fix occasional unitialised value
	Various performance enhancements
	Use Digest::MD5 instead of MD5

0.32	Sat Feb 25 16:53:39 EST 2012
	Call the correct Perl in t/output.t - RT75309.  Thanks to ANDK@cpan.org

0.31	Mon Feb 20 11:30:29 EST 2012
	Handle Range header
	Fixed t/output.t to look in PERL5LIB for locally installed modules

0.30	Sat Feb 18 12:37:29 EST 2012
	Don't gzip if the output is < MIN_GZIP_LEN
	Honour cache-control HTTP header

0.29
	Added tests to t/output.t to verify that Content-Length is correct
	Don't tell HTML::Packer to remove newlines.

0.28	Sun Jan 29 12:33:55 GMT 2012
	Fixed test count in t/output.t

0.27	Sat Jan 28 11:04:32 GMT 2012
	Use the domain name as part of the key, so that if two scripts of
		the same name are used in different hosts, ensure that they
		are kept separate
	Remove use of JavaScript::Packer.  It produces JavaScript that doesn't
		work - I don't know what's wrong.
	Fix bug removing our own name in links to our own domain

0.26	Fri Jan 27 11:07:13 GMT 2012
	Added t/output.t
	Use shrink not best in JavaScript::Packer
	Only import HTML::Clean and HTML::Packer if they are going to be used.

0.25	Wed Jan 25 10:40:25 GMT 2012
	Fixed the way HTML::Packer was called

0.24	Tue Jan 24 21:59:00 GMT 2012
	Include HTML::Packer. Eventually I will probably drop HTML::Clean,
		so I'm testing HTML::Packer.  HTML::Clean doesn't seem to
		be supported any more

0.23
	Improved header handling when caching few headers

0.22	Fri Jan  6 20:58:43 EST 2012
	Ensure that two newlines are appended when doing redirects
	If CHI isn't installed, get the count right in t/too_late.t

0.21
	Fixed spelling mistakes in the documentation
	More optimise content at end of paragraphs and hyperlinks

0.20	Tue Nov  1 13:48:55 EDT 2011
	More server side cache fixes

0.19	Mon Oct 31 22:52:01 EDT 2011
	Send 304 if server cache matches client's HTTP_IF_NONE_MATCH

0.18	Mon Oct 31 17:41:48 EDT 2011
	Small optimisations
	Fix is_cache handling of gzipped content_encoding

0.17	Mon Oct 31 16:09:09 EDT 2011
	Fix bug in X-CGI-Buffer header
	Carp if you're trying to set an unsafe option after output has been
		generated

0.16	Mon Oct 24 09:54:09 EDT 2011
	Improved client side caching

0.15	Sat Sep 24 16:08:32 EDT 2011
	Improved error handling

0.14	Mon Sep 19 14:54:10 EDT 2011
	Another attempt at fiximg HEAD requests

0.13	Mon Sep 12 14:17:55 EDT 2011
	Fixed the call to "new HTML::Clean"

0.12	Tue Sep  6 14:55:17 EDT 2011
	Improved handling of HEAD
	Added t/is_cached.t
	Use HTML::Clean

0.11	Thu Aug 18 16:10:27 EDT 2011
	Optimise table rows
	Added t/critit.t

0.10	Wed Aug 10 13:31:46 EDT 2011
	Optimise HEAD handling

0.09	Tue Aug  2 11:58:47 EDT 2011
	Fix problem with HEAD which meant that the headers weren't being sent
	optimise_content: Add optimisations for <br> and <div>

0.08	Mon Aug  1 10:57:32 EDT 2011
	Ensure that the correct encoding mechanism is passed to the client

0.07	Sun Jul 31 20:06:52 EDT 2011
	Fixed compilation error

0.06	Fri Jul 29 18:04:56 EDT 2011
	Added cache_key option
	Fixed problem with cache delivering gzipped version to non gzip
		aware browsers

0.05	Wed Jul 27 23:07:50 EDT 2011
	Fix the documentation
	Fix calls to is_cached() when there is no cache
	Added new optimise_content

0.04	Wed Jul 27 10:19:41 EDT 2011
	Fixed over zealous optimise_content

0.03	Mon Jul 25 16:38:00 EDT 2011
	Improvements to optimise_content

0.02	Sun Jul 24 20:18:15 EDT 2011
	If protocol isn't known, assume http

0.01    Fri Jul 22 13:09:17 EDT 2011
	First draft