2004-06-06 Darren Duncan * Release 1.09. * Updated CGI::MultiValuedHash (now v1.09 up from v1.081) to fix a bug in the to_html_encoded_table() method where key names were not being output in the generated HTML table. No other changes. 2003-03-16 Darren Duncan * Release 1.081. * This is a small maintenance release that is mainly for improving documentation. Similar maintenance releases are being done with all of my distributions simultaneously. A lot of it was formatting issues that were easy to miss before; the other bullet points below list the non-formatting issues. These files had formatting fixes: MultiValuedHash.pm, ChangeLog. * The distribution file layouts have also been standardized, with modules going in a "lib" folder and tests going in a "t" folder. "test.pl" was changed to "t/CGI-MultiValuedHash.t" and "MultiValuedHash.pm" was changed to "lib/CGI/MultiValuedHash.pm". * Updated NAME in MultiValuedHash.pm to match the official registered module list. * Updated all Copyright statements to a better (and longer) version, which is dated 1999-2003. Your rights to use these modules haven't changed, but they may now be easier to understand. These statements appear once in the ReadMe file at the bottom (under "AUTHOR"), and twice in each module (.pm file), a full version at the bottom (under "AUTHOR") and a shorter version at the top (in code comments). * Added the "use warnings" pragma at the top of all files beside "use strict"; every occurance of "use strict" gained a "use warnings" whether it was in the module proper or in its test script or in its SYNOPSIS. This should save users from having to put -w on the command line for them. * Incremented the module $VERSION number by 0.001, and all test scripts to require the new version. Also, where any of my modules use others of my modules, they now require the latest version as of this writing, for simplicity of maintenance. * Updated all test scripts to read $verbose from @ARGV (command line arguments), and other test script improvements. * Updated these sections of the ReadMe file: INSTALLATION. 2002-12-01 Darren Duncan * Release 1.08. * The modules Data::MultiValuedHash and CGI::MultiValuedHash are now being distributed separately from each other, largely so they can be maintained as independant entities. The 2 new distributions, Data-MultiValuedHash-1.08 and CGI-MultiValuedHash-1.08, replace the older MultiValuedHash-1.0703. That said, to use the CGI module you still need the Data module, since the first is a subclass of the second. * As part of the above separation, each of the new distributions has a copy of all the old supporting files with alterations to focus on the one module it is staying with: ChangeLog, Makefile.PL, MANIFEST, ReadMe, test.pl. Both ChangeLog copies are identical for all releases before and including 1.08 (applying to both modules), but will differ for any later releases. * Minor bug fix release to remove any 'uninitialized value' warnings that appear when running the test suite with the -w option. All of the uncovered warnings were in the Data::MultiValuedHash module, of which the test suite has already covered every method and method argument variation (but the module had correct output before). * However, many of the methods in CGI::MultiValuedHash are not yet included in the test suite, and may conceal bugs that a thorough suite would expose: to_file(), from_file(), to_html_encoded_table(), to_html_encoded_hidden_fields(), trim_bounding_whitespace(), batch_to_file(), batch_from_file(). That said, many of those methods have been used in a production environment without problems to date. These methods *have* been tested, with all their argument variations: to_url_encoded_string(), from_url_encoded_string(). * Updated the HTML generating methods in CGI::MultiValuedHash, to_html_encoded_table() and to_html_encoded_hidden_fields(), to add support for the stricter requirements of W3C's XHTML standard where possible. See "http://www.w3.org/TR/xhtml1" for details on how this standard compares to HTML 4. Mainly this means that the HTML now has lowercased tag and attribute names, and solitary tags like 'input' and 'br' have a trailing slash in them (eg: '
'). However, the changes should be backwards-compatible with older user agents (web browsers). A bug was also fixed in the first method involving a mis-matched pair of 'table' tags. * Minor documentation fix in Data::MultiValuedHash for the splice() method; the new version indicates which arguments are optional. * Several documentation updates in the README file; one is so that someone can get an idea what this module does after looking only at the README. 2001-07-16 Darren Duncan * Release 1.0703. * This minor release deals with some more "ambiguous call" warnings of the sort that release 1.07 addressed, since I didn't get to them all in 1.0702. Both modules are now at version 1.0703. In CGI::MultiValuedHash, the POD was updated in Synopsis and The Default File Format; for the latter case, I hope that I removed ambiguity regarding how compatible my default file format is with Boulderio. Thanks to Jonathan Snyder for alerting me to the fixed problems. 2001-05-28 Darren Duncan * Release 1.0702. * This minor release deals with some more "ambiguous call" warnings of the sort that release 1.07 addressed, since I didn't get to them all that time. There are also minor POD updates in this release. Both modules are now at version 1.0702. 2001-04-20 Darren Duncan * Release 1.0701. * This release should take care of some install problems because the modules are now inside folders like lib/CGI and lib/Data whereas they were just in CGI and Data before. Thanks to Jost (kriegjcb@CMC01SE.rz.ruhr-uni-bochum.de) for alerting me to this problem. I'm afraid I still don't know much about Makefiles. 2001-04-12 Darren Duncan * Release 1.07. * This release should take care of a number of warnings like "Ambiguous call resolved as CORE::values()" which appear on some systems when -w is used. These spring from the fact that Data::MultiValuedHash has methods with the same names as the following built-in functions: keys(), values(), exists(), push(), unshift(), pop(), shift(), delete(). To fix this, I have changed all such calls to built-in functions so that they explicitely say CORE::*. However, the modules were working before, as Perl seemed to be resolving the ambiguity in the correct way. * Thanks to Geir Johannessen (geir.johannessen@nextra.com) for alerting me to the above warnings. 2001-03-17 Darren Duncan * Release 1.06. * This release is the first one that includes the CPAN standard file "test.pl", which was created following the previous release. The install instructions in the ReadMe file were updated to reflect that "make test" should now work. "Test.pl" was used to discover several bugs that my production use of the modules didn't; their fixes are detailed below. * Removed gratuitous arguments from clone() that allowed one to specify a subset of keys for the clone to have. This way, clone() is fully standard and does what one would intuitively expect, namely make an identical copy. * Added new method fetch_mvh() to Data::MultiValuedHash which replaces the functionality dropped from clone(). fetch_mvh() has the same calling conventions as fetch_all(), except that an MVH object is returned instead of a literal hash of array refs. * Fixed bug in Data::MultiValuedHash where the store_all() method didn't properly handle the import of case-sensitive keys into a case-insensitive MVH. Specifically, keys were added but associated values were lost. * Fixed bug in Data::MultiValuedHash where the splice() method didn't properly handle missing LENGTH arguments. That is, if one called "splice( KEY, -1 )" then that should remove the last element of the array, but it instead acted like LENGTH had a value of 0, so nothing was removed or returned. This issue appears to be related to Perl's built-in splice function, which treats "splice( KEY, 1 )" and "splice( KEY, 1, undef )" or "splice( KEY, 1, () )" differently. So the fix for my splice() method involved having two separate calls to Perl's splice() depending on whether its OFFSET argument is defined or not. * Added several new POD sections to the end of CGI::MultiValuedHash that give an overview of using url-encoding/decoding and file-import/export features, mainly describing what the file format or encoding formats look like. 2001-03-14 Darren Duncan * Release 1.0501. * Fixed bug in CGI::MultiValuedHash where the method to_html_encoded_table() was misnamed as to_html_encoded_hidden_fields(), which is the same as a separate method that also exists. This module's version was increased to 1.0501 to reflect the change. 2001-03-08 Darren Duncan * Release 1.05. * Added method splice() to Data::MultiValuedHash so as to round out the set of standard array manipulation functions. * Added static function batch_new() to Data::HashOfArrays which emulates the symantecs of calling new() multiple times with a list of initializers; as many new MVH objects are made as there are elements in the list. * Added programmer documentation to the private method _reduce_hash_from_subset() so people know how to use this method. Due to its private nature, it is not covered in the POD. * Added static functions batch_to_file() and batch_from_file() to CGI::MultiValuedHash so that an entire list of MVH objects can be written to or read from a file at once. Symantecs are similar to calling the to_file() and from_file() methods. batch_from_file() will read to the end of the file by default, but can be restricted by an argument to stop after a certain number of objects are retrieved. 2001-03-07 Darren Duncan * Release 1.04. * This release is the first one that includes the CPAN standard files "Makefile.PL" and "MANIFEST", which were all created following the previous release. The content of "Manifest" was previously inside the "ReadMe" file. Improved the install instructions in my ReadMe file. * Added a hook to the initialize() method in Data::MultiValuedHash for handling initializers that aren't hashes. CGI::MultiValuedHash now uses that hook for handling more initializer types, rather than overloading all of initialize(). * Many small changes were made to methods in Data::MultiValuedHash so that they work better or the code is more readable. * Added "Method Relationship Overview" section to the POD of Data::MVH. * Changed the ignores_case() method so that you can now change that property. When changing this property from false to true, any existing keys are lowercased, which can not be reversed (except from a backup). * Added method fetch_hash() to Data::MVH which returns a hash containing all of the keys and one for each; values are all taken from the array index specified with an argument. fetch_first() and fetch_last() have been changed to use this to implement themselves, since they are the same except with implicit indexes of 0 or -1. * Added INDEXES argument to fetch() and fetch_all() so that a caller can restrict a range of values to be returned for each key, much as the KEYS argument of some methods restrict which keys are returned; the new argument does not have a COMPLEMENT modifier, however. * Added a method to CGI::MultiValuedHash for outputting the object's keys and values in an HTML table. 2001-01-03 Darren Duncan * Release 1.03. * This release is the first one following official registrations with "The Perl 5 Module List". The updated entries would look like this: Data:: ::MultiValuedHash bdpO Hash whose keys have multiple ordered values DUNCAND CGI:: ::MultiValuedHash bdpO Store and manipulate url-encoded data DUNCAND * This release contains my "Data::MultiValuedHash" and "CGI::MultiValuedHash" modules, which were previously a part of the following distributions under the temporary name of "CGI::HashOfArrays": - "libdwg": v1.0 r2000-07-23, v1.11 r2000-08-23 - "CGI-FormGenerator": v0.9 r2000-09-04, v0.9201 r2000-12-26 * These modules require Perl version 5.004. * Code changes since CGI::HashOfArrays consist mainly in its separation into two modules, named above. This took care of a longstanding logistical problem concerning whether the module was a generic data structure or a tool for encoding/decoding CGI data. The CGI:: module is a subclass of the Data:: module, so you can use it like you used the old module. Thanks to Johan Vromans for suggesting the split. * The POD also received significant updates. Thanks to Steve Benson for suggesting improvements to initialize(). * Added method store_value() to Data::MultiValuedHash. 2000-12-26 Darren Duncan * CGI-FormGenerator 0.9201, the last version of any distribution to include CGI::HashOfArrays, was released. 2000-07-23 Darren Duncan * libdwg 1.0, the first version of any distribution to include CGI::HashOfArrays, was released on CPAN. 2000-04-21 Darren Duncan * Completed primary development on this module. 2000-01-28 Darren Duncan * Began development on the final version of this module. * Module based on a template created by h2xs 1.18. 1999-07-29 thru 1999-11-13 * Worked on prototype of code that ended up in this module.