Changes

There are a number of things that have changed since the previous version.

Changes in 0.5.3

Added: support for inline comments in the input file, see Files for more information.

Bugfix: for some reason there was a problem with UTF-8 strings on Mac OS X systems, causing really weird malformed output. Fixed.
Bugfix: VSCA had problems understanding end-of-line bytes that were, according to Perl, not native to the system VSCA ran on. In 0.5.2, VSCA understands *Nix, Windows, and Mac line endings, as well as the unicode characters End Of Line and End Of Paragraph. Even mixed line endings within a file work now (I really saw a file that had that, very odd).

Improved behaviour: *sigh*, there was still something wrong with polygraph recognition in words. 0.5.1 still offers rule-specific and global polygraphs, but from now on the global list is created by only taking polygraphs that appear inside variable values. Early tests ran by me point out that this works pretty good, but seeing how the whole polygraph thing has changed in almost every release, expect this to be changed again.

Added: the -d and -c command line switches
Removed: the -m command line switch. Due to the improved behaviour of the symbol splitter (see changes in 0.4), it didn't really make sense to have the search modes anymore. The new search mechanism offers both of best worlds.
Improved behaviour: the major change in 0.4 caused another misunderstanding of some polygraphs. Version 0.5 offers the best from both worlds of 0.4 and the 0.3.x versions. Long story short: polygraph symbols are now stored per-rule and globally, and word-chopping appears in two phases. The first phase chops the word according to the rule-specific polygraphs, and all left-over monographs are combined into polygraphs according to the global polygraphs.

Added: optionals (still very experimental, though)
Improved behaviour: sometimes VSCA would misunderstand the way you mean to chop a word into monographs and polygraphs. This was caused because it did the chopping right at the moment when it read the word from the lexicon file, and then never looked back to the result of that. However, sometimes this caused the word to be chopped in a way that's useless to a rule. This behaviour is now changed: VSCA does no longer collect a global list of polygraph symbols during compile time. Instead , when your ruleset is converted to VSCA's initial format, a list of polygraphs that is required for a specific rule is stored with that rule. Now before a word is ran through a rule, VSCA splits the word into symbols according to the polygraph symbols that are relevant to that rule.

Bugfix: in some cases Perl would throw a warning at you. Nothing harmfull, and everything worked just fine, but the warning was a bit annoying and confusing. Well, I had some goofy code but it's fixed now, warning is gone.

Bugfix: utf-8 files that somehow start with a BOM (Byte Order Mark) are now properly dealt with

Added: the command line switches -H, -h, -m, -u
This feature has been taken out in v0.5.. Changed: added "best match". This is the default search mode, although you can still choose v0.1 behaviour by selecting "first match" mode (-m f).
You might want to learn about search modes if you have used version 0.1: thanks to the "best match" mode, writing rules is less of a pain.
Improved: the behaviour of the "+" symbol inside polygraphs has improved. This involves all polygraphs, whether they are Position or Exception fields, or polygraphs within variables, Original, or New fields. Long story short: a plus now only doubles the variable immediately preceding it (and not just everything preceding it), and double (triple, quadriple, ...) pluses now work too.