My future is static
It's already had it
I could tuck you in
And we can talk about it
(Sonic Youth, "Schizophrenia")

Future improvements

Minor enhancements

• Improve hscpitt: Add a copy/move command.
• Rename some message classes: message classes ``bad style'' and ``fatal error'' will probably be renamed to ``flaw'' and ``failure'', so the class name will also consists of only one word. Furthermore, ``fatal'' sounds too hard.
• Rename internal attributes and filenames: currently, internal attributes are named like HSC.XY, and for filenames there are no templates at all. I will probably use the prefix ``hsc'' and a hyphen (``-'') as separator, resulting into attributes like hsc-click-here and filenames like hsc-XXXX.tmp.

Major enhancements

• New project database: instead of a single project file, hsc should store it's data in separate files for every document. A document data file is only loaded during parsing if some information about a specific document is needed. This should speedup huge projects with more than 100 documents.
• Additional document data: Without remarkable changes, hsc could also collect useful information about a document, like document title, URIs referenced etc. Of course, some access functions will be provided, too.
• Index document creation: A new tag <$index> will allow the user to attach some private information to the document data, which later can be retrieved using hscpitt. A script the user will have to write can use these to create a html-object. Easy to implement for me, awfully cryptic and complicated, but rather flexible for the user.
• Checking of external URIs: I'm definitely not going to mess around with shitty TCP/IP-stuff, but a script program could retrieve information about HREFs from the document file, and pass them to an external program (like GetURL)
• Functions with multiple arguments: Functions like Exists() or GetFileSize should use the same attribute syntax like tags, for example GetEnv(VAR="HOME").
• New copyright: Version 0.913 is probably the last release distributed under GPL. The major lack of the GPL seems to be that I can't prevent someone else porting hsc to a MS-DOS-based system. But I don't think about not including the source or getting incredibly rich.

Only if I'm bored

• More conditionals: <$select>, <$when> and <$otherwise> as an extended version of <$if>
• Indention: Add option INDENT for <$include> and <$source> to indent preformatted text
• Plausibility checking of external URIs: unknown protocol, missing domain etc.
• Improve expression parser: the current implementation is an embarrassment for a student of computer science, but.. well, it does it's job for now.
• GUI-Version for AmigaOS: I ought to play around with MUI sometimes anyway
• Filename compatibility mode: this mostly means that a ``..'' should act as parent directory on all systems without having to run additional system patches.

Things someone else can do

• Autoconf-support: In the unix-world, a cryptic tool package called autoconf has become quite popular, as it tries to compensate the lack that even at the end of the 20th centaury, no one knows in which header file to expect strftime() on this ridiculous system (ANSI has specified this in the early 80ies, bye the way). Although I don't care much about outdated unix-versions, and I don't like unreadable and badly documented macro languages, I might include the required changes if someone else wastes his time with setting up a working autoconf-installation.
• Improve hscdepp: As I do not consider dependency generators a reasonable concept, someone else will have to add things like excluding specific documents or pattern matching stuff. Hey, it's only about 700 lines of code..
• Document relation editor: It would be nice, if one could edit all those next/prev/up/.. relations with a program, and include these into hsc-sources later. It would make sense to store these relations in the document data file.

No future

• XML-support: It's too ridiculous. If you want to have a jolly good laugh, try a html validator on http://www.w3.org/pub/WWW/TR/WD-xml-lang-970331.html, to get an impression of how competent these people are. After that, read http://www.w3.org/pub/WWW/TR/WD-xml-lang.html to laugh again... This time I'm really glad if all browser developers will ignore it; just another prove that w3c has not produced anything useful after html-2.0.
• Precompiled include files: This would also speed up reading hsc.prefs; I experimented around a bit with this, and it seems that more time is wasted scanning those bloody linked lists then by checking the data. It's more likely that some sort of balanced binary tree will make it into the source.
• Undefine macros or attributes: I do not consider ``undefining'' a clean concept, but a result of mental impotence.
• Type checking for attributes: Anything more then the current (very relaxed) checking would not fit into the typeless concept if plain html.
• Move around in text using <$goto> and <$label>: This one's perverted to the core.
• Support other output-formats like texinfo or AmigaGuide: There are converters around for this task, and several people smarter than me have already failed to introduce the ultimate hypertext authoring tool.