I wouldn't go the way of PCL but rather the way of Postscript, together with any template toolkit that isn't HTML specific (that leaves TT2 from what I see, but I don't know whether one couldn't abuse HTML::Template). This should take care of the layout side.

For the printing process, either have Postscript capable printers (you most likely already have, as you are under Linux), or use Ghostscript on any box to handle the ugly details of converting Postscript to something printable on your printer of choice.

I personally would turn to TeX for the superior layout quality (and ease of producing PDFs), but Postscript should be as readable as (La)TeX, if not more.

perl -MHTTP::Daemon -MHTTP::Response -MLWP::Simple -e ' ;    # The  
$d = new HTTP::Daemon and fork and getprint $d->url and exit;#spider
($c = $d->accept())->get_request(); $c->send_response( new   #in the
HTTP::Response(200,$_,$_,qq(Just another Perl hacker\n))); ' #  web
[download]

To what degree and in what ways are template toolkits made HTML (or other language) specific? I know that, despite its name, CGI::FastTemplate is completely content-agnostic and I've used it to generate TeX output without having to resort to abuse.

I haven't worked with any of the template toolkits, but I imagine that the HTML oriented toolkits might implement automagic HTML entity quoting, which would get in the way or at least be inconvenient. Also, such toolkits might expect to be able to parse their templates with HTML::Parser, or expect their templates to be valid XML / XHTML, except for their own special tags. Some toolkits even bear HTML in their name, which suggests that they cater specifically to HTML output (HTML::Template and HTML::Mason).

But as I said, I didn't work with any of the toolkits, so this remains to be evaluated by the actual user.

perl -MHTTP::Daemon -MHTTP::Response -MLWP::Simple -e ' ;    # The  
$d = new HTTP::Daemon and fork and getprint $d->url and exit;#spider
($c = $d->accept())->get_request(); $c->send_response( new   #in the
HTTP::Response(200,$_,$_,qq(Just another Perl hacker\n))); ' #  web
[download]

Look at LaTeX if you truly just looked at TeX. It's a somewhat higher level system (sort of C to TeX's assembler) and might be more suited to your needs.

Also consider that you don't have to generate the complete (La)?TeX output each time. You could use a templating system to seperate the actual output presentation from its generation.

There are other typesetting systems you could also look at:

• good old troff. Nowadays there are really nice tutorials coming with groff, and I got the book "Unix text processing" for free ages ago.
• Lout. Similar functionality to TeX but much simpler to learn.
• PostScript, if you first get some nice functions to turn it into something more high-level.
• XML + XSLT -> XSL-FO, for example using Apache FOP. XSL-FO is the modern-day equivalent to TeX and Lout and, like everything these days, uses XML syntax.

Once you figure out what high-level concepts you'd want to use (headings, paragraphs, tables?, font changes, etc.), you'll find that those have all got direct mappings into roff and Lout (and TeX) constructs. If you go for PostScript or PCL, you'll have to implement them yourself.

Depends on what you are aiming to output. Plain text monospaced output? There are always Perl formats ...

Arrowhead touched on troff. It's a good lightweight system, with excellent typographic control. The excellent tutorial book Unix Text Processing is available for free download from http://www.alltel.net/~kollar/utp/.

You might want to look at PDF. If you need to create simple PDFs from scratch, PDFlib is your (admittedly low -level) friend.

If you are getting forms in as PDF blanks (it happens), I've seen clever things demonstrated with PDF::API2.

Most workflows accept PDF as a native file format. Good luck!