.pre-process-main.pl can be simplified or rolled into Wattsi

[domenic]

I'm idly investigating if we can simplify html-build by removing the Perl scripts. Here's at least one piece of low-hanging fruit.

If you run

HTML_SOURCE=html HTML_CACHE=.cache perl .pre-process-main.pl < html/source > .temp/source-expanded-1

and diff the results, there are only three transformations actually performed:

• Replacing <!--REPRESENTS article--> and friends in https://html.spec.whatwg.org/#usage-summary-2 with the values defined elsewhere in the document
• Replacing <!--BOILERPLATE cldr.inc --> and friends with some files from the build process
• Replacing <pre>EXAMPLE offline/clock/clock.appcache</pre> and friends with their contents from the demos/ folder.

All the stuff in parserExpander and its sub-function expand appear to be unused.

I think we should add a facility to Wattsi to do arbitrary file insertions, e.g. <!--INSERTFILE filename-->. Then we can eliminate the <!--BOILERPLATE x--> and <pre>EXAMPLE x</pre> specializations of this from the Perl script and leave it doing only the more complicated <!--REPRESENTS --> business.

The REPRESENTS thing is part of a more general class of problems (basically, substituting text or HTML from one part of the spec into another) that we could try to tackle in Wattsi as a separate effort, after we've done the easier stuff.

[sideshowbarker]

👍 yeah, all looks doable — and getting rid of the Perl dependency would be a nice win

[domenic]

One thing I realized is that, unless this is done in a failure-tolerant way (which seems subpar), then this would probably break wattsi-server and thus PR preview :-/.