Hi Deimi - wow, thank you for all the work. It all sounds excellent. Unfortunately I am going to be tied up producing the TUG conference proceedings for some time. But I will come back to this as soon as I can. For now, a couple of quick replies.
Re the separate mktex4ht.cnf, yes. Except manual maintenance doesn't sound good. Seems like we could generate the separate file outside of TeX, e.g., by grep/sed/perl whatever of the source files? Then we'd avoid the mutual dependency, I think. Agreed on no benefit to literate version of the .cnf. Also really like the idea of being able to split up tex4ht-4ht. As you saw, we had to disable our tiny attempt at beginning that with lettrine because of the mktex4ht.cnf problem. Having a different file for each "class" (and defining the classes more clearly) sounds right to me. Alternatively, if (since, I guess) the classes are essentially arbitrary, I wonder about doing it alphabetically (one file for a*.4ht, one file for b*.4ht, etc.), to avoid making random judgements. But maybe that is worse than trying to impose some kind of semantics. Thanks again, so much. -karl