I am starting to think this whole mixing of HTML and math is not practical.
Not for a document with 50,000 equations. I can imagine some sort of checksum-based infrastructure such that only changed math gets reconverted, like Martin described, but imagination is all it is. I myself will never have the energy to implement anything like that. But why do you need to generate the HTML every time, as opposed to developing/writing doc using PDF and then just generate HTML at the end, when you are ready to release/publish it? Another thought: can you break the document into parts? Such that you only need to compile something smaller while you're working on it, and only compile the whole thing together at the end. I don't believe tex4ht knows anything about \includeonly, but with regular latex and a document in pieces, you could potentially use that in order to cut compilation time even more. Info at, e.g., http://svn.gnu.org.ua/viewvc/latexrefman/trunk/latex2e.html?view=co#g_t_005cinclude-_0026-_005cincludeonly Best, Karl
