HTML minification modifies content re: whitespace

[connorjclark]

This only happens for release (debug does not minify HTML).

Related code:

emscripten/emcc.py

Line 3661 in 011fabe

def minify_html(filename):

Can be avoided with -s MINIFY_HTML=0

Personally I don't think emscripten should bother minifying HTML at all. For my shell file, minification reduced the file from 10KB to 6.5KB, so the savings just aren't worth it. And given the complexity of the minify_html function (a lot of flags passed to this external tool, many knobs to turn) there may be other unwanted semantic modifications going on here.

[DanielRuf]

7fcc11a

#8058

Maybe there were reasons. Since this can be disabled with the documented s-flag it should not be a huge deal.

[kripken]

I think there are cases where HTML minification helps significantly. cc @juj

[DanielRuf]

@kripken please see my opinion at terser/html-minifier-terser#122 (comment)

[juj]

Removing the collapseInlineTagWhitespace minification does sound good. The intent of the default set of minifications is not to do any potentially unsafe minifications, and when I was looking at the list of minifications to do, there was no good documentation that I could find to see which ones were unsafe and which ones weren't. PRs welcome!

[sbc100]

Are folks really depending emscripten to do HTML minification? It seems that most serious users of emscripten are probably not using emcc to generate their html.. maybe I'm wrong? Its only users who also specify --shell-file= on the command line who would benefit, IIUC? Is it common for emscripten users to use -o <foo>.html + --shell-file?