Simple: out of curiosity, I found a 11MB file that compresses to 200k, 55:1.
I'm curious if that was the Holy Bible, Jochen? What about Alcoran or Torah for a change?
Bad design IMHO (what about style sheets?), but in any case this is criticising the big software vendor who designed chm, not you
From my standpoint, every little thing that's based upon interpreted
markup languages is bad by definition. Aside from being able to compress deeply and thus to transfer efficiently over cable and air, interpreted plain text is horrible to work with. Those innumerous tags are aesthetically and practically much, much worse than the C language's dreaded curly braces, and the sheer amount of them is the cause of ever-increasing enthropy that every such file potentially brings along to the system. Take
any Linux GUI: it
may look nice -- at a first glance and in some rare cases -- but it is entirely impractical and ungreatful to work with design-wise and it is the direct cause of most crashes many more or less advanced GUI programs, let alone the entire frameworks such as GTK & Co. themselves, experience on that platform.
Laurence's H&M project is triply vulnerable in this regard. First off, the sources are designed in XML. Then they are translated together with their XSL style sheets to HTML files. And finally these are in their turn utilized by JS scripts that include yet more markup for the HTM topics to display as intended. How stable such a system can be, do you think?
"The big software vendor" once took a questionable decision to try their hand at the newly emerging markup language market and (quite expectedly) failed. But the mere sense of responsibility before the nearly two-billion strong user base regarding the backward compatibility of their system is what really keeps the CHM concept still going today.
I should also mention that 10..12MB large CHM files like Win32.chm (~100MB of raw data) are probably a physical maximum to what the 12 y.o. stock HHC compiler is able to handle. It was a nightmare experiencing regular system memory crashes in the HTML Help Workshop environment (under any Windows OS I had) with every three new entries to the file's TOC back in 2014, and it is what I'm seeing again working on this recompilation now. Oh my...
You take everything personally - a bad reflex, my friend ;-)
I'm just trying to take everything I'm working with s.e.r.i.o.u.s.l.y. Is that really such a bad reflex, do you think?