It provides a single interface to all your help docs.
So does the currently CHM system, and nearly any other helpsystem pretends to be universal if other people would just convert the existing helps :-)
But anyway, it seems that dash needs special tags embedded in the html, so you are probably looking at adding a new fpdoc backend for the RTL/FCL/LCL docs. You can look at how the CHM backend wraps around the basic html backend for that.
The latex docs (prog/user/ref etc) are a bit harder, see the fpcdocs/src/compilelatexchm.pp that mutates html generated from latex for CHM purposes.
Then finally there is the lazarus IDE help, and afaik it is only stored in the wiki, with no systematic available offline help, so somebody's else would have to explain the practicalities of that (*)
(*) yes, there are tools to downloads part of them, and sometimes that is even rumoured to have worked, and sometimes not. I never saw it work.
Then finally there is integration into the IDE. I don't know much about that either, but afaik it is currently fairly static for the existing set, and you can't embed random docsets into the IDE. How easily that can be changed I however don't know, so I don't know if that is a showstopper or not.
Note that any system that has umpteen files per docset (like each html lemma stored in separate files) is a bad match for Windows and embedded systems like Dos. Probably all non *nix systems (either by direct genealogy or spiritually). The original reason for the html to chm conversion is minutes of installation time of the FPC docs on win9x machines, and half an hour on light weight dos machines due to the large amount of files.
Modern machines are a bit better, but unfortunately SSDs without cache handle writes of many small files rather badly, sometimes even worse than a harddisk.
Recent
