I've been looking around the documentations but I still don't understand how ACS works&runs. + I couldn't find any explanations about the differences between ACS3, ACS4 and OpenACS, and why the packages couldn't be ported from ACS3 to 4, what's the reason?
Because ACS/OpenACS 4 was a complete rewrite of ACS, (ACS as described by "The Book" (http://philip.greenspun.com/panda/)).
You can find some history on OpenACS (and ACS to some extent byt looking at the history page (https://openacs.org/about/history) on this site.). Really breif: OpenACS was a Postgres port of ACS. When ACS-TCL was let go by Arsdigitia, OpenACS because the offical ACS-TCL code). A quick search of the web would have shown sites like http://eveander.com/arsdigita/ for some old history of ACS, and of course RedHat (who bought some of the old aD has some archives at http://ccm.redhat.com/bboard-archive/).
You'll need to read through them if you want to understand ACS, OpenACS and the changes that have gone through.
Anyway, so I attempted to look at the code. It's big, too big if you ask me, don't know where's the head, where's the tail. So I'm kind of overwhelmed, I'm stuck. I think what I need is the bare-minimum of *ACS (code+sql), forget the packages, or whatever, I just need the thing that's behind it all, I don't know if you guys understand what I mean... any help?
Of course the code is big, look at what it does. If you do a bit of research you'll learn that what is behind OpenACS is AOLserver, SQL, and TCL. Personally, and you'll probably take this wrong, but I don't think YOU understand what you are asking. If you want the "bare minimum" look at acs-core, but like UNIX itself OpenACS is built from from the "many parts working together" model, not a monolithic core of everything. It is possible to build from just the core but most of the features you'll need to develop for OpenACS have been moved to packages. The system is built on top of the core model.
