Independent differently-configured instances running concurrently

[Hans BKK]

I'm investigating how to run completely separate instances of emacs 
concurrently.

For example,

 * highly customized but relatively stable "production" setup used for 
exploring source-code, docs, creating my own self-doc-set

 * basically unmodified "base" setup for learning vanilla and comparison 
purposes

 * scratch-experimental setup for testing packages, comparing their mods to 
vanilla, etc


It seems many people to do this with something like:

emacs -q -title custom -l "~/path/to/custom/init.el" \
      --eval "(run-hooks 'after-init-hook)"

but from a bit of research, it seems there are two disadvantages to this 
approach. 1 the "-q" option disables saving normal customizations and 2 
non-standard eval ordering compared to running via the normal 
"~/.emacs.d/init.el" process.

So I'm looking for feedback on an alternative approach that would give me a 
stock-standard setup for each instance. So far, it seems to me that launching 
via a script that sets  a separate $HOME for each instance could work, e.g.

#!/bin/bash
HOME=$HOME/emacs/homes/vanilla
export PATH="~/.cask/bin:$PATH"
/bin/bash
# emacs

Many of these instances are being tracked via git, hence the preference for 
opening a prompt from which I can launch emacs rather than going straight in.

Note I'm a bit of a *nix noob, so feedback on the bash scripting side would be 
most welcome, in addition to anything emacs-specific.

So do you think this will work? Is there any potential problem of the different 
instances "stepping" on each other, either in RAM or perhaps writing back to 
global init configs?

Any other comments welcome, expect perhaps those along the lines of "your 
learning process is too complicated" 8-)