Hi all,

I've arrived here at the advice of a developer of my fave schematics
tool, who likes this lean distribution and control over its workings.
I share his interests if not his abilities.

I have a goal of standing up an IC design platform comprising the
three key design tools I use, and a minimal graphics (i.e. more like
twm / fvwm than Gnome / KDE). I would like to make a deployable
archive which covers all the dependencies of those tools, and the
whole OS, such that I could leave it on a hard drive in a drawer and
install clean, at will, on any PC-architecture machine.

I've been using the more popular, low-information-user distros just
for ease of install but I believe I am paying a performance price
(GUI bloat on GUI-intensive activity) and also find that these installs
leave out a lot of the tools' dependencies, so now rounds of update.
This might well "break something" if I do not bundle up all the time-
coincident versions of the (presumably) working distro and tools and
dependencies.

I'd like to ask for advice on how such an effort ought to proceed.
Perhaps it is to just download every .deb (?) that the package tool
(? - this is my first day knowing that Devuan even was a thing)
says I need, to a side archive. But that wants some serious
attention at every step.

I am wondering whether there might be some "choose your own
adventure" type script that can pull and arrange "top to bottom"
install file-pile based on some input regarding packages and third
party software identified by the user? Or, since HDDs are cheap
and well exceed the software load, just "every damn thing in the
distribution", to pull from offline at any future point to install a
working, whether or not up-to-datest, design system on demand.
It's my belief that self-consistency is key and a stable system
could work forever update-free if that is the discipline.

Any subsequent updating should be consensual, not forced. I
am after a goal of eliminating software obsolescence as a chip
design / product life cycle threat (having lived through some
$$$$$$-tool earthquakes in my career, and then dealing with
the aftermath).

The two styles of machine I envision setting up are

- A dual 4K "mostly just graphics" machine that will VNC to a
more-cores, more-memory, more-disks server

- That server, with boot-to-console but a good X service bandwidth,
optimized for execution of circuit simulators and other "crunching"
along with graphics rendering for X service at its end.

Any advice is welcome, clue deficit should be obvious.