Re: [MiNT] Where should a user put GEM apps

[Jo Even Skarstein <joska@online.no>]

On Tue, 2011-01-11 at 07:10 +1100, Paul Wratt wrote:

> 1st rule of programming, the simpler and more flexible you wish a user
> interface to be, the more complex its underlying structure becomes. It
> can not be avoided, it is exponential, it is a fact of life, just like
> in a marrage :)

It can be avoided. If the code and the structures gets very complex,
it's time to look at the design. You can probably break down the problem
further and make things simpler.

> Even those Software Centers you mentioned are full of scripts

I'm fully aware of that. These scripts runs on GHz computers with
virtual memory and gigabytes of RAM. If you want to invoke a shell +
shellscript + some binaries each time you swap keyboard layout, that
would probably work on these machines. On an Atari running FreeMiNT, it
would make you tear your hair out.

> This is fine for "zips", not so for stuff from repo packages which
> require structure

I hate that stuff. Having an application's files spread all over a
filesystem is not "structure". It's madness. I understand that it's
unavoidable when you port stuff from the unix-world. I accept that you
have to adopt to the unix way of messing things up when you use their
stuff. But I can't see a single good reason to complicate
GEM-applications in this manner.

> It would therefore be nice for package maintainer to be able to place
> environment variables in paths for installation purposes, where the
> end user could then choose there own filesystem structure or layout

That would in many cases (when we're talking about ported unix-stuff)
require a recompilation of the programs to install. Not very
userfriendly to me.

Jo Even