Re: [MiNT] Re[2]: GEM boost

[Evan Langlois <Evan@CoolRunningConcepts.com>]

On Mon, 2005-07-18 at 20:04 +0200, Maurits van de Kamp wrote:
> There is /usr/GEM, analog to /usr/X11 :o) And for programs that want to have 
> their own directory for everything, /opt/appname would be in order. Well you 
> could always distribute both.. so an RPM for MiNT users who can then type the 
> "user unfriendly" install command and be done with it, and a ZIP with a long 
> textfile explaining you where to copy everything and what config files to 
> change for non-MiNT or non-RPM users. And then people can start to 
> re-evaluate for themselves whether MiNT is really more complicated than TOS 
> or MagiC. ;o)

If you are forcing the Linux Filesystem Standard, of course its easier
to use an installer - but be please have a GUI installer available.  You
could set the GUI installer to be the default application for RPM files
and that should take care of most GUI users.  Now, I prefer Gentoo to
RPM, and I don't use porthole to install stuff, but thats the first
thing I install when I install Gentoo for someone else.  "Here's where
you download and install new software or do updates".

As for which is easier.  Papyrus for Windows comes as a single EXE.
Here is a platform in which nearly every program there is comes with
multiple DLLs, weird config files, registry entries to install, and all
sorts of cruft.  Windows programs almost always have an installer and an
uninstaller to take care of this, but Papyrus can run from a USB drive
if you want - just carry it from computer to computer and it runs just
fine with no installer.

Which method should the next kick-ass application be distributed?  An
RPM with files all over the place, or as a .PRG and .RSC you can put
whereever you want?

Are we going to use the .desktop files from Linux for desktop and menu
shortcuts as well?  I guess MiNT developers want to start reading the
freedesktop.org site to be sure we follow all the upcoming Linux
standards since MiNT is going to be another Linux alternative.  Why did
we bother making XaAES a MINT module and not a Linux module?  Change to
a Linux kernel and you get every kernel feature you could want.