[Divunal-devel] Faucet on Windows

[Glyph Lefkowitz]

Although I agree in principle with the point that Evil Ivan has made here,
the fact is that *our* users should not need to know how to use a JVM.
It's nice if they do, but it is our responsability as developers is to
make it so that the program Just Runs.

If you're an author and you can't figure it out, we'll provide a minimum
of assistance and then you're on your own, but if you're a user, well...
a user doesn't want to pay you to hear that they have to figure it out
themselves because if they can't figure out a JVM they're not smart enough
to play.

On Fri, 25 Jun 1999, Evil Ivan wrote:

> --- James Knight <jknight@MIT.EDU> wrote:
> > At 6:31 PM -0400 6/24/99, Evil Ivan wrote:
> > >Part of the "responsibility" of working with Java
> [snip snip snip]
> > Bullshit. Needing to know anything about a JVM in
[snip]
> > means the programmer messed up.
> [snip snip snip]

Yes, it does, and the programmer ought to be more careful.  It's my fault
that the Windows version doesn't run -- and I am sorry about that.

Hopefully once we have a GTK version it'll run more smoothly and require
less download time.  (Raffi?)

> That attitude is what's ruining the computing world.

This attitude is also what's *running* the computing world.  If you want
people to learn to use a JVM and enter command line options correctly, the
best way to do that is to give them correct, working examples, and the
option not to learn.

If you give them broken examples, such as our current runscript, they will
probably just become frustrated and decide not to use the software.

The main reason I opted to ship the thing with a runscript at all is
because it is a laboriously long commandline to type in a shell w/o tab
completion (We're already making users download the 10M JRE to run this --
I'm *not* making them download CygWin! ^_^)

> We DO expect people to know simple things about what OS they're using
> when they install a game or other software application.

Do we?  How many windows users do *you* know who can actually figure out
where they installed an application when they have to find it later?
Don't they usually say "in the start menu"?

> Then why shouldn't they be expected to know how to set a CLASSPATH and
> correctly type a full package name? Hell, I'm not saying they'd even
> have to know that; all THEY'D have to know is to read a README file.
> 
> -phil

Perhaps the best solution to this is to stop giving a (broken) runscript
out for Windows and just explain how to get it running in the README --
for Linux, the runscript is easy to write and easier to execute, and it
can give a detailed error message if it can't find the JVM instantly (none
of these 15M runscripts like the ones in Swing...)

The best solution, of course, would be to have a program that the user
didn't need to perform a huge download to install and run -- either by
distributing this via CD with its own JRE, or by making a smaller, native
version (reference posts to this list about getting the GTK+ faucet
running on windows).

Write-once, run-anywhere is a great idea, but not if you need to offer
Java literacy courses to people just to get the software functional.

Finally, posts on Slashdot about TR recently have been complaining that
it's impossible to get running, even on Linux (and these aren't all
lUsers, some are other java developers).  Has anyone on this list had (and
fixed) similiar problems getting the pump to just start up?

If so, I'd really like to hear about them and fix them before we release
1.0 (this saturday).