Re: Random thought...

[Gleef <gleef capital net>]

On Sat, 15 Aug 1998, Dan "Effugas" Kaminsky wrote:
> Tom Vogt wrote:
> >Dan Effugas Kaminsky <effugas@best.com> wrote:
> >> In a way, I'm arguing that there's going to need to be a standard
> >> "hintreader", if you will, much like Amaya.  KDE, for all of its faults,
> >> is a complete environment.  In fact, it is so complete that it is
> >> impossible to extend.  That's Bad.  What I'm saying is that GNOME
> >> *will* have a standard WM.  This isn't a question, this isn't
> >> something we can choose not to have.  Whatever ships with Redhat 6 is
> >> going to be the standard. 

A) You overestimate Redhat's power in the Linux world.  Whatever ships
with Redhat 6 will be commonly used, but by no means the standard.  Other
common Linux distributions are not going to go away.  SuSE and Caldera are
both VERY popular purchases, and Debian is growing as well.

B) Redhat 5.1 currently ships with two window managers (Afterstep and
FVWM2).  I see absolutely no indication that they will ship a single,
'standard' WM anytime soon.

C) In a separate email, you implied (without saying explicitly, a bad
habit) that since Rasterman is a Redhat employee, Enlightenment will be
the window manager shipped with Redhat 6.  For one thing, from everything
I've heard, the only support that Redhat is giving Enlightenment
development is CVS and mailing list space.  My understanding is
Enlightenment is a spare time project for him, and he doesn't work on it
on the clock.  Redhat paid him to do Electric Eyes (so they could dump
almost-free xv), not E.  Enlightenment is too demanding of old hardware
and missing some key features (like root menus) for Redhat to be likely to
standardise on it at this point.


> >a) you are VERY quick to convert ideas to fact. "gnome SHOULD have..." is a
> >much better way to state your opinion.
> 
> No.  Should implies I like this state of affairs.  Will implies there's
> nothing that anybody can do to stop this:  Due to the inordinate amount of
> power the American media has, in general, the amount of free press Redhat 6
> is going to recieve will be massive.

Yes, but Will implies inevitability, which I certainly don't see as the
case.  Perhaps "I fear that..." would have been better, to make it clear
that you think it's likely, but you don't like it.

The American media has given Caldera almost as much press as Redhat.
Foreign media seems to give SuSE the edge.  Slackware gets no press, but
still makes up a hefty chunk of Linux installs.  I don't think the
American Media is the most important force at play when it comes to Linux.


> >b) redhat ships with various wms at the moment. anyone from redhat here who
> >can enlighten us to why this should change? I doubt it will.
> 
> Now, as far as I've heard, Redhat has been saying that GNOME will be the
> standard UI for its next distribution.

They said it will be their standard UI, I heard nothing about when.


> We're already using a panel, Tom.  That's headlong deep into "window
> manager" territory--guess what, this is a good thing.  If Caldera has a
> complete interface and we have a mess...welp, KDE wins and we go away into
> the night.

The panel is nowhere near "window manager" territory.  Window manager
territory is decorating, controling and managing windows.  The fact that
some of the more popular window managers shipp with applications (like
GoodStuff and Wharf) that are compared to the panel does not mean GNOME
touches window manager territory.  Rather, those window managers touch
application territory.


> >> I do not want GNOME to be a bad thing for WMs, but we have three choices.
> >> Either GNOME becomes a standard but extensible windowing system(what I
> >> want), or a standard and inextensible windowing system(KDE), or it fails.
> >> It's That Simple.
> >amen
> >
> >I'm afraid you are TOTALLY missing the point. gnome is a very specific
> >project, and a LOT of the ideas you have are simply outside of its scope.
> >in the same way that a style guide has no job talking about reboots,
> >gnome is not a window manager and not a replacement for them.
> 
> So then you have a problem, don't you.
> 
> If the ideas are good, but don't belong in GNOME, where do they belong?
> Some single, specific window manager?  Oops, KDE goes ahead and integrates a
> few of them, suddenly KDE is better...

Depends on the idea.  Some ideas are better implemented on the Window
Manager level.  If those ideas are good enough, it could be possible to
add them to the list of GNOME-compliant wm requirements, but this sort of
thing cannot be unilateraly decided on by the Style Guide.

Yes, KDE's scope includes the WM.  This would hypothetically allow them to
integrate such features.  It does not make it likely, and I think that
their requirement of "a single, specific window manager" is more a
hinderance than a help.


> if the ideas are bad, then say so.  I don't think they are, Tom.  I think
> Cluehunting, Keyboxes, Minbars, and so on are advanced components that are
> damn well good enough to be standard.

I must of missed discussion of Cluehunting, I have no idea what it is.  I
have heard a little of Keyboxes and Minbars (btw, big trademark/name
recognition problem with that name).  The Minbar sounds like an alternate
Panel, and certainly seems in the same scope as GNOME. Keyboxes sound very
hard to implement properly.  Personally, from the screenshots I've seen of
Minbar, it looks too busy for me to want it as standard, it is likely to
overwhelm new users.  However, it looks like it would make a very good
alternate choice for the person who wants a panel with the kitchen sink 
(like me).


> We can either do a half assed job designing an interface, hoping that we
> don't do anybody else's work, or we can do a complete job that other can
> choose to build off of or choose to stay with where we are.

We need to design a style guide for application developers to use to make
good, consistent programs.  If we set our scope too broad, there will be
no way to implement some items, much less finish the document.  Operating
system issues (such as reboots) have no place here.  

Window Manager issues we are not at liberty to dictate.  The window
manager developers have been told time and time again that the
requirements of GNOME-compliant window managers are:
  1) Proper X Session Management support
  2) Supporting MWM window hints
  3) Supporting extra GNOME window hints
And that the extra GNOME window hints will be discussed only with their
input.  If we have something to add to that list, not only had it better
be good, but we can't put it in without more general discussion than here.
Partiularly since the WM developers have given up on this list, Maciej for
instance.


> >we have the kernel - basic system functionality
> >next layer is X - basic graphics functions
> >window manager - basic window functionality
> >gnome - inter-application consistency and interaction
> >app - user actions
> >
> >does that make sense to you? of course ALL of it is part of the "user
> >interface", but only 1/5th is part of gnome. we can make limited demands on
> >the apps, because they will rely on gnome for parts of their functionality.
> >we would be idiots putting requirements to the kernel. we would have to have
> >not one or two, but a vast collection of really good reasons to ask for any
> >changes to X. we should work using the established and existing ways to
> >interact with the window managers. if we can ask for a favor or two (gnome
> >specific wm hints), it's mostly thanks to gnome having a higher profile
> >than the wms.
> 
> GNOME isn't tied to Kernels(Litestep), nor is it tied to X(Berlin).  It can
> make "demands", if you will, of anything that pisses of the user.  I'm
> sorry.  If a computer pisses me off, I don't want to hear "well that's not
> my job".  We're supposed to describe how to build an interface that *DOES
> NOT PISS PEOPLE OFF*.  What part of MAKES PEOPLE HAPPY is a bad thing?

GNOME is not here to make people happy.  GNOME is here to make a good,
implementable network-object-desktop environment.  Making people happy is
a good thing, but leaving our scope to do so is foolish.


> You know, this reminds me of my reaction when I heard people complaining
> about how PCs in general are smart enough to recognize that a floppy isn't a
> boot floppy but too dumb to move on to the bootable hard drive.  I thought,
> "That's not something an operating system can deal with", then I realized,
> damn, that's a bug, and if Microsoft deemed this functionality a bug you can
> be sure it'd be gone in the next flashable bios.  Microsoft hasn't been too
> hot on calling rebooting a bug, because that'd require major re-engineering.
> But, guess what:  The need to reboot continually is a huge interface FLAW,
> MISTAKE, etc.  Refusing to admit it as such because of some kind of "layer"
> issue is as foolhardy as refusing to complain about the boot floppy bug
> because "oh that's the BIOS".

Saying that Microsoft has the power to dictate hardware issues is a far
cry from saying that GNOME does.  The need for Windows 95/98/NT to reboot
constantly is a flaw, a bug, a sign of poor engineering.  On the other
hand, it is not a user interface issue, it is a system stability issue.
>From personal experience, if you ignore the reboot request message, and
continue on in a Win32 environment, you risk spurious errors, protection
faults, and other signs of system instability.  This tells me that
Microsoft put in the reboot requests to hide poor underlying engineering,
rather than to make an annoying interface.  Any hypothetical GNOME port to
Win32 that touches those areas where Win32 would have to reboot, SHOULD
reboot, because if they don't, GNOME will become unstable.


> I'm not afraid to call a cigar a cigar.  If that means I step on a few toes,
> well, thing about the fact that, according to your definition of GNOME up
> there, about the only thing GNOME refers to is copy/paste, save file
> windows, and anything else that's consistent between applications.

That's not entirely true.  We have the power to tell applications what to
do, since applications presumably would want to call themselves GNOME
compliant.  But even there, if we put something stupid or impossible to
implement as a GNOME requirement, just because someone wants to be happy,
we will get categorically ignored.


> >--
> >Those who do not understand Unix are condemned to reinvent it, poorly.
> > -- Henry Spencer
> 
> Those who do not understand the few flaws in Unix are condemned to reinvent
> it, poorly.
> 
> Being afraid to set a standard--even one that we accept can be overridden
> and expanded upon--is classic Unix.

Let the LSB project determine the Linux standard.  Let the Open Group
determine the Unix standard.  Let Microsoft determine the Windows
standard.  We have enough on our hands without telling operating systems
what to do.


-Gleef