Re: NSToolbar (was Re: Portability/Compatability betweenGNUstep<---> Cocoa...)

[Lars Sonchocky-Helldorf]

Jeff Teunissen <deek@d2dc.net> wrote on 14.01.2004 15:21:03:

> Lars Sonchocky-Helldorf wrote:
> 
> > 
discuss-gnustep-bounces+lars.sonchocky-helldorf=bbdo-interone.de@gnu.org
> > wrote on 14.01.2004 05:52:07:
> > 
> > > Lars Sonchocky-Helldorf wrote:
> 
> [snip]
> 
> > > > most of the time I read something from you on the list it is a 
rant
> > > > against any new feature to GNUstep that is not covered by the
> > > > OpenStep specs.
> > >
> > > It appears to me that you are somewhat unfamiliar with the concept 
of
> > > evolution vs. revolution.
> > >
> > > Why do I feel the need to say this stuff? In part, it's because it
> > > saves me from having to start a fork; in part, it's because being a
> > > clone of Cocoa (or of OPENSTEP, for that matter) is a sure-fire path
> > > to perpetual obsolescence for GNUstep.
> > 
> > O.k. I see you are asserting some things. But can you also 
substantiate?
> 
> Of course I can.
> 
> Remember OS/2? Even though lots of people used it, even though IBM 
created a
> very good Win16 (and later a Win32) portability library, vendors 
generally
> didn't port their Windows software. There were a few ports, but not 
nearly
> enough.
> 
> How many vendors have ported their Windows software to Linux? At last 
count,
> one, even though there exists a mostly-complete Win32 API (libwine) for
> Linux/BSD on x86...one which can even use DLLs. MFC even works, and 
there is
> ONE vendor.

Well to me this is a completely different story. Look, when those apps 
already run there as a Win binary it is logical that there is no need to 
port anything, the vendors would say: "It runs, where is your problem?" It 
is also important to note that those Win APIs are not the primary API of 
the target System, that means that you don't get real support for all the 
OS features and the look and feel might conflict. This pisses off OS users 
("why should I buy this OS when I can't use its features anyway") and 
makes Software vendors lazy. More comparable to VirtualPC on the Mac.

GNUstep is different here. It provides the primary API to the given 
System. So it is more like a cross platform solution where one part is 
made by another vendor. So it competes more with Qt/C++ (which is also 
available for the Mac meanwhile) than with Cocoa (there are some Cocoa dev 
that consider switching to Qt/C++ because of the lack of a "usable cross 
platform (here they mean Windows) solution" for Cocoa (YellowBox/NT still 
exists somehow (if you install WebObjects on Win you'll get it, but you 
can not get a License for distributing it)) Here GNUstep fits in. I think 
GNUstep and Cocoa can get in to a win-win situation if both decide to 
colaborate: GNUstep gets developers and apps, Cocoa gets a cross platform 
solution. Both sides just have to overcome prejustices and arrogance (and 
this exists on both sides, believe me).

I think only the following situation would "kill" GNUstep: Apple offers 
YellowBox/NT, YellowBox/Linux and maybe YellowBox/Solaris. Even if it 
would cost some royalty and would be only available as runtime binary (no 
development on Linux etc. possible). On the other hand Cocoa devs would 
stop bugging Apple for this if GNUstep would be the cross platform 
solution of choice for them.
 
> 
> Lessee, what else is there...oh yeah, how about the fact that GNUstep 
can
> never possibly implement the functionality that Mac OS X developers 
expect?
> It's an impossible task. Apple releases a new OS X every 18 months, and 
they
> have teams of people working full-time on new stuff. GNUstep hasn't even
> fully caught up to Rhapsody, much less OS X 10.1. Further, their new 
stuff
> is nearly all patented out the wazoo, so even if people WANTED to do it,
> they couldn't.

I don't know what stuff is patented, but all the stuff for which prior Art 
exists (which invalidates any patent) and is in heavy use should be 
implemented. 
Of course it would be good to have the full OpenStep specs implemented as 
a milestone at a certain point. But how do you want to enforce this? You 
can't forbid somebody to work at a certain class he/she (no she here 
anyway - o.k. that's another story ...) wants to have nor can you force 
somebody to implement certain things. This is a project where everybody 
works on voluntary. The only result of your anger in this relation is 
turmoil, anger again and resignation. Political fundamentalism most likely 
scares people that could/would like to help away.

> 
> [snip]
> 
> > > But mostly, it's because what I'm saying is true.
> > 
> > Oh, I understand! You are the holder of the ultimative and universal
> > truth! I bow low before you!
> 
> Nice hyperbole. Are you going to be serious, or should I just tell you 
to
> fuck off and stop discussing it with you?

Maybe I forgot a "SCNR" or a smiley here. This was meant to be sarcastic.

> 
> [snip]
> 
> > > Please keep religious arguments out of this -- I have no religion,
> > > least of all a pathetic Cult of NeXT (a subset of the 
equally-pathetic
> > > Cult of Steve). My opinions -- all of them -- on user interface 
design
> > > (graphical and otherwise) have been reached through research and
> > > reason, not faith.
> > 
> > Who did the research? You? Or were it NeXT and Apple? Which company 
was
> > it to publish "Human Interface Guidelines" for their OS first?
> 
> All of the above, plus many other HCI researchers.
> 
> Apple were the first company to publish human-computer interface 
guidelines
> for a graphical operating system (and have repeatedly broken them since 
NeXT
> took over Apple).

There is a crack through the middle of Apple. After Apple bought NeXT and 
NeXT took over Apple there where two conflicting HIGs. Which one should 
survive? If you ask the NeXTies they where opting for NeXTs HIG, opposite 
the Apple die-harts. To not completely oust one party they settled on a 
compromise, which makes no one really happy. Some things already changed 
since the public beta other things where taken back again and some 
revived. Still lots of things need to be integrated. I think it will take 
some years and OS versions till the dust has finally settled.
http://www.asktog.com/ has some interesting comments on this (also 
criticism)

> NeXT followed up, having improved on the Mac's UI in
> certain spaces (that is, big screens -- the original Mac UI remains
> altogether superior for small screens). 

And on big screens there should be sufficient space for a toolbar. And if 
you _really_ dont like toolbars, you can hide them.

> In the same time-frame, Microsoft
> kinda threw together a GUI by throwing elements at a wall and seeing 
what
> stuck, researchers at IBM designed the Presentation Manager system, Sun
> developed Open Windows, and the UNIX vendors co-developed Motif.
> 
> The original NeXT GUI was good, but had lots of rough edges that weren't
> fully hammered out until the third major version. And unlike most 
systems
> (including the Mac), NeXT's later designs got more simple instead of 
more
> complex.

If this simplicity doesn't get in the way I want to acomplish several 
tasks, it is good (not like those "Simple Finder" or even the older 
"AtEase" (a very simplistic graphical shell mimicking a file cabinet) 
stuff Apple created which hampers the experienced user) Maybe it is better 
to say that the interface should be "natural" (I don't know a better word 
here, but in nature everything is as complicated as it's needs to be and 
not simpler (dolphis, whales and orcas ;-) for instance despite mamals 
degenerated/lost their hind limbs, they where no longer needed in the 
oceans)

> 
> > Btw. somebody already pointed out that toolbars have been part of the
> > NeXT Interface ever since (even NSToolbar as private class)
> 
> Both of them (two people -- Phillippe Robert, and Greg Casamento -- not 
one)
> were wrong, and as I said earlier, I don't have a problem with toolbars
> being in GNUstep.
> 
> To expand on that first sentence:
> 
> NeXTstep did not have toolbars.
> OPENSTEP did not have toolbars.
> In the OPENSTEP days, NeXT had a private framework that implemented a 
class
> called NSToolbar, which was nothing like the class that we know of today 
as
> NSToolBar. NSToolbar provided Mail v4 and Project Builder with their
> toolbar-like icon lists, which were: 1. not user-configurable, and 2. 
just
> another view, which is the reasonable way to implement them in a NeXTish 
UI.

O.k. maybe it was not _the_ NSToolbar we know today, but what does this 
matter to the enduser? The look and feel is the same. What do you don't 
like with the configurability of the toolbars?

> 
> > > Why have I not "jumped on the bandwagon"? Because the alternatives
> > > that have been presented are more ugly, less consistent, and less
> > > efficient.
> > 
> > Again, which OS is known to have the most consistent user interface?
> > Windows? Unix with CDE, KDE or GNOME, possibly Motif?
> 
> NeXT has been consistently regarded by researchers as the most 
consistent
> GUI computer user interface in history. Macintosh System 6.x comes in
> second, followed by System 7.x. Few others even come close. The NeXT
> interface was so good that when it came to design Chicago, Microsoft 
ripped
> NeXT off shamelessly, making it the first (and possibly last) time that 
the
> Mac wasn't their primary inspiration.
> 
> > You purpot a lot but I miss the prove.
> 
> That's "proof", and I have given it before, the last time this was
> discussed. Do try to keep up.

oops.

> 
> -- 
> | Jeff Teunissen  -=-  Pres., Dusk To Dawn Computing  -=-  deek @ 
d2dc.net
> | GPG: 1024D/9840105A   7102 808A 7733 C2F3 097B  161B 9222 DAB8 9840 
105A
> | Core developer, The QuakeForge Project http://www.quakeforge.net/
> | Specializing in Debian GNU/Linux http://www.d2dc.net/~deek/
> 
> 

Lars