discuss-gnustep
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [OT] GentooLinux GNUstep Developer


From: Chris B. Vetter
Subject: Re: [OT] GentooLinux GNUstep Developer
Date: Wed, 26 Nov 2003 20:02:15 -0800

On Thu, 27 Nov 2003 03:21:27 +0000 (GMT)
"M. Grabert" <xam@cs.ucc.ie> wrote:
> On Wed, 26 Nov 2003, Chris B. Vetter wrote:
> > Take a look at http://boblib.sourceforge.net/ ...
> >   As wonderful as GNUstep is...let's face it, it's huge. The
> >   installation process isn't just a matter of the "./configure
> >   && make && make install"- no, that would be too easy.
> >   Instead, one must install the makefiles first, followed by the
> >   base. THEN one must configure the defaults database for timezone
> >   and such, install the gdomap service at boot, and run an
> >   environment initialization script at login for every user.
> >   Not to mention the dependency hell from the very start.
> Don't speak about dependency hell.
> GNOME/KDE? I agree.

In case you haven't noticed, the above was a quote.

> GNUstep? I requires objc support (naturally), X11, maybe openssl.
> and of course a development tools (make, binutils and so forth).
> Nothing much really ...

Wrong.
You need quite a few more libraries in order to get GNUstep work
properly, ffcall, libffi, libaudiofile, libart_lgpl, freetype2,
gmake (on BSD), tiff, openssl, libxml2, ...

Those are just the ones from top of my head, not counting the libraries
needed to install those libraries.

Plus, (at least) on BSD, in order to install the documentation as well,
you need to have teTeX installed, which adds another whole slew of
dependencies.

> > While this isn't entirely true, he/they are right. GNUstep needs to
> > be easier (for a user John Doe) to install. It's easy enough for a
> > developer and/or someone who essentially know what he's doing. While
> > you CAN argue that John Doe isn't the target, I disagree -- because
> > the majority doesn't know, doesn't want to know and doesn't care
> > enough how and why. All they care about is "does it install in one
> > go? if not, I'll ditch it".
> What is wrong with compile-all ?

For one, it checks for gcc3 (as /usr/bin/gcc3) -- if it can't find it,
it assumes 'gcc' -- bad, because (at least) on BSD 4.x, 'gcc' defaults
to version 2.95.x (x = 3, I think) that has NO Objective-C support
installed.

Second, compile-all installs back, which is fine, but (just IMHO)
back-art looks nicer...

So, bottom line is, you NEED to know about that, otherwise you're either
stuck or get a system you may not like -- and discard the whole idea.

Sorry, if you (in general) don't like that, but that *IS* how people
react -- I see it every day at work.
If software doesn't work (as expected) immediately, people get
frustrated and ditch the thing -- especially if it doesn't cost them any
money. They just go and look elsewhere.

> > On the other hand, no one will give a sh*t, whether GNUstep
> > eventually installs smoothly on Debian, Gentoo or
> > FlavourDistributionOfTheWeek if there's nothing, no Tool, no
> > Application, no nifty Thing that draws attention.
> Yes, true.
> We need a set of popular/usefull GNUstep apps bundled, and have a
> compile-all for them aswell.

Correct. People may look at GNUMail, GWorkspace and Gorm and think "oh,
pretty, I want that" but to install *another* "X11 toolkit" is almost
not worth it, as they very likely already have GTK (GNOME) and/or QT
(KDE) installed.

So why bother? Just for less than a handful of apps? Hardly.

-- 
Chris




reply via email to

[Prev in Thread] Current Thread [Next in Thread]