GNUstep 0.6.5 freeze

teferi writes " The GNUstep project, a GPL'ed implementation of the OpenStep environment, has gone into a code freeze for the 0.6.5 release. The base library is 94% done, and the various parts, including the DPS/DGS graphics backend are coming along well. "

Comments

That's great...

[Mr. Piccolo]

but where are the applications?

There's like maybe two that have been released. Having the library is great, but if there's no real use for it, no one will care.

I don't know if there are very many open-source NeXTSTEP apps out there either.

Re:That's great...

[QuMa]

I think afterstep uses it. It puts it's preferences in ~/GNUstep, so...


BTW, if you don't use afterstep: It's real nice, try it!

Re:That's great...

[Anonymous Coward]

Well, no.

Afterstep is a window manager. it could put its prefs in ~/Windows98, but that wouldn't make linux a microsoft product.

Windowmaker is the official GNUStep Window Manager (and also the official GNU Window Manager), and has a look/feel more reminiscent of NeXTs.

Re:That's great...

[Anonymous Coward]

Where are the applications?

Very simple.. Once GNUstep is 100% done, it should
require little effort to port tons of NeXTSTEP/OPENSTEP/MacOS X
apps to GNUstep... The process is already on its way to port
some of those.

Not to mention the fact that ProjectCenter.app, once completed, will
provide a really nice environment to create new GNUstep apps with
a minimum of effort. (kinda like Steve Jobs demonstrating how to
create a word processor in 5 minutes using NeXSTEP's ProjectBuilder.app).

The final word is that the applications are just around the corner...

Cheers.

This is AC of Borg.
Accounts are futile;
Trolls are irrelevant.
You will be slashdotted.

Re:That's great...

[Ian Bicking]

Ironically, while Window Maker is part of GNUStep, it doesn't use any of the GNUStep functionality, being written entirely in C (not Objective C). I guess this was done because the author (Alfred Kojima, I think) wanted to make it functional apart from GNUStep, and before GNUStep -- both of which have been accomplished.

But as a result, it's only a matter of policy that Window Maker is part of GNUStep, while I imagine the rest of GNUStep will be connected at more of an architectural level, being based on Objective C, the Foundation libraries, and all that.

Re:wmaker?

[jcr]

GnuSTEP aims to re-implement the entire OpenStep spec under GPL. When this is done, you can expect about 200 NeXT apps to suddenly appear on Linux, FreeBSD, etc.

-jcr

NEXTSTEP / Mac OS X relationship?

[Erich]

What is the relationship between NEXTSTEP and the new MacOS X? If the OpenSTEP project is successful in making a NEXTSTEP-compatable system, would it be trivial to compile the new MacOS stuff to run under it?

Re:NEXTSTEP / Mac OS X relationship?

[smart2000]

Mac OS X has several programming interfaces, one of which is called Cocoa. Depending on what you combine with it, this is also similar to OS X Server, Rhapsody, Yellow Box, OpenStep and NeXTSTEP.

NeXTSTEP (an operating system) was first, and had a damn good programming model, but originally only ran on NeXT's hardware.(which was pretty sweet, but also pretty expensive) (and also one version on RS/6000, running on AIX, again fast, sweet and expensive.).

NeXT then began making moves to dump hardware and make NeXTSTEP available on Intel hardware.

Right around that time, they also started development on OpenStep, which has a very similar object hierarchy to NeXTSTEP, but uses a better object allocation model, and has renamed many methods to make the API "cleaner".

To move an application from NeXTSTEP to OpenStep, you ran a series of scripts that would convert to the new API.

OpenStep was made to run on several OS's including Solaris, Mach (from NeXT), HP/UX and Windows. OpenStep meant two things at the time, both the API and the NeXT delivered operating system as a whole. One was called OpenStep, and the other was OPENSTEP. You guess which was which.

Any ways, Apple bought NeXT so that the NeXT management team could take over Apple, and now all that API is part of OSX Server and soon OSX.

It is easy to move a program from NeXTSTEP to OpenStep or OSX Server. I moved Xox, an arcade style shooter with a few days of work.

It is trivial to move the average program from OpenStep or OSX Server to GNUstep. In many cases the same code compiles on both.

We moved our entire development over to GNUstep, and haven't looked back yet. We found the Foundation kit to be more stable than Apple's and easier to explore.

D'oh

[Erich]

I mean GNUStep, not OpenSTEP. OpenSTEP == NEXTSTEP. I knew that. Sorry.

Hooray!

[Anonymous Coward]

This is great news. IMO, GNUStep is (will be) more important than KDE or GNOME, even though they get more attention. OpenStep is a well designed spec. Programming for it is great. The only thing that matches it could possibly be BeOS (both are object oriented and similar in many respects, though BeOS is multi-threaded to the max).

Plus, OpenStep has a better look & feel than KDE or GNOME*
(Yeah, I know, OpenStep is a spec, not an implementation. It still looks nice).

they are there, just...

[Anonymous Coward]

waiting to run. The problem is that GNUStep hasn't reached a point where GUI OpenStep apps can just be compiled. There are too many inconsistencies and too much not done yet in the AppKit Foundation. But once it's done, in theory any openstep app should be able to be run on linux. They also run on MacOSX, and Windows. OpenStep is the cross platform app development environment that Java wishes it was. A JVM is stupid, it slows things down to much especially considering what a hog swing is, and you have to use swing to do decent interfaces. OpenStep got it right. The basic philosophy of OpenStep is write once, compile anywhere as opposed to Java's compile once, run anywhere...

Re:they are there, just...

[smart2000]

The problem is that GNUStep hasn't reached a point where GUI OpenStep apps can just be compiled

We find most command line stuff "just builds". We type make, and it's off to the races. The GUI stuff is a bit harder. It compiles, but the interface format used by NeXT / Apple is un-documented, so you have to convert the interfaces. The GUI apps compile, they just run a bit funky.

Ah yes!

[Yarn]

I remember lusting after NeXT boxes when I was like 12 after we got our 1st 386, they looked so sweet, and the UI was so *smart*. Consistancy is something that almost everyone agrees is a problem with X, but plans to make is nice always seem to fall over. CDE is/was hell (imo), gnome and kde are diverging and converging at the same time, and I'm getting confused. SAVE ME GNUSTEP :)

Re:Ah yes!

[Surak]

Believe it or not, CDE is still in wide spread use. The engineering workstations at GM all use CDE (they run Solaris and HP-UX) and I have to agree its GARBAGE.

KDE and GNOME are good steps in the right direction.

I don't think GNUStep/OpenStep/NeXTStep whatever is really going to ever be any sort of standard, IMHO. Nice interface, nice technology, but too far from mainstream interfaces.

Re:Ah yes!

[scrytch]

Sun, of course, also uses CDE. It's also pretty much universally hated there. It does do some things well that KDE still doesn't quite manage, and gnome doesn't even try for. But overall it's ugly, klunky, and unfriendly. The file manager is a joke (except when you want to set ACL's, there it's the best interface going). The mail client is primitive (except it does IMAP perfectly, something free email clients are always lagging on). And it uses the classic motif look, which can only be described as "boxy but good" if only it were good.

All that said, CDE does some things right. Like a web browser icon on the panel that runs sdtwebclient, which acts as the equivalent of the netscape-wrapper script for netscape or hotjava. The panel isn't very flexible, but it's far more intuitive and easy to navigate than any START button or knockoff thereof.

I'd love KDE on Solaris if it were actually as functional as CDE. Give it a year or two and it'll probably get there, faster than CDE, for sure.

Re:Ah yes!

[Ian Bicking]

I don't think GNUStep/OpenStep/NeXTStep whatever is really going to ever be any sort of standard, IMHO. Nice interface, nice technology, but too far from mainstream interfaces.

I don't think the interfaces are a problem at all. NeXTStep isn't really all that novel in its interface -- basically a polishing of the Mac interface. Windows looks more like NeXT than Macs.

But that won't be the problem -- the biggest issue is, IMHO, Objective C. Now, I'm not saying anything bad about Objective C, but it's a new language to most people, and people get weird about that sort of thing. It doesn't have the hype behind it that Java does, the history and maturity of C, or the mainstream acceptance of C++. Too bad, it's better than all of them.

Re:Ah yes!

[Surak]

I fail to see how KDE is less functional than CDE. I use both on quite a regular basis and I am always saying that CDE is not as functional as KDE.

KDE's panel is very similar to CDE's (this is not by mistake). CDE *does* have the ability to customize it a bit better by hacking scripts (you can have multiple rows of icons and customize their size, for instance), but overall, KDE offers more "point-and-click" functionality right out of the box.

Re:Ah yes!

[scrytch]

> KDE's panel is very similar to CDE's (this is not by mistake).

Not where it matters, namely the drawers. All of the CDE panel's icons have arrows for the drawers above them, with the ability to add new items to the drawer via drag and drop, and the ability to put one of the drawer items on the main panel. It's extremely intuitive, and KDE isn't quite the same (possibly no worse, but I didn't feel it was as straightforward). Secondly, unless they've changed this recently, every time I click on the netscape icon in KDE, it attempts to launch a new netscape process (and whines about lockfiles, etc). To say nothing of detecting preferred browsers or launching an alternate one. Sound simply didn't work at all in Solaris KDE.

The most damning thing of all is that KDE does not understand multiheading. Getting it going on the second monitor tended to make it conflict with itself and do very ugly things.

Re:Another (L)GPLed project that will never finish

[smart2000]

Anyone who would devote any time to GNUstep is an idiot.

Please, tell us what you really think. I devote time to GNUstep because it allows me to quickly write cool software that we need and distribute it on commodity hardware that runs an operating system that supports the cards and features we need. I make money doing this. How does this make me an idiot?

It is possible to work with GPLed software for the wrong, and wasteful reasons. But just choosing GNUstep over the competition doesn't make one an idiot.

This project will clearly never finish.

Probably so. Like most free software it will continue to live and improve each day. However they have met several important goals, and they stuff the code as it stands is useful TODAY.

GNUstep is in the same hopeless position as projects such as GNU Classpath, forever trying to catch up to an evolving standard.

At the time GNUstep started, OpenStep was supposed to become a certified standard. Since then Apple bought NeXT (who saw it coming?) and is trying to take their existing code base and justify the purchase. Whether they actually ship it is another question.

.. snip.. The bizarre GNUstep mission page claims that the project will accomodate both commercial and free software. How's that supposed to happen with GPLed libraries?

The libraries are LPGL, not GPL. There is nothing in the GPL that prevents commercial projects from using the source. There is nothing that requires them to distribute the source unless they distribute binaries. For MCCA users, there is no conflict

About the *step interface...

[extrasolar]

My knowledge of the *step interface is based on WindowMaker. I read somewhere that much of the design of the *step interface was based on large computers with large monitors. That was why they have root menus instead of a global menu. Is this true?

Re:About the *step interface...

[rangek]

I read somewhere that much of the design of the *step interface was based on large computers with large monitors.

I can believe this. I started using X under Solaris and CDE (and I didn't think it was too bad. Yeah the file manager sucked, but I never really use a file manager... (am I missing something?)). Then I used a customized fvwm2 set up under RH5.x. I liked that well enough, since I constantly twiddled with it over the course of a year or so.

Then I installed RH6.0 and dealt with GNOME for a while. It is great for newbies, most every one in my lab uses it. It is just point and click configurable enough for them to make life livable for themselves. They are all used to fvwm-95 or whatever that default config was under RH5.x (YUCK!), so GNOME/E is just peachy as far as they know.

I soon got sick of GNOME though. It just wasn't configurable enough. Granted, I didn't spend a lot of time figuring it out, but shit, I have work to do man. I can't just fuck around with my WM all day.

So I looked around and decided to try GNUstep (WindowMaker). It is awesome on my machine at work (1600x1200 on a 21" monitor). But when I installed it at home (1024x768 on a 17") I really wanted smaller icons. You can change the icon size, but then you have to use pictures on those icons that are the right size. So I know it is possible to do, but once again, too much effort. WindowMaker should be smart enough to use a 32x32 set of xpm's or whatever if I tell it I want 32x32 icons.

I still like WindowMaker and GNUStep. But I think it would be pretty impossible to use on a system Now maybe we are all complaining about something that is easily fixed but we just don't know about. Any WindowMaker/GNUStep guru's know something we don't?

Re:About the *step interface...

[extrasolar]

NS has this kind of menus because this structure offers better usage then the orginal menu bar idea *if* your screen is large enough. These vertically stacked menus do not waste space anymore than a standard menu bar do. Being able to move then around makes them *very* convenient. I prefer them even today over every menu bar/start menu crap.

That is a very good point. It is probably a reason why WindowMaker has the Root Menu able to be dragged around, most users don't have that much screen space.

This NextStep way of the user interface is interesting to me. I only now realize how much of a mess our interfaces are that try to *hide* the information from the user.

Re:About the *step interface...

[Anonymous Coward]

NS has this kind of menus because this structure offers better usage then the orginal menu bar idea *if* your screen is large enough. These vertically stacked menus do not waste space anymore than a standard menu bar do. Being able to move then around makes them *very* convenient. I prefer them even today over every menu bar/start menu crap.

Like many NeXTSTEP idiosyncracies (scrollbars on the left, close box on the top right, 17" minimum screen, black hole / recycler icon instead of trash can, etc.) the NeXT menu is touted for its "benefits" -- but was actually not designed for that reason but instead to avoid lawsuits from Apple. As part of his severance agreement, Steve Jobs agreed to a number stipulations from Apple about the then-unnamed NeXT computer. One was, humorously enough, that the NeXT cube had to be more powerful than any of Apple's present offerings (which it was, with a 25 MHz 68030 processor, soon changed to 68040).

Although there are lots of nice features about the NeXT menu bar design, and I personally love it, it nonetheless has three flaws which override all of its niceties and make it distinctly inferior to the Mac menu bar.

1. It violates Fitt's Law. (If you do not know what this is, don't respond to this message until you look it up on the web; you'll just look foolish otherwise.) The single most serindipidous feature of the Mac menu bar (Apple lucked out) is that it is very easy to hit with the mouse, because the menus are backstopped by the top of the screen. While NeXT (and Windows, and XWindows) users generally choose menus by carefully and precisely moving the mouse to the menu button and pressing, Mac users generally choose menus by slamming the mouse up to the top of the screen, then dragging down, all in a very crude, hit-the-barn-door fashion. They can get away with this because it is impossible to overshoot Mac menu options -- it's as if they have infinite height. Published studies generally show that experienced Mac users choose menus at almost five times the speed of Windows (or NeXT) users for the same accuracy.
2. Apple's submenu-dragging algorithm is vastly superior to XWindows, Windows or NeXTSTEP. On the NeXT, in order to drag onto a submenu, you must move your mouse carefully to the right, out of the submenu parent's menu area. Don't move the mouse too far up or down! Or you'll have chosen another parent menu option and the submenu will disappear. This is also the case for Windows and for nearly every XWindows application with submenus. On the Mac, if a submenu is up, you can move at a significant angle to hit the submenu, even crossing over other menu choices for short periods of time, to reach the submenu. This is a real problem for NeXT users because nearly everything is in a submenu in the NeXT GUI.
3. The NeXT's menu design might seem to take up "less space" than the Mac menu design, but in reality it is significantly more obstructive. Like most GUIs, the NeXT GUI is built around large rectangular windows. On the Mac, the menu leaves a workspace below which is still rectangular, whereas the NeXT menu leaves a workspace in the shape of Utah (horizontally flipped). As a result most NeXT users tend to leave the entire desktop area directly below the menu unused. One occasional hack is to move NeXT "inspector windows" (information-modification dialogs) over there, though inspectors are rarely thinner than the menu width so this isn't a great solution. Usually the space, about 15% of the screen, is just left wasted.

Many NeXT users, like Windows users, avoid problem #2 by chosing menus click-by-click, rather than dragging through the menu tree. Mac users are not, of course, forced to do this. The catch-22 for the NeXT GUI is, however, that most experienced NeXT users avoid problems #1 and #3 by moving the menu nearly off the screen and using the right mouse button to pop up the menu at the mouse position -- the problem with this is that when using the popped-up menu, you must drag to choose menu items, hence problem #2 rears its devastating head again. Right-mouse-menu-choosing is very slow.

As a longtime NeXTSTEP developer and experienced GUI hacker, I love the NeXT interface. But let's call a duck a duck: the NeXT menu really really sucks in some places.

GNUstep clarification: Support GNUstep!

[RevAaron]

People seem to be kind of confused about what GNUstep is and implies, I'll try to clarify:

GNUstep is: an implementation of the OpenStep API. The OpenStep API makes it quite easy to develop programs for it, as the developer doesn't have to worry about the little things, and spend their time innovating and writing great code . It's cross-platform (between Windows w/ the YellowBox, anything running GNUstep, Mac OS X/Cocoa). It's a dream to develop with, and the Objective-C language, to me, is much nicer to use than C++ (although I think there's wrappers for Java, and perhaps C/C++).

GNUstep will: Allow for easy ports to platforms running GNUstep from source written under OpenStep, Rhapsody, or Mac OS X (using Cocoa/YellowBox). This encourages cross-platform development, and hopefully will help bring many apps to Mac OS X/Cocoa, as well as Linux/FreeBSD/etc.

GNUstep is not: a window manager or a desktop environment. Desktop environments can (and quite easily) built with GNUstep. In fact, someone is working on a NeXT-like file manager right now, which is working and developed under OpenStep, and easily recompiled on a FreeBSD box using GNUstep.



For more information, see the GNUstep website or the unofficial GNUstep website, both of which have plenty of information on the OpenStep spec, and where GNUstep is going.

In short-- definately check it out!

Aaron

Tools will pop up VERY SOON

[Anonymous Coward]

My colleagues and I wrote some OPENSTEP tools that are Interface Builder and Project Builder workalikes.

They are running right now on Intel and NeXT machines running OPENSTEP. As soon as GNUstep can host the code base, these tools will be released.

Linux is about ready to get the best application development environment on any platform--and it'll all be open source.

Interface Builder is a GUI-building tool that works something like VisualAge in that it allows you to (a) visually build connections between GUI controls and methods and instance variables in objects and (b) create new instances of non-UI objects.

Finished UIs in Interface Builder do not contain code. Instead the connections and controls are archived into "NIBs." This allows you to create and maintain UIs without having to write a single line of code. If you've ever written a Java Swing application, you know what a pain in the ass it is to write GUIs by hand.
However, the tool does generate stubs for the custom, non-UI-related objects you create. This allows you to visually create a new object, connect up its stub methods to your controls visually, then "fill in the blank" to generate the core logic of your object's methods.

Project Builder is similar to a Smalltalk code browser, with a Miller column view of your class hierarchy and an integrated editor window that displays source code and documentation depending on the currently browsed object. If you've used JBuilder, Visual Cafe, or VisualAge, you already know how this kind of thing works.

Re:Nextstep Hype

[rm -rf /etc/*]

Have you asked anyone who has used it? Everyone I know who has done Openstep development would be happy to share with you. My list? I wrote a TextEditor with search and replace, multiple undos/redos, selectable font size and style, spell checking, cut and paste, and drag and drop integration with 7 lines of code in under 4 minutes.

Show me somewhere else I can do that.

sort've

[/]

The philosophy of unix is "write once, compile on all conforming platforms" (ie, other unices, as long as you don't use any platform-specific code). What openstep/gnustep buys you is "write once, compile anywhere, including all the silly proprietary have-their-thumbs-where-the-sun-doesn't-shine platforms (ie, Windows et al). The latter is a godsend for programmers everywhere.

Interface feel

[Julian Morrison]

I've used a proper OpenStep derived OS (Apple's Rhapsody betas) and I've used the current front runner linux UIs, Gnome and KDE. OpenStep is a lot nicer. Why is OpenStep nicer? Because it's integrated. Because it has things like color wells you can drag out of onto things, document proxy icons in the window manager you can drag to the filer, app bundles that contain an app and all it's global config in a single unit, system services that dynamically detect what you have selected and offer relevant options to tweak it.

Gnome and KDE can be snazzified with themes and config and whatnot, but in the end it's mostly just chrome. They are struggling to retrofit the same degree of dynamism and integration that OpenStep had from the get-go.

Objective C

[Christopher B. Brown]

Part of the big deal comes from the Smalltalk-like structuring of Objective C.

Note that there is an Objective C binding for GTK...

The apps are at Stepwise.com.

[jcr]

There are about 200 apps for NeXTSTEP that will show up on Linux as soon as GnuSTEP is able to support them.

GnuSTEP isn't done yet, but when it is, hold on to your hat.

For a sample of the apps you'll see, check out next-ftp.peak.org, and www.stepwise.com/softrak/

-jcr

Re:Another (L)GPLed project that will never finish

[Le douanier]

was thinking of learning objective-c (objective-c==openstep?) but if i gotta buy some devkit from apple to get the full functionality and only be able to run stuff on OSX then i guess i wouldn't be bothered...

Objective C != OpenStep.

Objective C is a programming language that adds OO style programming to C but with a philosophy closer to Smalltalk than to C++ (i.e. Java also has a Smalltalk-like philosophy but with C++ syntax), thus Objective C has got great merits on its own.

OpenStep use Objective C, so it integrates well with OpenStep, but you can use it without OpenStep.

By the way, Next made the Objective C front end for GCC, and first tried to release it in binary only (they thought that releasing the .o binary was compliant with the GPL), RMS contacted them to tell them it was in violation and they re-released it under the GPL.

I have also heard that Apple has got a compiler in which you can mix Objective C and C++, must be pretty cool.

I haven't used it personally so it is things I have heard about it in various tutorials/explanations/... When I get the time I definitively must check it (I'm checking Guile right now).

Anyway, for more informations see this.