newsNews: Hotspot #8: the project Étoilé

Show feedback again
Latest News
Gna! closing soon posted by beuc, Tue Jan 31 22:31:06 2017 - 5 replies
mail subsystem upgraded posted by beuc, Sat May 23 18:17:18 2015 - 1 reply
dl/home/cvs/svn/ upgraded to debian8 posted by beuc, Thu May 21 22:12:23 2015 - 2 replies SSL certificate renewed posted by zerodeux, Thu Apr 24 07:13:06 2014 - 4 replies
Heartbleed status and measures posted by beuc, Wed Apr 9 09:47:29 2014 - 2 replies
[125 news in archive]

Hotspot #8: the project Étoilé

Item posted by Mathieu Roy <yeupou> on Mon Jan 16 15:53:36 2006.

On a regular basis, Gna! people pick and interview one project hosted here to highlight the software and personalities that drive Libre Software community.

This is the 8th hotspot, dedicated to the Étoilé project. Étoilé is a GNUstep
based user environnement.

Interview with David Chisnall, Jesse Ross, Nicolas Roard, Quentin Mathé, Yen-Ju Chen:

[ This interview is also available in HTML at ]

DC: David Chisnall
JR: Jesse Ross
NR: Nicolas Roard
QM: Quentin Mathé
YJC: Yen-Ju Chen

| Gna: Hello, Can you give a two line description of your project your grandma(s) could understand?

JR: Étoilé is a Desktop Environment that focuses on components, workflows, and collaboration. It's about simplifying, streamlining, and sharing.
DC: We are a group of people who have all come to the conclusion that modern computer systems are almost unusable - hacks piled on top of hacks to produce a design and an interface with very little thought for what a regular user wants to do. The purpose of Étoilé is to do something about this problem.

QM: I would just add that on mainstream computer systems we use everyday, there are usually several ways to do a single thing... yet they often don't deliver a truly identical result, or they deliver different results in a rather unpredictable way.

NR: We want a powerful, yet easy to use Desktop Environment, based on small, cooperating applications (or components). We want to focus on collaboration features, documents/projects management, and obviously take a great care of any usability issues.

| G: Who are you? How many developers contribute regularly? How is the project leaded? What is the profile of the main developers?

YJC: My real job is biomedical research. I work on LuceneKit (a search and index library) and related parts occasionally. It is an Objective-C port of the Java framework Lucene (Apache Foundation). The Étoilé project is lead by Quentin Mathé and Nicolas Roard. I simply focus on my part.

JR: My name is Jesse Ross, and I am the artistic lead on the project. I build mockups, icons, and themes, and do all the branding and identity work for the project. I got involved in the project after having developed some icons and a theme for GNUstep.

DC: I am a PhD student at Swansea.I was introduced to GNUstep by Nicolas after I started using Cocoa, and to Étoilé after I started becoming frustrated with the limitations of the OS X User Interface.

QM: I'm Quentin Mathé, student in Applied Mathematics, before turning back to Science studies recently I studied Multimedia during two years. I initiated Étoilé project and I'm currently leading it with Nicolas Roard. Being a Cocoa developer in the past, I started to contribute to GNUstep two years and half ago because I was getting more and more disappointed and frustrated by Mac OS X user experience: I was expecting way more improvements over Mac OS 9 and older versions, but they never came, we were just getting a new side step with each release. From an User Experience standpoint, various minor gripes were well addressed, but the most important flaws were totally ignored. It is a rather strange thing because Apple started to investigate them in 90' with various projects like Newton, OpenDoc or Copland.

When I discovered GNUstep, unlike GNOME/GTK, KDE/QT or whatever, it looked like an exciting possibility to build a new Desktop Environment I would like to use. I thought initially Backbone could be a such environment, Nicolas was involved in Backbone, which brought us to exchange our ideas on User Interface and Desktop Environment topic, but I wasn't interested by official Backbone vision which intends to recreate NeXTSTEP before starting to rethink it. In parallel I became more critical on Cocoa API, in a perspective taking in account both its historical design and how it was evolving right now. Finally GNUstep being still not really usable two years ago, I decided to contribute to GNUstep, moreover it was a good opportunity to have a better understanding of Cocoa/OpenStep API. Ultimately I thought it would open various possibilities to improve it.

NR: I'm a PhD student at the University of Wales, Swansea. I'm working on distributed visualisation systems, and use a bit of Objective-C/GNUstep/Cocoa here and there ^_^... I had always heard incredible praise about NeXTSTEP/OPENSTEP (I think the first hints were some comments in STMag, and there was also the Phaleon Demo on Atari with a mock up NeXTSTEP workspace), so I was rather curious about the whole thing. GNUstep beeing an implementation of the OpenStep APIs thus looked like a very interesting project.

I actually remember looking into it a long time ago -- perhaps around 98 or 99 -- and at the time GNUstep was rather complex to install, and not really in a very usable state for end users or casual progammers, as it was based on the half-working Display PostScript implementation from Aladdin, etc. So I forgot a bit about it, and used C++ with QT, which is a very nice library. But then, something like 3 or 4 years ago I had another look into it, and this time actually installed it and started playing with it. I completely fell in love with Objective-C, which is a much more dynamic OOP language than C++, astonishingly simple and powerful (and I'm now completely on the side of dynamic languages and systems - Smalltalk, etc.). Aehm. Anyway, I wrote a couple of applications using GNUstep at that time, and was blown away by the incredibly well designed framework. I thus started beeing more involved with it, wrote a few articles about GNUstep with Fabien Vallon in Linux Mag France! , participated to the FOSDEM, etc.

At one point I started to think that having a GNUstep-based Desktop Environment would be something really nice, and contributed to the Backbone project. Alas, lack of contributions and too few persons involved lead to a stall. Around that period Quentin started to work on Étoilé, which was heading toward a different, more radical goal. I was interested by this idea, and became involved.

The main contributors to Étoilé for the moment are probably Quentin and Yen-Ju. We try to lead the project in a democratic way, using the wiki and the mailing lists. Overall, we have 3 or 4 regular committers.
G: When and why was the project started?
QM: The project was started in June 2004 but didn't really begin until December 2004 with Uli Kusterer participation to IconKit and February 2005 when Yen-Ju Chen joined us. During this month, we also decided to merge with Garma led by Banlu Kemiyatorn, Garma was a desktop environment project hosted on Gna! but focused on creative tools unlike Étoilé. Few months later Banlu Kemiyatorn retired for personal reasons, Garma development was interrupted, therefore our planned merge was delayed until we have time to handle it, but... Yes, it is still on our roadmap!

To talk about reasons to start the project, the first one would be that Apple needs a real competition on the User Experience front ;-)... Well, a bit at least... they are still setting the trends these days and often not perfectly in my opinion, even if Mac OS X is still way ahead of its competitors. When you look at Windows, GNOME or KDE, and even Mac OS X, they don't seem to really want to change the way computers may work, partially because of historical reasons and the need to support existing software, and probably also because the main model in the last fifteen year, for many people has been Windows, not even Mac OS, due to its market share. The end result is that very few try to innovate, and when there is innovation, it is usually baby steps. Notwithstanding they are interesting efforts like WinFS or CoreData+Spotlight. Even though it might seem a bit crazy, we would like to try a more radical departure.

The second one is GNUstep whose development is still slow specially on AppKit side, because it misses very active contributors and looks truly ugly to attract any new users or developers (although Camaelon answers most of these issues). I have contributed a bit to AppKit development previously but I decided it was more important to attract people by working on something like Étoilé. Ultimately that would lead me back to work on GNUstep for Étoilé needs.

The third reason lies in the fact that I have been studying graphic design few years ago and I was and I'm still really disappointed by software tools used everyday in this area. To take an example... Photoshop is powerful but unintuitive and not really flexible. I'm pretty sure there is room for many cool improvements with a component-based approach.

| G: Could you provide us with a short history of NeXTSTEP, OPENSTEP and GNUstep?

QM: I'm sure Nicolas or someone else might do it, I mean better than I would... :-)

NR: OK... Not sure I can be brief ^_^

Well... to sum it up, it's all Steve Jobs' fault :-) After he left Apple in 1984, he started a couple of companies: Pixar and NeXT. Everybody knows Pixar, but NeXT was in fact its main focus for a long time. NeXT's goal was to produce the best workstation ever done. They more or less succeeded, with the famous NeXT Cube in 1988, and later the NeXT slabs (more infos in french on ). For a variety of reasons though, they completely failed commercially (selling pricey workstations to students is... well... not the best business plan apparently, even if the stations are nice ^_^).

But in fact, more than the hardware, and particularly looking from now, it was the software that was truly innovative and interesting. NeXTSTEP was based on Unix (a BSD subsystem on top of a Mach kernel), with an incredibly well designed UI (not everybody had the chance of having a Keith Ohlfs on board) that also completely hide the Unix layers. In addition, there was innovative UI features - vertical menus, tear-off submenus, the Dock, the use of a Display PostScript (everything on screen was actually Postcript, so it was great for WYSIWYG and for printing), an excellent pasteboard system coupled with a way of making apps collaborating by providing system-wide services. And that was for the UI alone.

The real power came from the use of Objective-C, an Object-Oriented language that actually manage to add OOP to the C ANSI with just a dozen of additional keywords and one syntactic change, and follows Smalltalk's philosophy (rather than Simula, like C++). NeXT was able to leverage Objective-C to provide a completely object-oriented user interface. One of the crown jewel of the development environment was InterfaceBuilder, a program to create user interfaces by simple drag and drop, written by Jean-Marie Hullot by simple drag and drop. At the time it was a revolution. In fact, InterfaceBuilder is still far ahead of any GUI Builder, because, more than beeing a simple UI Builder, it's really an Objects Connector application - you can manipulate any kind of objects, not just UI objects. The whole programming environment meant that you reused existing objects a lot, and thus, the typical programming time was severely reduced.

Alas, NeXT was in financial difficulty, and in 1993 it stopped producing hardware and started to focus on the software only. They wrote WebObjects, an outstanding Web Application toolkit, and again had a interesting way of selling it (licenses were in the hundred of thousands of dollars, afaik).

More interestingly for us, in september 1994 they published jointly with Sun an API specification inspired by the NeXTSTEP API, called OpenStep, and released an implementation of it called OPENSTEP, an operating system running on various architectures - 68k, hp stations, sun stations and x86.

A bit before, NeXT was forced to release its GCC modifications integrating Objective-C to the FSF, and some people started to write an implementation of the NeXTSTEP APIs. Then after the publication of the OpenStep specification, the focus changed to implement those specifications, and the GNUstep project was launched by the Free Software Foundation to provide a cross-platform OpenStep implementation - in theory it should have been the foundation of the FSF Operating System. For various reasons (lack of developers, lack of interest in the project, particularly if incomplete, relatively few people had contacts with a NeXT station, etc.) it progressed very slowly, although now it's quite usable.

In the meantime, Sun decided to drop OPENSTEP in favor of Java (which actually is quite inspired by Objective-C ), a rather severe case of Not Invented Here syndrom. NeXT itself was basically focusing on WebObjects, and not much on OPENSTEP anymore. At this point things were looking a bit dull, until Apple, which was looking for a replacement OS for its aging Mac OS system, decided to buy NeXT in december 1996. Considering most of NeXT engineering heads (Tevanian, etc.) ended up in command at Apple and that Steve Jobs itself became CEO... you could very well argue that NeXT bought Apple and was even payed for it ;-)

OPENSTEP then had a facelift (Aqua), dropped Display PostScript for Quartz (a "Display PDF") and a compatibility layer at the API level for old Mac OS application was coded (Carbon) as well as an emulator to run Mac OS 'classic'. The strange result was named Mac OS X. It's very similar to OPENSTEP in a way, the Cocoa API is still an OpenStep implementation (all thoses classes names beginning by NS mean NeXT/Sun), and the general system architecture is roughly the same. But at the UI level, they needed to act mostly like the old Mac OS to not scare users, so instead of vertical menus, applications uses the top-horizontal one, etc. The sad thing with this is that it resulted in a not very good mixture, particularly the Finder, a mix between the OPENSTEP Workspace Manager and the old Mac OS Finder. Also, many other interesting aspects of OPENSTEP were left behind (the Services menu is not very accessible, remote display is not here, no system-wide filters, no possibility to lin! k live documents, no default dictionary service, etc). The general feeling is of a setback from the more advanced OPENSTEP, although it's arguable that Apple didn't have much choice if it wanted to keep its user base and its existing softwares (and even so the transition wasn't that easy).

Now, for GNUstep this turned out to be a pretty good thing, as all of a sudden a myriad of books, documentation about OpenStep (well Cocoa), Objective-C programmers and Objective-C programs started to appear, and a (small) fraction came to GNUstep. Perhaps partly because of that, and partly because GNUstep was starting to be usable as all the pieces started to be more or less here, more people are interested in GNUstep than ever. The recent 1.0 version of GORM, the GNUstep equivalent to InterfaceBuilder, is also a very good thing. In the near future GNUstep should be able to read directly IB files, and vice versa, and porting Mac OS X applications to GNUstep (and vice versa) should be even less work.

| G: And what audience are you targeting exactly? Geek, Grandma? Could you describe the profile of your community? Which users?

QM: In a wide sense, our target is every computer users. We intend to provide real simplicity, high scalability, but as David outlined I'm pretty sure in the short term Étoilé will interest developers and power users before grandma. Providing efficiency and flexibility initially is one thing but in an evident manner we will lack applications and a packaged solution (GNU/Linux-like distribution with Étoilé) that is probably crucial. We cannot avoid to be a bit rough on edges too. My secret hope or desire is to have a set of very high quality components for graphic design (page layout, image manipulation etc.), but right now we are really short on developers for such development.

JR: The audience of Étoilé are people with an interest in collaboration and organization - people much like ourselves. All the developers on this project are extremely interested in usability and in finding alternative ways of dealing with the mass of documents on our computers. Additionally, we all want to see certain technologies that are critical for global communication -- such as IM, VoIP, email, file-sharing, white-boarding -- integrated into the entire desktop experience.

DC: Most of us come from Mac or NeXT backgrounds. We are all, as far as I am aware, current users of one flavour of UNIX or another. Our target audience is anyone who uses a computer. Usability is a big focus for us -- I am a very lazy person, and so I believe that using a computer should require the absolute minimum of effort possible. If a task can be automated, it should be. Nothing should ever prompt me to confirm -- it should just go ahead and let me undo it if it turns out that wasn't what I meant to do.

OPENSTEP is well regarded as a development platform, and so Étoilé is very lucky to have GNUstep to build on. Our aim is to abandon some of the metaphors that are no longer relevant, and provide new ones for developers. My personal pet hate is the file. What is a file? What, in the real world, do I interact with that behaves like a file? From a developer's perspective, a file is an irritating abstraction - you have a collection of bytes, but you have no way of knowing how you are meant to interact with them. From a user's perspective, a file is a... thing. It doesn't map nicely to any concept the user is familiar with.

We replace the file with an object. Objects can be anything - images, text, video clips. Since we are using GNUstep and Objective-C, we get things like introspection for free. We can grab an object and say 'can you behave like a piece of text?' or 'I would like a PDF version of this, please.'

Having objects as first class citizens is only part of the puzzle. How do you arrange them? Our solution is to use projects. A project is something like a cross between a folder and a virtual desktop. Objects in a project can be arranged spatially and edited directly - you don't have an icon representing an object, you interact directly with a view of that object. Projects, like folders, can contain other projects.

The idea is to maintain the UNIX philosophy of 'a tool does one thing, and does it well.' Different components will be responsible for editing different types of object - you won't need to have a dozen text editors, or images editors, or drawing tools, for example.

We extend the clipboard to allow a link to another document. You might have one project containing your photo archive. If you want to create a newsletter, you would create a new project for it. To insert an image from your photo archive, you would select the image, copy it, go to the newsletter and say 'paste link.' It would then be easy to jump back to the original image, and any modifications you made to the original would be updated in the linked copy. At a fundamental level this is similar to the concept of a symbolic link, but exposed at a much more useful level to the user.

NR: As highlighted, our current "target" is indeed developers, because we don't have much "user" applications for the moment. We only provide programming tools and frameworks, plus a couple of applications or bundles like Camaelon. Also, we don't have any packages for the usual GNU/Linux distributions, only 'tgz' snapshots of the repository. In the short term, we'll probably stay that way. However, I expect normal users to be interested by Étoilé in the medium term (some before, like people wanting to use Camaelon), and our "final" goal is obviously to target your everyday user, by providing a complete environment.

| G:Which applications and environment (business, home, scientific, anyone) are you targeting ?

DC: I think the main targets are, as always in community-driven Free Software projects, the developers. For me, the main requirements are a good way of editing structured text, a decent development environment, and the standard set of Internet tools. For Jesse, I would imagine high quality image editing capabilities are high up the list.

QM: For the moment, business and scientific, no. But definitely yes for home, graphic design, photo, developer tools... By home I suppose: Mail, Web Browser, Text Editor, Minimal Page Layout and various Collection Management Tools (by collections I mean mails, music, photos, and other documents)

JR: For me, targeting creative applications and users is important: web developers, designers, photographers.
NR: Well, I'd guess general use. Also, as it is so easy to program a GNUstep (and by extension, Étoilé) application, I think it could be quite interesting from a scientific/research point of view. I know that I'm quite happy to develop very quickly applications on GNU/Linux with GNUstep, or on Mac OS X with Cocoa, and beeing able to easily port applications between the two.

G: Is your software already usable, in your opinion? What features are you missing/planning? Do you have a defined roadmap?
DC: Some parts are usable, but a lot is still in the planning stage. Quentin is probably in the best position to answer this one...

QM: At this date, Étoilé is not usable on user side because it is mostly made of frameworks except for few things like Camaelon (our GNUstep theme engine) or WildMenus. WildMenus allows to replace NeXT menu with a Mac-like top menu bar. If we focus on what might interest developers right now, we have some frameworks in a usable state and currently in testing phase:

- LuceneKit
- OgreKit
- PreferencesKit
- UnitKit

YJC: Since my part is porting from other projects, I simply follow their progress. That is pretty much all I can afford right now. Both the Apache Lucene and OgreKit are quite usable and mature. Therefore, my work is to make the Objective-C ports work and bug-free.

NR: Some "end user" applications are also planned in the short term, in particular we would like to finish the file manager based on Uli Kusterer's code, a good tabbed shelf, integrate the desktop/sessions code, etc. The plan is to have something partially usable by the end of February, for the Fosdem even if only as an alpha release.

| G: Do you plan/wish native "*step" ports for famous cross-platform applications (Mozilla suite,, GIMP, etc.)? Does it make sense to you?

YJC: To me, it is not practical. In the near future, better communication among applications is an acceptable solution. In the long run, the separation of GUI and underlying libraries will make porting easier. For example, if GIMP can be separated into the GUI and image processing libraries, we can build a new one with GNUstep without reinventing the libraries. In this way, users can take the advantages of GNUstep and developers don't rewrite everything from scratch.

DC: Not really. The concept of an application doesn't fit very well with Étoilé in many cases. In a few, such as a preference panel, you will have an application, but in others the interface will appear to allow you to interact directly with your data, without an application getting in the way.

The recent release of GCC finally added support for Objective-C++ to the GNU runtime. This allows C++ and Objective-C to be mixed in the same source file. While in general this is not something you would want to do very often, it is important since it is used heavily by Apple's WebKit to interface with the C++ KHTML. Hopefully a GNUstep port of WebKit should appear soon, which would give Étoilé HTML rendering capabilities backed by a very good rendering engine. is an interesting one. While a GNUstep port of OO.o wouldn't make much sense, the OASIS file formats used by OO.o 2.x are very much of interest to us - we intend to provide native support for them in Étoilé.

QM: At first glance, it wouldn't make sense to port such big monolithic applications because they have not been thought to match Étoilé philosophy.For Mozilla, we could talk about Camino, it wraps Mozilla core in a simple but elegant Web Browser built with Cocoa. It was created many years before Firefox to provide what Firefox has achieved only recently. It is 'just' a web browser without a bunch of features and with an important focus on usability. However porting Camino involves rewriting the drawing code done in Carbon. Furthermore Camino won't give us the excellent WebKit API, which is why we are probably going to (or somebody will) port WebKit to build a browser around it rather than porting Camino. It is worthy to mention that various excellent open source Cocoa applications like Adium would be interesting to port. Adium is an IM application built around gaim library.

| G: We see PDAs are targetted. Does Etoilé and GNUstep in general has some specially fitting features for PDAs? (Étoilé_pda_ui_01.png)

YJC: To me, it is just a concept. We are still too far away from that.
QM: I agree with Yen-Ju, but I will try to sketch an answer... For special features I would say no, but GNUstep gives us interesting possibilities; it is an extremly light environment when you compare it to GNOME or KDE. Objective-C is a very dynamic language, relatively close to NewtonScript which was one of the Newton secret weapon. Newton PDA are still unmatched today even when you take in account they have been discontinued more than five years ago. I like to say Newton is the best product Apple has ever built, even better than the first Mac or Mac OS X. Its user experience is very good, specially in term of consistency when you compare it to a Mac.

Finally an Objective-C runtime bridge exists for the Io language and is planned to be integrated to Étoilé. Io is one of the most impressive language I have ever used and it is in part inspired by NewtonScript.

DC: Two of us, myself and Nicolas, are on the Nokia 770 developer program. The 770 is a nice device for developers - it's Linux based and runs X11 making it very easy to port existing code to. The relatively small screen makes it an interesting challenge for UI design. I would like to see Étoilé scale from huge high definition displays right down to small form-factor devices. A user is not expected to learn a new UI when they move between a desktop and a laptop, and they shouldn't have to learn a completely new one when they move between desktop and PDA either.

NR: Well, of course for the moment it's a pure concept :-) I think it's worthwhile to keep that possibility somewhere in our mind.

I think that something using GNUstep/Étoilé would be a very interesting platform for a PDA, and I'm personally rather interested by that idea, as I think a good Desktop Environment should share many common traits with the Newton User Interface. GNUstep itself gives us with a nice foundation, and not relaying on right-click user interfaces, as well as existing hooks in the API for managing "virtual ink", should be good things. There is actually a fork of GNUstep that runs on the Zaurus, called mySTEP. Myself I'm more interested by something that would work on the Nokia 770, and a few persons in the GNUstep community got a Nokia 770, so... we'll see what happens !

| G: What is your relationship with MacOS X, which is supposed to be "next-based"?

YJC: Since Mac OS X supports Cocoa, which is mostly compatible with GNUstep, most programs are cross-platform.
QM: We are related to Mac OS X because we are using on GNUstep. More precisely our relationship is based on OpenStep, the shared ancestor of both Cocoa and GNUstep. In addition, GNUstep is trying to keep up with Cocoa evolution, which implies it is possible to port Cocoa application on GNUstep relatively easily.

However, Étoilé is a layer on top of GNUstep. In the next development phase, we will probably override and extend some parts of GNUstep when it is necessary to reach our aim. In spite of this fact, we plan to design and implement this layer as cleanly as possible, which means Cocoa/GNUstep applications would still work in a compatibility mode when they haven't been tuned for Étoilé.

DC: Mac OS X includes a number of development environments. Of the two main ones, Carbon is based on the old Mac OS 9 API ands Cocoa is based on the old NeXT APIs. GNUstep and Cocoa are now both super-sets of the NeXT Foundation and AppKit, making it relatively easy to port code between the two. The main benefit we get from this is that we can run parts of Étoilé on Mac OS X, allowing people to get a feel for our system without having to abandon their old platform. Once Étoilé is in a state where it does everything they need, then they can leave Mac OS X behind, and move to a platform that doesn't encourage vendor lock in.

| G: Which license did you choose and why?

QM: The GPL has gray areas when it's about collaborating programs or using them remotely. We think the LGPL or the BSD are much clearer and permissive on this point, and we tend to favor them as our architecture is based on components.

YJC: It depends on probjects and developers. I personally prefer BSD-like licence for more freedom. But since my work is porting, I have to adapt the licenses from the original works.

NR: We used LGPL and BSD licences, as our work is based on other work, and as for libraries LGPL is in our opinion a better choice than GPL.

DC: My contributions are 3-clause BSD licensed. I believe that Free Software is intrinsically more valuable than closed software, and doesn't need the protections of a viral license. I also share Theo De Raadt's aversion to any license that requires a lawyer to understand it - if the license isn't 100% clear, then it's aimed at lawyers, not developers.

| G: Do you have any industrial or institutional support? If any, how so?

QM: Nope.

YJC:As far as I know, there is no support from industry or institute. GNUstep hasn't draw enough attention yet.

DC: Two of us are research students at the University of Wales, Swansea. While this doesn't officially count as institutional support, it does give us access to some shiny toys to play with and a lot of bandwidth. We also received two heavily discounted Nokia 770 Internet Tablets from their open source developer program.

| G: Are you looking for contributions? If so, what kind of contributions could be of use to the project?

YJC: That's definitely true. More programmers are highly welcome. In order to have a good desktop environment, we need a good infrastructure. And I believe we are short of people on that.

QM: Yes. To be a bit more concrete than Yen-Ju, I can mention what kind of contributors we would really need now:
developers (writing frameworks like BookmarkKit, maintaining frameworks like IconKit or designing new frameworks or components)
web site mainteners (keeping the content and the layout up-to-date through the whole web site, keeping published documentation in sync with the repository)
packages mainteners (taking of care of packaging policy with GNUstep and keeping our whatever-system-packages up-to-date)
developers on GNUstep itself to work on AppKit development (that's very important too)

| G: What tools do you use when working on the project? Why?

DC: NeXT gave the world the first Rapid Application Development environment - the system that Visual Basic is a low budget, poor quality imitation of. GNUstep and Mac OS X provide similar tools to those found on the NeXT platform. For editing code, I use a mix of Vim and Apple's Xcode. For User Interface design there's Gorm, which recently hit version 1 on GNUstep and Interface Builder on Mac OS X. I have a PowerBook which is my main machine, and so I tend to work on Mac OS X, but I also have a ThinkPad running FreeBSD that is used as my testbed, and when my PowerBook is in for repair, which happens distressingly often.

JR: I use Photoshop, Flash and SubEthaEdit.

YJC: I still use vi most of the time. GNUstep provides an easy way for compilation.

QM: When it is possible I work on Mac OS X where I'm more productive, I use tools like Xcode, InterfaceBuilder, SubEthaEdit, Subversion, svnX. When I work on GNUstep specific development on other UNIX (under GNOME, KDE or WindowMaker) I use: Gorm, Kate, Make, GDB, Subversion. We are also using UnitKit ported from Cocoa for Unit Testing.

It is important to acknowlegde GNUstep is currently broken on Mac OS X 10.4, it wasn't the case with 10.3 where it was usable in Apple X11 environment.

I hope development tools state will continue to improve on GNUstep, Gorm 1.0 release is paving the way for such progresses, in turn making really possible to spend more time developing with GNUstep tools for GNUstep. There is an IDE for GNUstep called ProjectCenter but in my experience it doesn't give any interesting features over make, GDB and whatever-is-your-favorite-editor combo. Another one has recently emerged, it looks promising and is called ProjectManager . We will see what happens to both ProjectCenter and ProjectManager Because of this IDE issue, I tend to use Xcode when the project module is developed for both Cocoa and GNUstep/Étoilé. Xcode is quite decent and steadily improving but it is still missing refactoring support. In a better world, we would have an IDE on steroids similar to the one discussed on

NR: I mostly use vim on GNU/linux - the GNUstep makefile system is awesome, writing a makefile is a piece of cake with it. I'm also using ProjectCenter sometimes, and on Mac OS X I use Xcode. Of course I use a lot Gorm - you can create your own palettes of objects, that you can drag and drop and instanciate as well as the "default objects". It can really simplify a lot the development of custom applications.

| G: Why did you choose Gna! as host? What Gna! tools do you use? Which features do you like most? Which features miss you most?

QM: It was easy to register and registration were getting handled quickly. Finally Gna! web tools interface look decently clear and user friendly compared to SourceForge to take an example.

YJC: I think the SVN is mainly the reason.

| G: What is the question we didn't asked you would like to answer? (and the answer is...)

JR: Q: Where can we learn more about how Étoilé will work?
JR: A: The Precognative FAQ and Making Computers Suck Less

YJC: It is worthy to mention other attempts to have a GNUstep desktop environment like Backbone . To me, it is sad that many people want to have a GNUstep desktop environment but cannot work together because they have different visions. If we can put everything together just like GNOME years before, I believe we can draw more attention and people to work it.

QM: Q: Why did you choose Étoilé name because it sounds strangely French ?
QM: A: Yes, it is a French adjective that might be roughly translated to 'Full of stars'. It emphasizes various notions like connection, distribution, constellation that evocates how we envision Étoilé components model... However just six months ago, we discussed other name possibilities with easier pronunciation for English speakers, but we were unable to reach a solid agreement except for Étoilé itself (and actually the French speakers were not the one pushing for the status quo! ;-). That's why the name persists today.



No messages in Hotspot #8: the project Étoilé


Start a New Thread:

You could post if you were logged in
Show feedback again

Back to the top

Powered by Savane 3.1-cleanup