First of all, I’d like to apologise for the lateness of this episode. As you may know, if you follow me on social media, I’ve started a new job as GTK core developer for the GNOME Foundation—yes, I’m actually working at my dream job, thank you very much. Of course this has changed some of the things around my daily schedule, and since I can only record the podcast when ambient noise around my house is not terrible, something had got to give. Again, I apologise, and hopefully it won’t happen again.
Over the course of the first chapter of the main narrative of the history of the GNOME project, we have been focusing on the desktop and core development platform produced by GNOME developers, but we did not really spend much time on the applications—except when they were part of the more “commercial” side of things, like Evolution and Nautilus.
Looking back at Miguel’s announcement, though, we can see “a complete set of user friendly applications” in the list of things that the GNOME project would be focusing on. What good a software development platform and environment are, if you can’t use them to create and run applications?
While GIMP and GNOME share a great many things, it’s hard to make the case for the image manipulation program to be part of GNOME; yes: it’s hosted on GNOME infrastructure, and yes: many developers contributed to both projects. Nevertheless, GIMP remains fairly independent, and while it consumes the GNOME platform, it tends to do so in its own way, and under its own direction.
There’s another issue to be considered, when it comes to “GNOME applications”, especially at the very beginning of the project: GNOME was not, and is not, a monolithic entity. There’s no such thing as “GNOME developers”, unless you mean “people writing code under the GNOME umbrella”. Anyone can come along, write an application, and call it “a GNOME application”, assuming you used a copyleft license, GTK for the user interface, and the few other GNOME platform libraries for integrating with things like settings. At the time, code hosting and issue trackers weren’t really a commodity like nowadays—even SourceForge, which is usually thought to have always been available, would become public in 1999, two years after GNOME started. GNOME providing CVS for hosting your code, infrastructure to upload and mirror release archives, and a bug tracker, was a large value proposition for application developers that were already philosophically and technologically aligned with the project. Additionally, if you wanted to write an application there was a strong chance that you had contributed, or you were at least willing to contribute, to the platform itself, given its relative infancy. As we’ve seen in episode 1.4, having commit access to the source code repository meant also having access to all the GNOME modules; the intent was clear: if you’re writing code good enough for your application that it ought to be shared across the platform, you should drive its inclusion in the platform.
As we’ve seen all the way back in episode 1.1, GNOME started off with a few “core” applications, typically utilities for the common use of a workstation desktop. In the 1.0 release, we had the GNOME user interface around Miguel de Icaza’s Midnight Commander file manager; the Electric Eyes image viewer, courtesy of Carsten Haitzler; a set of small utilities, in the “gnome-utils” grab bag; and three text editors: GXedit, gedit, and gnotepad+. I guess this decision to ship all of them was made in case a GNOME user ended up on a desert island, and once saved by a passing ship after 10 years, they could be able to say: “this is the text editor I use daily, this is the text editor I use in the holidays, and that’s the text editor I will never use”.
Alongside this veritable text editing bonanza, we could also find a small PIM suite, with GnomeCal, a calendar application, and GnomeCard, a contacts applications; and a spreadsheet, called Gnumeric.
The calendar application was written by Federico Mena in 1998, on a dare from Miguel, in about ten days, and it attempted to replicate the offerings of commercial Unix operating systems, like Solaris. The contacts application was written by Arturo Espinosa pretty much at the same time. GnomeCal and GnomeCard could read and export the standard vCal and vCard formats, respectively, and that allowed integration with existing software on other platforms, as an attempt to “lure away” users from those platforms and towards GNOME.
Gnumeric was the brain child of Miguel, and the first real attempt at pushing the software platform forward; the original GNOME canvas implementation, based on the Tk canvas, was modified not only to improve the performance, but also to allow writing custom canvas elements in order to have things like graphs and charts. The design of Gnumeric was mostly mutuated from Excel, but right from the start the idea was to ensure that the end result would surpass Excel and its limitations—which was a somewhat tall order for an application developed by volunteers; it clearly demonstrates the will to not just copy commercial products, but to improve on them, and deliver a better experience to users. Gnumeric, additionally, came with a plugin infrastructure that exposed the whole workbook, sheet, and cells to each plugin.
Both the PIM applications and the spreadsheet application integrated with the object model effort, and provided components to let other applications embed or manipulate their contents and data.
While Gnumeric is still active today, 20 years and three major versions later, both GnomeCal and GnomeCard were subsumed into what would become one of the centerpieces of Ximian: Evolution.
GNOME 1.2 remained pretty much similar to the 1.0, from an application perspective. Various text editors were moved out of the release, and went along at their own pace, with gedit being the main survivor; Electric Eyes fell into disrepair, and was replaced by the Eye of GNOME as an image viewer. The newly introduced ggv, a GTK-based GUI layer around ghostscript, was the postscript document viewer. Finally, for application developers, a tool called “Glade” was introduced as a companion to every programmer’s favourite text editor. Glade allowed creating a user interface using drag and drop from a palette of components; once you were done with it, it would generate the C code for you—alongside the needed Autotools gunk to build it, if it couldn’t find any. The generated code was limited to specific files, so as long as you didn’t have the unfortunate idea of hand editing your user interface, you could change it through Glade, generate the code, and then hook your own application logic into it.
Many projects of that era started off with generated code, and if you’re especially lucky, you will never have to deal with it, unless, of course, you’re trying to write something like the history of the GNOME project.
For GNOME 1.4 we only see Nautilus as the big change in the release, in terms of applications. Even with an effort to ensure that applications, as well as libraries, exposed components for other applications to reuse, most of the development effort was spent into laying down the groundwork of the desktop itself; applications came and went, but were leaf nodes in the graph of dependencies, and as such required less coordination in their development, and fewer formalities when it came to releasing them to the users.
It would be a long time before somebody actually sat down, and decided what kind of applications ought to be part of the GNOME release.
With this episode, we’ve now reached the end of the first chapter of the history of the GNOME project. The second chapter, as I said a few weeks ago, will be on January 17th, as for the next four weeks I’m going to be busy with the end of the year holidays here in London.
Once we’re back, we’re going to have a little bit of a retrospective on the first chapter of the history of GNOME, before plunging directly into the efforts to release GNOME 2.0, and what those entailed.
So, see you next year for Chapter 2 of the History of GNOME.
history of gnome gnome podcast