Friday, February 1, 2008


Story here, referencing this blog.

It was quite hard to find an actual claim in this article: there is no single piece of text that seems to be something we can evaluate in terms of facts. The nearest is this statement:

"It is, by this stage, becoming clearer that Jeff Waugh’s promise was not trustworthy. GNOME is apparently becoming Mono-dependent, which is a shame."

But the implication is pretty clear; GNOME is now, or soon, dependent on Mono. The source is a bit better, though, it says:

"[T]he GNOME guys want not only to replace libdbus with ndesk-dbus, but they want to nail down everything so that the new ndesk-dbus/Mono bindings are used in as much as possible!"

This is about, then, the acceptance of of ndesk-dbus into the GNOME external dependency set, which is an entirely managed DBUS implementation. The main point missed by the authors of the posts is that there already was a DBUS binding - it was buggy and unreliable, but this isn't about replacing libdbus itself, but replacing the C# binding to libdbus with a managed DBUS implementation. Fundamentally, we're talking about C# apps losing a non-managed dependency, not GNOME gaining a managed dependency. Specifically, it's not about replacing libdus.

More widely speaking though, the fact that it makes it into the external dependency list doesn't make it a GNOME dependency per se - the original decision on Mono/Gtk# hasn't changed, it's allowable as a desktop module. The addition of an external dependency is only of interest to those Gtk# apps in that module set anyway - those are still removable, and thus that external dependency is removable.

Friday, January 11, 2008

Knocking OLPC...

Article here, referencing stories such as this one.

As often happens when commercial deals happen within the free software world - as outside the free software world - the exact terms of the OLPC and Microsoft deal are somewhat sketchy. It's very easy to assume something is true, when in fact there isn't really any way of justifying that assumption.

There's one particular statement in the above article I wanted to pick out (the rest of it is mostly an opinion piece):

"Nicholas Negroponte has always been willing to go where the wind blows: the original OLPC prototypes ran Debian, notable because it's produced by a public-benefit non-profit. Once Red Hat offered money and resources, Debian disappeared from the system. Now it's Red Hat's turn to disappear."

An OLPC employee responded thusly:

"OLPC is not taking Microsoft’s money, and we are not being assisted in any way technically by the company. Bruce also claims that “original OLPC prototypes ran Debian […] OLPC XO prototypes never officially ran Debian."

(It should be obvious that both people can be correct - Bruce may have seen the hardware run Debian, and Ivan may be right that it was never officially supported: what is clear is that I can't find any evidence that the Debian project was officially involved in the OLPC project)

Is it right to impugn a project as being fiscally motivated, rather than motivated by the project goals, especially when the project is being run by a non-profit, when there is no evidence of money being involved? There is also no evidence that RedHat is somehow being pushed out; their status within the project has not changed (as far as the public is aware) recently.

Thursday, December 20, 2007

GNOME: a "cowboy project"?

Article here.

There are a couple of claims in the article which seem worth examining:

GNOME had not included support for the Open Document Format, which was accepted by the ISO as a standard in 2006, in either the word processor Abiword or the spreadsheet Gnumeric, which are part of GNOME.

Not only does Abiword support ODF, and has done for a while, but its deployment within the OLPC platform is ODF by default.

Gnumeric support is also there - it's difficult to say when it was "added" because it emerged from the StarOffice support, it seems. The OASIS "ODF Advantages" PDF puts support for Abiword and Gnumeric at September 2005.

Asked whether, in view of the principled stand taken by the K Desktop Environment on the OOXML issue, the GNU Project would officially acknowledge this (Stallman has acknowledged it in a posting to a GNOME mailing list) and consider nominating the K Desktop Environment as one that now met all the requirements for a free software project, Stallman replied: "That would be a very drastic thing to do."

I wonder if either the question was badly put, or the answer badly reported - RMS has accepted KDE as a free software project for a long time now, and it's listed in the FSF directory.

Tuesday, December 18, 2007

GPL and sublicensing

Story here, repeated almost verbatim here

A common criticism of the MS OOXML specification is that the patent promise is insufficient to allow an open source implementation without problems, and this story makes a specific claim about the patent license that we can examine:

"What Microsoft intends to do with its OSP is to forbid sublicensability, which is one of the cornerstone for distributing GPL code."

Legally, there are really two aspects to this: first, is sublicensing a corner stone for distributing GPL code (covered by copyrights), and second, does that apply to patent rights (rather than copyrights)? It's important to address both those aspects, since either could be true and would affect distribution in different ways.

First, the copyrights themselves. The GPLv2 doesn't address copyright sublicenses explicitly, but does say that "the recipient automatically receives a license from the original licensor" (section 6) which isn't a sublicense. GPLv3 is explicit, though: "Sublicensing is not allowed; section 10 makes it unnecessary." (section 2, last para).

Now addressing patents, the position GPLv2 takes on patent licenses is "a patent license [must] permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you" (my edit, the GPL states that in the inverse, section 7). It doesn't require the ability to sublicense, or request it.

GPLv3's section 11 (Patents) is considerably more complex than the v2 counterpart, but in essence requires contributors to grant a "non-exclusive, worldwide, royalty-free patent license". It doesn't ask for the ability to sublicense, and it's difficult to see why you'd need one given the initial license granted by contributors.

Monday, December 17, 2007

Accidental mob justice

Sometimes, the free software community decides to "mobilise" on an issue, and it pays dividends: a good example of this could be software patents in Europe, where everyone spoke up at the same time to ensure that they were heard.

Other times, it doesn't work so well. More often that not, problems occur because someone got the wrong end of the stick about something, blogged it, and the wider community decide to get active. Two good examples recently:

  1. A number of sites, including Slashdot, have run a story about a school student being given detention for running Firefox. It turns out that this is a hoax based around a doctored detention letter, and the school had to respond after meny people started contacting them.

  2. The WHAT-WG mailing list was deluged earlier this month after everyone and their dog blogged about Ogg being removed from the HTML 5 spec. Many people didn't realise that decision was merely one step on the path rather than the final position, and consequently many angry contributions were received by WHAT-WG, rather than more considered arguments in favour of standardizing on Ogg.

It's very easy to publicize "scandalous" issues, and get people to respond. It's a lot harder to undo. Sometimes, all it takes is a simple e-mail to check a fact out before releasing the hounds.

Tuesday, December 4, 2007

False attribution

Story here, referencing this.

When you write a story claiming that somebody said something, you'd better make sure they actually said it. In this case the headline is "Novell Vice President Again Defends Microsoft’s OOXML", and the story says "As far as attendant of the XML 2007 conference are concerned, Novell, represented by de Icaza in this case, opines that OOXML is needed."

Let's see what statements are attributable to de Icaza in the source material:
  • without a direct quote, he "said at the outset of his remarks on OOXML and ODF interop that he was not going to get into the corporate politics surrounding the two formats. De Icaza noted that Novell supports both OOXML and ODF via its use of Open Office."
  • “In 2006, there was lots of FUD (fear, uncertainty and doubt) about the problems behind OOXML and it went downhill from there,” de Icaza said.
  • De Icaza said Novell’s major issue around the dueling file formats was “there is no one-to-one mapping” between them. There are features in ODF, like Page Styles, which have no equivalent in OOXML.
  • “Neither group (Microsoft nor the ODF camp) is willing to make the big changes required for real compatibility,” de Icaza said.
  • “We need Microsoft’s (ODF interoperability) commitment to go beyond 1.0,” he said.
  • Objection over the size of Microsoft’s and the ODF’s file-format specs is a red herring, de Icaza said. What’s really needed is more developers who are building applications using OOXML and ODF to offer critiques of what does/doesn’t work, in terms of interoperability.
  • There’s no end in sight to the ongoing disputes between the two file-format camps, de Icaza said. “Sadly, there is a lot of money at stake here,” de Icaza concluded.
Which of these statements is defence of OOXML? And why is no attention paid to the statements he makes against OOXML, that it is missing features found in ODF, and that MS have no long-standing commitment to ODF interoperability.

It's always worth following the links behind stories to see what bloggers base their opinions on, you may find you come to a different conclusion sometimes.

Sunday, December 2, 2007

OpenDocument Foundation saga

More than enough words have been expended on the sorry saga of the Foundation, the story of which I will not repeat here.

However, this bit of investigation by David Berlind at ZDNet is worth a read. The public words and some of the private ones are sewn up in rough chronological order, and though the author draws an implicit conclusion (at least, to this reader) he rightly points out the different scenarios which explain the facts. Written relatively neutrally, it seems to be a decent sum-up.