LWN.net Weekly Edition for April 6, 2006

Thunderbird looks forward

The Thunderbird mail client developers have recently posted a Thunderbird 2 page describing the changes they anticipate for the next major release. According to the roadmap, this release is expected in the "late Fall 2006" (presumably northern hemisphere) time frame. The task list is ambitious, but perhaps not sufficiently so.

One of the planned changes is to introduce multiple views of the folder pane - the list of mail accounts and folders which appears on the left of the window. Thunderbird users with vast numbers of folders would evidently like to be able to filter the display in various ways to make the list easier to work with. So there will be options to display "favorite" folders, the most recently used folders, or those with unread messages.

Current Thunderbird implements "labels" for messages; the user can mark a message as being "important," "work," "personal," "todo," or "later." There is no facility for adding new labels, so those which might be useful to your editor ("muchmuchlater") are not available. For 2.0, the developers have realized that (1) any self-respecting application must allow users to apply tags to objects, and (2) labels are really just a form of tags. So labels will be "rebranded" as tags, and users will be able to create their own tags. The association of colors with tags will be possible, preserving the color-coding capability that Thunderbird has now.

Another new feature is called "improved phishing support," which, one assumes, is not exactly what the developers intend to implement. Plans include integrating the Firefox2 safe browsing extension and making use of both local and network blacklists. There are also (unspecified) plans for improving the internal bayesian filter for spam filtering.

Then, there's the animated new mail alerts and a tooltip-like popup which can provide a summary of new messages in a folder without actually opening that folder. Your editor must confess to being unconvinced that inflicting even more little popup windows on the desktop will truly improve the overall experience.

There are a few other things which might be nice to have on this list. Your editor has been using Thunderbird with a (non-LWN) account for a while now, on the notion that there must be something to these graphical mail clients which makes them worth using. Based on this experience, he has a few suggestions for features he would like to see implemented ahead of animated alerts:

• The ability to configure the printing of messages - or, at a minimum, a realization that, most of the time, there is little value in using half a page of paper for every single header, causing even short messages to be split between two pages.

• Some flexibility in the on-screen header display would be nice as well. Why should it be necessary have all headers displayed just to see who a message was sent to?

• A provision for feeding a message to a shell command.

• Replace the confusing "Junk/Not junk" toggle with a non-modal interface.

• In your editor's experience, the internal bayesian filter is not as effective as it should be. Rather than try to improve it, why not fill out Thunderbird's fledgling support for integration with external filters? Being able to easily train SpamAssassin, say, from Thunderbird would be a great thing.

• Make it possible to send plain text (such as a patch) without having to go through strange rituals to keep it from being reformatted.

• Cause Thunderbird to not send HTML mail by default.

• Somewhere along the way, a bit of attention to reducing Thunderbird's memory footprint would not be entirely misplaced.

Thunderbird is a nice mail client in a number of ways, and its developers look like they plan to make it nicer yet. Your editor supports this work, but hopes that attention to some basic usability issues will not suffer as new features are added to this application. In many ways, graphical mail clients are still slower, more awkward, and less powerful than the text-oriented clients they ostensibly replace. Sooner or later, it would be nice to close that gap.

Comments (24 posted)

Fedora and MP3

The Linux world hears relatively little from Eric Raymond these days, a fact which maybe bothers some people more than others. Be that as it may, Eric recently broke his silence, in classic form, on the Fedora-devel list. It seems that Eric has come to save the Fedora distribution and set it back onto the path of Total World Domination.

There are a few small details that Eric would like to see fixed, including the FC5 artwork ("...backgrounds that look like a Teletubby hocked loogies into a dish full of soap scum."). But the real issue is in a different area: media support - DVD playback, Java applets, Flash media, and, especially:

The problem, of course, is that MP3 is a patented format. Since Fedora is, by design, a 100% free distribution, it is unable to include patent-encumbered software. So no MP3 format in Fedora. Adding MP3 support to an installed Fedora Core system is not a particularly difficult task for somebody who knows where to look (or how to ask a search engine), but it does require some extra steps. Red Hat's lawyers do not even allow Fedora (or its web sites) to even include a pointer to where this software can be found for fear of "contributory infringement" charges. As a result, adding MP3 support is too hard for many desktop users, especially home desktop users.

One option might be to get a distribution license for the GStreamer MP3 plugin. With such a license, Fedora could ship a fully licensed MP3 decoder, with BSD-licensed source. There remain issues with just how that plugin could be shipped with certain GPL-licensed players, but the real problem is elsewhere: a Fedora distribution with this plugin would no longer be redistributable by others. It would, in other words, no longer be a 100% free distribution.

Another option would be to put together some sort of third-party, repository with a carefully-chosen set of Fedora additions, a few of which just happen to include MP3 support. Said repository would naturally be hosted in a carefully-chosen country. Fedora could come with instructions for configuring the system to use that repository as a source of "useful extra software," with no mention of what is to be found there. Such a scheme might be vague enough to make the lawyers relax - though they have not made their feelings known on the matter.

Yet another approach would be for Eric to make his own, MP3-enabled Fedora offshoot distribution - call it Fully-Armed Fedora or some such. Eric, however, has declined that opportunity, saying:

What is really being called for here, in other words, is for Red Hat to stick its neck out and take the legal risk that comes with providing easy MP3 capability to Fedora users. Red Hat is understandably reluctant to do that. The company's relatively high profile and significant cash pile (around $800 million) make it a more likely lawsuit target than many others. Red Hat management probably sees much risk and little benefit in inviting lawsuits by including MP3 support, directly or indirectly.

Eric's claim is that companies like Red Hat need to make a business decision to solve the MP3 problem in one way or another, even if it means making deals with patent trolls or shipping proprietary software. A Linux desktop which cannot deal with MP3 files is simply too crippled to be taken seriously by a large portion of the potential user base. If Fedora is ever to succeed in that market, it must do what the target users want it to do.

There is a point here. Using Ogg for one's CD collection is no sacrifice, especially if one's portable player (running Rockbox, say) also supports that format. But there is an increasing amount of interesting content on the net which is only available in the MP3 encoding. All of that content is inaccessible using a stock Fedora Core system. That is, indeed, an unacceptable situation for many users.

Solutions must be approached carefully, however. Future systems are likely to present other problems: DRM-encoded video formats, broadcast flags, locked-down computers which only run officially-signed software, and more. Any solution which does not also offer at least some hope of addressing those issues will not get us very far. So, in other words, to properly solve the MP3 problem, we must (1) continue working to encourage the creation of content in free formats, and (2) face the legal issues which are at the root of these problems. Those goals will not be helped much by bolting proprietary or otherwise encumbered software onto our free systems.

Meanwhile, some other issues may be amenable to easier solutions. To that end, Warren Togami has announced the creation of a new mailing list for the discussion of artwork for future Fedora Core releases. Fedora Core 6 still won't play MP3 files, but maybe it can look a little nicer.

Comments (36 posted)

Sveasoft and the GPL

Sveasoft is a small company which makes its living by selling supported versions of Linux-based firmware for a number of wireless routers. Paying subscribers can download current versions of the firmware, which adds a number of features not normally found on those routers. They can grab updated versions as they become available, and participate in support forums as well.

Sveasoft's products are based on free software - Linux in particular. The company's approach to GPL compliance has raised eyebrows for a couple of years now. One tactic employed by the company has been to terminate support accounts for any subscriber who further redistributes the Sveasoft binaries or source. The GPL says that customers are entitled to that code (for the GPL-licensed portions of Sveasoft's products, at least), and that they have the right to pass it on to others. Sveasoft has responded that, when this redistribution happens, it is no longer obligated to provide future versions of the software. The company has employed various schemes for determining which subscriber has redistributed any particular version, and has been quite aggressive at shutting down accounts.

To some, it looks very much like Sveasoft is attempting to add restrictions to the GPL-licensed software it uses for its products. It is, in essence, imposing a penalty on anyone who redistributes its products. In the end, however, challenges to this model have not gotten far, and the Free Software Foundation has stated that Sveasoft is in compliance with the GPL - at least, with regard to its support agreements.

It seems that the story does not stop there, however. Sveasoft makes "pre-release" versions of its firmware available to subscribers. In practice, it seems that these "pre-release" releases are the actual product; the "public" releases tend to lag far behind. It also seems that the corresponding source is not made available to anyone - not even subscribers. Sveasoft argues that, since this is a limited, "pre-release" distribution, it is not obligated to provide source as well. The GPL, however, makes no exceptions for "pre-release" distribution.

The OpenWRT Project, on whose work Sveasoft's product is based, has had enough. So, in March, the project notified Sveasoft that its OpenWRT license was terminated due to GPL violations. From OpenWRT's point of view, Sveasoft no longer has any rights to be distributing OpenWRT's work in any form. Sveasoft responds that it remains in compliance with the GPL, and that OpenWRT has improperly incorporated Sveasoft code which was never meant to be licensed under the GPL - a charge that OpenWRT developers deny.

Since then, there has been a great deal of discussion, and Sveasoft's proprietor has come forward with an offer to create source tarballs on request for any subscriber who has received a copy of the binary firmware. There is also apparently an updated source tarball available to subscribers, though there has been no independent confirmation, yet, that it contains all of the source it should. The OpenWRT project has not, in any public way, rescinded its revocation of Sveasoft's license. Still, it would appear that public pressure has helped to move things in the right direction.

For now, at least. History suggests that Sveasoft will continue to push the boundaries of the GPL. Recent history also suggests, however, that Sveasoft may become less relevant in this area; by many accounts, the fully-free alternatives - beyond OpenWRT itself - go beyond the Sveasoft offerings in a number of ways. See this page on LinksysInfo.org for a detailed comparison of a few projects.

Comments (5 posted)