LWN.net Weekly Edition for September 20, 2007

Geographic display and input using Marble

At first glance, the KDE Marble project might look like a competitor to other 3D mapping applications, like Google Earth or NASA's World Wind, but it has a very different focus. It has a similar globe view and the navigation is familiar, but, unlike the others, it does not rely upon enormous data sets accessed via the internet; it is, instead, self-contained and fairly lightweight. The intent is to provide a framework for other applications to use so they can incorporate geographic information, while the Marble application is a demonstration and testbed for those ideas.

The project wants to see Marble used by many different applications, both for input of geographic information and for presenting it. The project lives under the KDE Education Project as one of the applications for Marble is for geographic learning. Many other applications could use a standard framework for displaying maps of various sorts, from games, using their own, possibly fictional maps, to GPS and other visualization tools.

Marble does not rely on OpenGL or any hardware support for 3D, in order to reduce complexity and dependencies, which will serve it well when porting it to embedded devices. The dataset that comes with the program weighs in around 9M and provides reasonable, worldwide, detail. The interface is meant to work like other geographic tools to provide a "geo-widget" that behaves the way users expect, removing one barrier to its acceptance.

The project recently released its 0.4 version, which was easily installed on Fedora 7 using yum. When starting it for the first time, it goes through a setup process, lasting for 30 seconds or so (depending on hardware, of course), but after that, startup is very quick. It opens with a spherical projection view of the earth along the prime meridian allowing users to grab and rotate the earth in various directions.

The navigation is simple, with zoom and pan buttons in addition to the "grab and pull" style. One can also pan the view by moving the mouse to the edges of the display and clicking once the pointer has changed to an arrow indicating a direction. Left-clicking on the map will give the coordinates of the location, whereas right-clicking brings up a menu allowing a few operations to be performed. While not horribly painful, moving around is a bit jerky, tracking noticeably slower than the mouse pointer moves.

The default theme is the atlas view, which looks much like the name implies, with colors and relief shading to represent elevation and ocean depth. Other themes available include a satellite view, using NASA Blue Marble data providing 500 meter per pixel resolution, as well as an "earth at night" view showing populated areas by the amount of light they give off. The information overlaid on the map contains political boundaries, cities keyed by population, lakes and rivers, notable mountains, and a latitude/longitude grid, each of which can be disabled as desired.

Many of the map features can be clicked to bring up information about the location, both from the program data and Wikipedia. The main right-click option is a distance tool, which measures the distance between the two (or more) points. Marble also handles standard GPS .gpx data files, along with support for Google Earth's KML format. Overall, this release provides a limited subset of the capabilities eventually envisioned for the tool.

The main thrust of the 0.5 release is to fully integrate the contributions from three Google Summer of Code (GSoC) students. Improving the KML support, adding gpsd support to talk to to GPS devices, and "flat" projections were completed by the GSoC participants. They have not been fully integrated into the interface for 0.4, but will be for the next release.

Longer term plans include adding support for data from OpenStreetMap, a Wikipedia-like project to map streets and roads worldwide. The project also plans to offer optional OpenGL support to enable hardware acceleration for applications and users who want it. Better resolution satellite data is another area that will be addressed by adding Landsat 15m data.

Marble shows a lot of promise, the current release is stable and useful, though it lacks many features. The key to its success, as a library and framework as opposed to an application in its own right, is in defining an API that is flexible enough for most applications. If the project can get that right, there are lots of ways to use it. Once an API stabilizes, we can expect to see Marble-enabled applications, hopefully soon.

Comments (8 posted)

Mozilla forming new company for Thunderbird

Mozilla has made its decision and will be spinning off Thunderbird into a new organization. Back in August, we covered a discussion about the future of the Thunderbird project, which was spawned by a series of postings in the blog of Mitchell Baker, CEO of Mozilla Corp. At that time, it was recognized that Thunderbird was suffering from a lack of attention, mostly because of an intense focus on Firefox. This week, Baker announced that the results of that discussion were to start a new for-profit company to nurture Thunderbird.

The new company, as yet unnamed and referred to as "MailCo", will start with three million dollars in seed money from Mozilla. The intent is to use that money to hire a small team to foster email and internet communications through Thunderbird. To that end, they have hired David Ascher, currently CTO at ActiveState, as the CEO of the new company.

Ascher also posted a blog entry about the new organization, providing some insights into the role of MailCo:

The biggest job for Ascher and MailCo will be to determine what, exactly, Thunderbird should be. From the comments on Baker's blog and elsewhere, it is clear that there is no consensus on what an email client should and should not do. There are many constituencies; trying to please them all is likely to please none.

There are lots of questions about integration of email with other internet services: instant messages, RSS feeds, VoIP, etc. There are also questions of local vs. remote message storage and web vs. host-based clients. Each has its advantages and disadvantages along with a vocal set of users. If MailCo starts moving in a particular direction, to the detriment of supporting others, they may lose some significant portion of their [PULL QUOTE: While profit may not be a requirement, some kind of potentially sustainable business model will probably have to be established. END QUOTE] user base. But a decision will have to be made in order to concentrate their efforts; it will be hard to find the right balance.

While profit may not be a requirement, some kind of potentially sustainable business model will probably have to be established. It is hard to imagine that Mozilla will keep pumping money into the company, though Baker makes it clear that they will consider further investment. Thunderbird does not have the obvious 'sell eyes to Google' model that Firefox has so successfully used; it directly competes with Google and other, similar, ad-supported mail sites.

For various reasons, Thunderbird has never had a large development community in the way that Firefox or other free software projects do. There is a core group of developers, presumably strong candidates to be hired on at MailCo, but in order for the project to succeed, it will need a bigger army of volunteers. There can be friction between paid developers and volunteers, especially if the volunteers feel like they aren't being heard. Growing and working with the development community will be an important part of MailCo's first year or two.

Many folks point to the stagnation of the main competition, Outlook, and liken it to the situation, several years back, with Firefox and Internet Explorer. There are some similarities, but there is also one big difference: Exchange. It is relatively easy for a user to change their desktop applications, even in a controlled workplace environment, but companies are unlikely to toss out their Exchange servers anytime soon. Because Microsoft completely controls the mail client to Exchange server protocol, Thunderbird will have a hard time being a drop-in replacement, in the way that Firefox is. One possibility would be to work with Openchange or similar Exchange replacement projects to provide an end-to-end solution for the enterprise.

Obviously there are some challenges ahead, for email clients in general and for Thunderbird in particular, but there is reason for optimism as well. Many did not expect Firefox to achieve the level of adoption that it has – it has made remarkable inroads against an entrenched competitor – and many of the same folks are behind the effort to give Thunderbird a push. Though it may seem like Mozilla is kicking Thunderbird out of the nest, they are actually giving it some resources so that it has a chance to fly. It certainly will not suffer under an organization devoted solely to its development.

Comments (10 posted)

The case of the unwelcome attribution

A couple of weeks ago, LWN examined the dispute with the OpenBSD project over the copyright notices placed in (and removed from) the versions of the Atheros wireless network driver intended for eventual merging into the mainline Linux kernel. At that time, the files with the improperly removed license text had never made it anywhere near the mainline repository and an effort was being made to fix the problem. It really seemed like the whole issue should end then.

So why does a perusal of the OpenBSD lists (and, often, unfortunately, linux-kernel as well) turn up gems like these?

-- Jason Dixon

-- J.C. Roberts

-- Can E. Acar

One might well think that the whole issue is still open. In fact, much of the dispute has gone by the wayside. The files with the improperly removed copyright notices never were going to make it to the mainline. The allegations by Theo de Raadt that taking a dual-license notice at its word was illegal have been pretty well laughed off; the OpenBSD camp is no longer asserting that claim. In fact, there is really only one point of dispute left:

• The OpenBSD developers do not believe that developers Nick Kossifidis and Jiri Slaby should have added their own copyright attributions to the file ath5k_hw.c. Those two developers, it is claimed, have not done enough work on that file to have earned any copyright claims there.

For this offense, the OpenBSD community continues to flame, threaten lawsuits, and more. It seems that the developers named above should simply add some original haiku to the opening comments so that their right to claim copyright to portions of the file would be indisputable. Even in the absence of bad poetry, these developers have done some small amount of work and will certainly do more to get the code ready for Linux inclusion. Threatening legal action as a way of keeping them from adding their own attribution to the file seems gratuitous.

Part of what is going on here may be a simple culture clash. It seems that, in the BSD world, the adding of a copyright attribution to a file is usually done with the permission of the existing copyright holders. For a developer to just patch an attribution can come across as being a bit rude. In the Linux community, instead, developers simply add a copyright if they feel they have done enough work to justify it. It is hard to come up with cases where these attributions have gone in without merit.

Eben Moglen's one public contribution to this conversation includes this paragraph:

That is clearly what is going on here - this discussion is certainly happening on a strongly emotional level. But it must be said that the most harsh language seems to be flowing in one direction: from OpenBSD toward Linux. This was also true when the situation was reversed and an OpenBSD developer was found to have improperly relicensed some Linux code. In both cases (and in others) there is a clear sense that the OpenBSD people feel wronged by Linux.

One might well wonder why this is the case. To an extent, OpenBSD developers may be following the tone set by that project's leader. They may be irritated by the licensing asymmetry: BSD-licensed code can be incorporated into a GPL-licensed project, but GPL-licensed code cannot be brought into a BSD-licensed project. Or perhaps they feel that their system has been unfairly upstaged by an inferior rival. Whatever the reason, there is a certain hostility emanating from that camp which is unpleasant to see.

It would be a mistake, however, to let the public flaming obscure the fact that Linux and the BSD variants have much in common. There is certainly no shortage of Linux proponents whose "advocacy" makes our community look bad. BSD will have people like that too. Meanwhile, behind the scenes, there is a great deal of good will, information, and code which flows in both directions. We are all working toward the same ends, and there are plenty of places where we can learn from the BSD communities. This incident will pass, and hot heads will cool - before, undoubtedly, heating up again on a different topic - but, through it all, free software will just continue to get better.

Comments (34 posted)

Page editor: Jonathan Corbet

Inside this week's LWN.net Weekly Edition

• Security: Exploiting symlinks and tmpfiles; New vulnerabilities in kvirc, openoffice, qt, quagga...
• Kernel: Large pages, large blocks, and large problems; A generic tracing interface; A summary of 2.6.23 internal API changes.
• Distributions: Fedora Audio Creation SIG; Fedora 8 Test 2; Gentoo Council 2007 Election Results; Ubuntu 7.10 beta approaching; Fedora Interview Series
• Development: Google Summer of Code 2007 Conclusion, OpenLDAP Update, Device driver for ATI Radeon R500/R600, KDE 3.5 on Nokia tablets, Introduction to Erlang, new versions of PostgreSQL, BusyBox, Clonezilla, MailStripper, OpenNMS, Vuurmuur, CUPS, ClamTk, FLAC, nova_filters, GNOME, GARNOME, Adaptive Planning,Chandler, XCircuit, pyFltk, Wine, Claws Mail, CK-ERP, Liferea, Firefox, Tcllib.
• Press: ISP DDoS attacks increasing, why Linux won't make it on the desktop, KDE on Windows meeting, Linux World China, SCO files Chapter 11, unlocked iPod, Dell customizes Ubuntu, Canonical courts Oracle, Microsoft loses EU antitrust case, Loop-based Music Composition, OneWebDay celebration.
• Announcements: McHardy elected as Netfilter head, FSF on EU antitrust case, ActiveState Komodo 4.2, IBM's Lotus Symphony, paid support for Freespire, Angolan chooses Mandriva, Mozilla to launch MailCo, Seiko label printers for Linux, printed PostgreSQL Reference Manuals, Greasemonkey scripts for LWN comments, Zero Install survey, FLOSS+Art cfp, NDSS 2008 cfp, IRC meeting on AGPLv3and GPLv3, Linux Plumbers 2008.

Next page: Security>>