Other LWN stuff:
 Daily Updates
 Calendar
 Linux Stocks Page
 Book reviews
 Penguin Gallery

 Archives/search
 Use LWN headlines
 Advertise here
 Contact us

Recent features:
- RMS Interview
- 2001 Timeline
- O'Reilly Open Source Conference
- OLS 2001
- Gaël Duval
- Kernel Summit
- Singapore Linux Conference
- djbdns

Here is the permanent site for this page.

See also: last week's LWN.

Leading items and editorials

gnucash 1.6 and the dependency nightmare. The release of gnucash 1.6 was announced on June 11. gnucash is an important application - it is the only free package which provides comprehensive personal and business finance functionality. Your editor has been using it for over a year, and has been anxiously waiting for it to catch up to what the commercial finance packages can do.

From the release notes, the 1.6 release has gotten much closer to that goal; unfortunately, LWN is not, as yet, able to say more than that. You see, we have not yet been able to make it work on any of our systems.

gnucash is perhaps the prime example of shared library dependency hell. The executable requires no less than 60 different shared libraries, all, of course, with the right version. Upgrading to GNOME 1.4 addresses many of those dependencies, but not all of them. Dealing with the rest has proved tricky, even for people who are accustomed to this sort of problem.

There is no criticism of gnucash intended here. The gnucash developers are trying to produce the best package they can by taking full advantage of the work that has been done by others. That is how component-oriented software development is supposed to work, after all. And gnucash is certainly not the only application that presents this sort of dependency issue. But there is an important point that is worth raising here.

A program that needs 60 different libraries is depending on a very complicated software environment to support it. As of this writing, there is probably not a single distribution which, out of the box, provides that environment. Upgrading to that environment is helped by the various update services and tools that an increasing number of distributions are providing. It is worth asking, however, just how many of you would proceed with such an upgrade in confidence that it would work, and that nothing else would break?

As the Linux software environment becomes more complex and powerful, it also risks becoming more brittle. The desktop will not be won as long as users must upgrade dozens of libraries, with a good possibility of breaking their systems, to get a new personal finance application. The desktop developers have a serious challenge ahead of them here: make the environment robust and easy to upgrade, or see the users wander away in frustration.

(As an addendum, it's worth noting that the gnucash developers have plans to offer a CD with the application and all required libraries shortly).

Linus is not accountable? We got a pointer this week to a white paper published by Microsoft entitled "Linux in Retail & Hospitality: What Every Retailer Should Know." It is available from the Microsoft web site, but only in Word format. It contains a pretty serious copyright notice that prevents us from putting up a decrypted version, unfortunately.

It is a worthwhile read. While the company's executives make people laugh by calling Linux "a cancer," this document dedicates a dozen pages to flat out Fear, Uncertainty, and Doubt. It even cites LWN editor Liz Coolbaugh as an expert on the number of distributions available, which, of course, is presented as a problem:

Imagine how confusing it would be if Microsoft released 188 versions of Windows and multiple versions of the GUI, each with a slightly different functionality? Wouldn't that be confusing? Wouldn't it be extremely difficult to run an enterprise solution with confidence about your future and return on investment in Microsoft products? That is the exact scenario that Linux is presently in by having so many distributions.

You read it here: choice is bad.

Rather than get into a point-by-point rebuttal, however, we would like to focus on one issue in particular:

With everything being in Torvalds' hands, he is in total control over where the future of Linux goes. If he doesn't want a new retail feature to be included in the core operating system, it will not be included. Additionally, he doesn't have any accountability to the industry when the releases are delayed, if they do not work well, etc.

One could have a lot of fun examining the degree of "accountability" shown by Microsoft when its releases are delayed, when they did not work well, etc. But that misses the point. The important thing to point out here is that Linus has, simultaneously, less control and more accountability than Microsoft would like its readers to believe.

On the issue of control, it suffices to say that Linus' domain does not extend beyond the kernel. Most of what users see as "Linux" has very little to do with Linus; it is, instead, the responsibility of thousands of developers worldwide. In particular, almost anything seen as a "retail feature" is unlikely to involve him.

Linus has a level of accountability that far surpasses anything Microsoft can claim: he can only lead where users and developers will follow. There is no structure that requires anybody to run or develop for his standard kernels; if he mismanages development, he may well find himself in charge of an obscure fork while the real activity goes elsewhere.

For an example of how this can work, see the the coverage of the device number debate in the May 17 LWN Kernel Page. When Linus made an unpopular decision, Alan Cox refused to follow him. As a result, many prominent distributions will probably include kernels that implement a policy different from that decreed by Linus.

In fact, most of the major distributors employ high-profile kernel hackers, and almost all of them distribute kernels which have been modified in some way. They have, in other words, declined to follow Linus in situations where they feel that their users' needs call for something different. Thus, for example, SuSE users have had ReiserFS for some time, and Red Hat users had the current RAID implementation, even though the standard 2.2 kernel did not.

This is one of the great powers that free software gives to its users: nobody can prevent them from incorporating whatever functionality or changes suit their needs. And it is the core of Linus' accountability. If he tries to take the kernel in the wrong direction, his user community will simply go around him. Proprietary software vendors generally lack that accountability, and their users suffer for it.

The survey results are in. Thanks again to everybody who took the time to fill in the form. Here, for those who are interested, is a set of highlights from the results. There was much there that was interesting to us.

• We got just over 3100 replies.

• Most of you seem to like us - 94% said that they have the site bookmarked and come here regularly. The bulk of you come by once a week, but about 1/3 of the respondents said they hit the Daily Updates Page every day.

• People are generally happy with the content, with the Front, Kernel, and Security pages topping the list. The pages with the highest "dissatisfied" votes were On The Desktop and Commerce, at 11% and 12%, respectively.

• On the list of possible enhancements, the security incidents and updates database and detailed research reports came out on top. Two often-requested features - email delivery and a PDA version - were requested by less than half of the respondents; those who want those features, though, want them quite a bit. Comment posting came in at the bottom.

• You are a very male group - 98.7%. There are almost as many women writing LWN as reading it. Most of you are between 25 and 35 years old. Almost all of our readers have at least some college education, with 38% having completed a graduate degree.

• 41% of our readers live in the U.S, and 5.5% in Canada. Most of the remaining readers are European, with the U.K. and Germany topping the list. Almost every other country on the list (and a few that weren't - sorry) was represented, though; we have readers from Argentina to Zimbabwe.

• Most of our readers are employed in technical positions, and most of them at smaller companies.

• More of you run Intel systems than any other, which is not too surprising. But almost 30% of you have Alpha-based systems and 20% use PowerPC systems as well. 85% of you have Red Hat Linux around; all of the other distributions fall into the single digits, with one surprising exception: half of you claim to run Slackware on your networks.

• 667 of you took the time to write in additional comments, and we've read every single one of them. Took a while. There were a few common themes that we found; even after we got past the most common type of comment, which was complaints about the survey itself. Yes, the survey could have been done a lot better; we apologize (again) for the problems.

  Beyond that, numerous people asked us not to change too much. Quite a few expressed hostility to the idea of comment posting, which came as a bit of a surprise; we had been considering adding such a feature. Gripes about the "On The Desktop" page were fairly common. We also saw a number of requests for expanded development coverage and a better organization for the software announcements pages.

  Some of you took us to task for excessive coverage of Red Hat on the distributions page. Others complained that we ignore Red Hat in favor of other, more obscure distributions.

  We were also surprised by a number of complaints about excessive commercial coverage. We operate in the belief that you can not really cover Linux and free software while ignoring the commercial sector; corporations, at this point, have a lot of influence over what happens. Numerous readers disagree, however.

We're still digesting the results of the survey; there is a lot of information there. We would like to thank you all, one more time, for giving us a bit of your time. It will help us to create a better LWN for everybody.

New LWN.net events calendar. The LWN.net Linux Events Calendar has seen a much-needed, much-delayed major upgrade. The new, Zope-based calendar provides a more flexible interface, and the ability to filter events by type. And finally we've done something with that linuxcalendar.com domain name... Have a look, we hope you like it.

Inside this week's Linux Weekly News:

• Security: Non-executable stack and heap implementations, new vulnerabilities in LPRng, xfs, gdm, xinetd, exim, fcron and TIAtunnel.
• Kernel: 2.4.6pre3; the limits of what modules can do; kernel data formats.
• Distributions: 217 Distributions and counting ...
• On the Desktop: Usability testing, GNUStep interview, and PalmOS updates.
• Development: GStreamer, Cplant cluster software, Ganymede 1.0, GVD 1.2.0, Open Source Legal Issues, APL for Linux.
• Commerce: TV Linux Alliance; 'LPI Certification in a Nutshell' from O'Reilly.
• History: Debian 1.1 was released 5 years ago; The "Open Source" trademark officially died 2 years ago.
• Letters: Licensing and security; naive economics; Linux's astrological sign.

This Week's LWN was brought to you by:

• Jonathan Corbet, Executive Editor
• Elizabeth O. Coolbaugh, Managing Editor
• Michael J. Hammel, Senior Editor

Next: Security