[Fri Jul 30 17:54:21 CDT 2004]

I deeply dislike political and religious dogmatism. It really riles me up to see people wasting their life attempting to impose some idea on everybody else and make reality follow their own utopian plans of how the world should be. How is this related to technology, you ask? Well, it is. Unfortunately, we see this type of pie in the sky extremism all too often in the ranks of the open source movement. Take, for instance, an article by Joseph Mallett about OpenBSD I just read. Only someone who lives in his own world can write something like this:

More and more often, Linux is being touted as a desktop operating system for the corporate environment. Its being pitched as a resource that saves money, gets a company out from under Microsoft's upgrade cycle and also as a means to get middle management and department heads to use the same secure operating system that runs their servers.

Unfortunately, the BSDs are often overlooked for this role. Linux software can run on BSD so software availability is not a factor and yet BSD is often times regarded as more secure than Linux, so perhaps it should get more attention than what it currently does.

The author goes on to praise OpenBSD for its unchallenged security records and overall stability, after also mentioning FreeBSD. I do have an enormous respect for both operating systems, but only someone who has no idea whatsoever what a non-technical user is would have the guts to recommend either of them as a desktop choice for "middle management and department heads". Come on! Let us be serious! My manager has problems adding email addresses to his Outlook address book, configuring a plugin in Windows or understanding how an URL is built. How do you want him to sit down in front of an OpenBSD box? {link to this story}

[Fri Jul 30 14:40:24 CDT 2004]

We are so used to reading articles about Linux and whether or not it is truly ready for the desktop, that some times we fail to realize we are simply using Windows as the desktop standard when trying to figure out if other alternative operating systems are up to the job, even though it is far from clear that should always be the case. It certainly makes more sense to agree on some more objective or neutral definition of what a good desktop is, and then see how different offers stack up against it. So, when one reads an article such as Sean Parsons' opinion on why Microsoft Windows is not ready for the destop we get caught by surprise. Let us be clear. To some extent, Parsons is simply playing a joke and stretching things a bit just to show us how we fall for certain assumptions that are far from proven. Still, he does make some good points. For example, it is widely assumed that Windows is the best choice when it comes to word processing, but Parsons points out how only WordPad is about the best it has to offer (Microsoft Office does not come by default with the operating system, and one has to pey its hefty price to install it in the system which is, needless to say, not very user-friendly). But there is plenty more: Internet Explorer has security holes and lacks features when compared to many other browsers, it is not possible to create a presentation with the default software bundled with the operating system, there is no instant messaging application capable of connecting to multiple IM networks by default, the email client is riddled with problems, file compatibility even between different versions of MS Office is a nightmare... As I said, Parsons is definitely stretching things a little bit too far, but he still has a point that we should think about. {link to this story}

[Wed Jul 28 12:42:04 CDT 2004]

While reading an opinion piece published by OSNews on how to improve the Linux desktop I came across a couple of projects I had never heard of before:

• The Y Window System, which aims at building an alternative to good old X11.
• The Zero Install System , which allows to run applications remotely without a need to install them locally, therefore avoiding dependency pitfalls, among other things.

As it tends to happen in these cases, there is a pretty good chance these projects will make it nowhere. Still, they do sound quite interesting and full of promise. {link to this story}

[Wed Jul 28 12:29:27 CDT 2004]

Robert Kaye publishes an article on O'Reilly Developer Blogs about the long delayed Perl 6. It was not so long ago that many in the open source community laughed out loud about the slow pace and permanent delays that always characterizes Microsoft operating systems. Well, Perl 6 has been in the making for several years now, and according to its main developers it will still be a few years before it is finally released. Yes, one of the strengths of open source is supposed to be a certain freedom from marketing plans and the idea that one releases the product whenever it is truly ready to be used. Still, as Kaye points out:

I've always preferred many smaller releases rather then infrequent large releases (release early, release often, right?) and large releases set of warning bells in my mind. Is the technology becoming too heavy? Longhorn certainly is, but is perl6? Will perl6 be worth the wait?

Even worse, while we wait for Perl 6, Python is gaining more and more support, which is something that should definitely worry Perl fanatics. {link to this story}

[Tue Jul 27 10:56:20 CDT 2004]

A judge has dismissed SCO's case against DaimlerChrysler in what amounts to SCO's first major defeat in its legal war against Linux. Not that this comes as a big surprise though. The notion that a company could press charges against another vendor claiming that Linux is in breach of their copyrights and, at the same time, also bring anothe company to court for using Linux when the first case has not even heard is just too preposterous to understand. Yet, that is precisely what SCO was trying to pull out in this case. Again, it should not surprise anyone that the judge dismissed it although there is lots to be written and said about the topic. By the way, feel free to check their website if you are interested in learning about SCO's side of the whole story. {link to this story}

[Fri Jul 23 08:47:22 CDT 2004]

Let us be honest and acknowledge that in our enthusiasm for everything open source often we do not give commercial OSes the credit they deserve. Iain Roberts writes an article for NewsForge that tries to tell us the good in AIX when compared to Linux. Most of his points can also be extended to other commercial Unices, such as HP-UX, Solaris or IRIX. Yes, the action is in the open source world, and most of the innovation is happening there, at least since 1998 or so. However, there are certain important enterprise features still missing from Linux, as Iain points out:

• Scalable device management: one needs to crawl through the output of dmesg to figure out the devices present in the system, while other OSes make it very easy to find out this type of information.
• Systems management: yes, different companies out there have been working in this field and are making an effort to develop products, but they come separate from the OS (which means that one also needs to pay for an additional product) and are not nearly as comprehensive, scalable and reliable as their counterparts among commercial Unices. And yes, this does make a difference when one needs to administer tens or hundreds of systems.
• Installations and upgrades: it is still a real pain to upgrade any Linux distribution out there. Sure, it may work for the most part. However, it is just as likely that something will go wrong, it will be plain impossible (as in the case of Red Hat's and SuSE's enterprise offerings, where one needs to reinstall from scratch in order to upgrade between major releases) or, even worse, it will work but this or that application will behave in a significantly different way than it did before. The Linux community has always been too prone to justify this as a byproduct of a very dynamic development process, but it just does not cut it for enterprise deployment.
• Managing disks and filesystems: we already mentioned above that Linux's device management definitely lags behind that of its older UNIX cousins, and the same can be said of this other related category. There is LVM, yes, but it still is years behind what other companies in the UNIX market offer by default.
• Workload management: things like logical partitioning have not traditionally been used in the Linux world, but when one is running a large system in the enterprise it certainle comes in handy to be able to partition it for different uses, departments, etc. Companies such as IBM and SGI do offer proprietary products that run on top of Linux but they are not part of the default operating system yet.

Mind you, there is a lot to like in Linux. I am not intending this to be read as a piece of Linux-bashing propaganda. I am just trying to help point out several fields where Linux still needs more work if we ever want to expand its use in the high-end enterprise world and replace UNIX with it. {link to this story}

[Sat Jul 17 18:56:13 CDT 2004]

eWeek publishes an article about the ongoing battle between disk and tape for storage supremacy. The likes of StorageTek are scrambling for new ideas and continue putting up a nice fight, but to be honest I am not sure tape devices have much of a future. The cost of disks and bandwidth has come down so much that it just does not make much sense to use such a slow and error-prone backup method anymore. It is not only SANs (Storage Area Networks) or distributed filesystems that I have in mind as tape killers. It has reached the point where it just makes more sense for anybody to back up to disk even at home. Simply add one more disk to your workstation and save files there, or set up an old system running Linux or FreeBSD and use it as a backup server. It is easier, cheaper and less messy than bothering with tapes. {link to this story}

[Tue Jul 13 16:03:10 CDT 2004]

I just came across an excellent article on the HP-UX kernel published by InformIT that is well worth a read. Although it is an excerpt from a book about the HP-UX kernel, most of what is explained in this article applies to any UNIX out there: overall design and philosophy of UNIX, I/O, virtual memory, search algorithms, etc. Actually, the authors provide one of the best explanation I have seen of several major search algorithms (static lists, hash tables, binary searches, B-Trees...) that I have ever seen, including several great and easy to understand examples. If you are interested in OS internals you should check it out. {link to this story}

[Tue Jul 13 15:51:15 CDT 2004]

It may be unfair, but Java has been living for years now with the general assumption that its performance is less than stellar. Well, J.P. Lewis and Ulrich Neumann, from the University of Southern California, have released a paper that finds Java performance on numerical code comparable to that of C++. It seems to me that this is one of those cases where the first experience counts. Back when Java was released (the boom years of 1995-1996), its performance was certainly bad compared to native code written in C or C++. Most people decided that they did not like it on the client, that it was sluggish, that it used way too many resources, and dumped it. Ever since, Java has been used almost exclusively on the server side. However, the fact that one can find Java just as easily on embedded devices these days should already tell us something. The reality is that, in spite of its bad reputation, Java has improved quite a bit over the years and it runs just fine on any decent desktop machine. I run a couple of Java applications on my workstation daily, and never noticed a major problem with them. Yes, it takes a while for them to launch, but other than that I simply cannot tell the difference performance-wise. We live, after all, in an era where many people run applications based on interpreted code and do not even notice. Perhaps we should give Java on the client another chance. {link to this story}

[Mon Jul 12 16:08:04 CDT 2004]

I have already written somewhere else about my skepticism regardind the Sun's Project Looking Glass to build a 3-D desktop interface. Actually, as a matter of fact, I am quite skeptical of any attempt to apply the 3-D metaphor to our traditional 2-D screens, be it from Sun or anybody else. I certainly do not deny the possibility that at sometime in the future (perhaps when we leave our present desktops behind, and adopt some amazing new gadget) the three-dimensional interface could make a lot of sense and may even feel quite natural. However, I strongly believe we have not reached that stage yet. So, in the meantime, these 3-D projects will remain purely experimental, on the edge type of development that bears a lot of promise but not fruit for the time being. In any case, Jono Bacon recently wrote a piece for O'Reilly Networks on the 3-D desktop that, while still not managing to change my mind on the topic, it certainly does lay out a few instances where the 3-D approach may prove helpful. {link to this story}

[Mon Jul 12 12:43:35 CDT 2004]

Distrowatch publishes a review of OpenBSD 3.5 that is well worth a read. It is nothing extraordinary, but it definitely gives you a nice idea of what this flavor of BSD UNIX is like. I must start by acknowledging that I never ran it or even installed it on any of my systems, although I do run FreeBSD. I simply have no need of a firewall in my home network, since the Cisco 675 DSL modem/router already does what I need it to. Nevertheless, if you need a firewall in your business or house, take a look at OpenBSD and give it a try. There is a good chance it will be your best choice of a secure operating system for the job, especially if you want to be able to sleep at night. Mind you, it is not a child's play. The installation process is complex, and the fact that the box is locked down will make it more difficult to administer too. Yet, that is precisely what a secure system is supposed to be like, right? {link to this story}