Predicting Three Comebacks

Well, it's a common belief that comebacks in technology don't happen often, but this time I predict three comebacks. I cannot recommend that you base your decisions (much less investment plans) based on it, but here they are. <h3>Comeback #1: Non-x86 Machines</h3>

Once upon a time, companies spent a fortune on costy UNIX servers or even IBM mainframes, just to be able to run their operation. Since then, x86-based computers have become faster and more capable, enough to make people be able to run Microsoft Windows NT or Unix-based Operating systems on them. This is now the most popular setup on the Internet for servers (much less clients and workstations).

However, these x86 machines are not as integrated as their non-x86 equivalents (UltraSPARC, PowerPC, etc.), are still much slower (due to the limitations of the x86 architecture vs. the superior RISC family of architectures). This costs in a lot of time of maintenance, many hardware failures, and sub-optimal lifetime. It is well-known that there are some PC XT machines which are still running. However, finding a low-end desktop Pentium I machine that's still OK is an exercise in futility. And even the high-end x86 servers suffer from many problems.

Human time is much more costy than shelling some money for getting a good RISC machine. And with the portability and stability of the GNU system, and portable and free operating systems such as Linux, or the BSDs, you can easily run a server there at ease.

So what can Sun and IBM (and friends) do about it?. They should phase out their own OSes in favour of Linux, as people have grown to dislike Solaris (to say nothing of AIX, HP-UX, or greater brain-damages). Don't get me wrong - Solaris has its place and I'm sure is a fine piece of engineering. But GNU/Linux is better for trying to promote the hardware.

Sun should also make the OpenSolaris licence GPLv2 and GPLv3 compatible, to allow it to borrow code from Linux and other GPLed-products (and vice versa). And they should also start supporting Perl, PHP, CPython, CRuby, Mono etc. etc. instead of just Java or whatever has been hyped. (Similar to what Microsoft have been doing recently for its MS-Windows OSes).

All of this will allow selling more hardware, getting more support and ergo - more profits. I think the transition to non-x86 machines will happen anyhow, but that will make it faster, and much more profitable for them.

Note that it is not the end of either Intel or much less the end of Microsoft. I still see workstations dominated by x86 in the near future, and both MS and Intel are insanely profitable and have enough time to re-invent themselves. But I think (and hope) we'll be seeing more and more non-x86 servers and to a lesser extent workstations. <h3>Comeback #2 : Mandriva</h3>

Once upon a time, the Mandriva Linux distribution (formerly Mandrake) was the de-facto standard for home Linux desktops. Then vocal Debian people started spreading a lot of pro-Debian and anti-everything-else FUD, and Ubuntu also became the over-hyped distribution-of-the-moment. Meanwhile, naturally, Mandriva suffered from lack of popularity and negative FUD, but also still continued to improve.

Right now, as Ubuntu Hardy (which I nicknamed "Hardly") tends to hang up or worse so often, and Ubuntu as a general rule seems to be infested with red-tape, it seems that there's an opening for Mandriva and distributions that are very good, but were not as hyped as Ubuntu.

Here are some Israelis who converted away from Ubuntu lately:

• ik_5 has converted to Mandriva after using Ubuntu for a long time
• Diego converted to Mandriva as well
• Meital has many problems with Ubuntu 8.04 (Hardy) and wishes to switch back to Debian
• Nadavvin has spent the weekend installing Fedora instead of Ubuntu which kept getting hanged up

These are all Linux experts and FOSS contributors. It's harder to tell what the common joe who wants to install Linux at home will choose. But I think that they will tend to hear a recommendation from the experts, too.

Will Mandriva become the hottest distribution? Maybe not. But it doesn't matter - Mandriva was never about hype. Naturally, Mandriva has had and still have its share of bugs and problems (mostly non-critical). But so do most other distributions, including such whose users were spreading anti-hype against it.

Writing this from a Cooker (which is the Mandriva equivalent of something like Debian Testing or Unstable) system. <h3>Comeback #3: Perl</h3>

And now we come to the third and final comeback - Perl. You can often here that "Perl is dead", or that "Perl is dying", etc. It's a matter of image, more than technology or mindshare. Many people would blame it on the anti-Perl FUD we can hear from various sources. But I believe the problem has lied in the Perl world itself. To quote an email I sent:

The Perl technology is in an excellent shape. It's fast, feature-rich, powerful, easy to use and learn, under active development, relatively bug-free, with many automated tests, and with many 1st party, 2nd party and 3rd party enhancements (re CPAN).

However, the Perl community is sick. Not dying - but sick.

This sickness was not caused by external attacks. Instead it was created from within the Perl community. By a leadership/intelligencia who were more keen on writing and selling closed books (see my essay about it) than on promoting the core Perl technology, documentation and community. It was caused by an "official" IRC network (which is still the first hit on a Google search for "irc perl") which is full of ego, abuse and hostility - where kicking, bannings (and seemingly K-lines) are not only common, but tolerated and seem desirable. It was caused by administrators and leaders who became too "busy" (and as a result heavily unproductive) to have time for Perl, which is what have been placing food on their table. It was caused by central sites whose source code and data were hidden from the public due to a lot of unknown bad reasons, and for absolutely no good ones. (We advocate openness after all).

But most of all it was caused by the many competent and active Perl contributors who did not take the initiative to replace the existing and sick infrastructure by a more independent and decentralised effort.

(I had written something relatively similar in the pre-Web-2.0 era as part of my "Usability of the Perl World for Newcomers Essay". Nevertheless perlbuzz.com deserves credit for presenting it better, making it public knowledge, and making it more obvious.)

But the situation is now getting better. There are more independent developers who have started to set up sites and help resources of their own. More people blog about their Perl experiences. I'm also going to continue giving my share of the contribution by setting up Perl-Speak (Please don't register yet - the E-mail does not work yet due to problems with my hosting), by contributing to the FAQs, documentation and wikis, and by helping with coding. People like me would also like to complement the inadequate, "official", and completely non-open *.perl.org infrastructure, which has been suffering from bitrot and neglect.

Lately, I've had the pleasure of talking to many people on IRC who have decided to learn Perl for various reasons, and some of them are young or even underage. Along with CPAN and perl5, Perl offers some clear advantages that no language has.

Again, Perl is also not about hype. Some people have privately raved about Perl and recommended it to their friends, but otherwise Perl was promoted quietly, bottom-up and mostly by word-of-mouth. It's hard to know if Perl will become the "hottest" language again. But I believe it will still become more popular, even among new programmers.

Syndicated 2008-07-07 11:41:18 from shlomif