Linux Journal has gone through some major changes during its first few months of existence, and Linux itself has gone through a few. However, some things have not changed.
by Michael K. Johnson
That's me. One of the changes here at Linux Journal is the new editor. My name is Michael K. Johnson, and I have used and hacked Linux since it was first able to boot itself late in 1991. I have made some small kernel modifications ("hacks" in Linux-speak) and maintained several programs for Linux. I started the original "man project" in early 1992, and helped form the "Linux Documentation Project" which was formed later to write a complete documentation set for Linux.
SSC, publishers of UNIX reference guides and books for over 10 years, is now the publisher of Linux Journal. By consolidating most of the work on Linux Journal in one place (Seattle, Washington), we hope to avoid the early problems that Linux Journal encountered when it was produced cross-country between Washington and New York.
While Linux Journal was experiencing its birth pangs, Linux version 1.0 was released. Version 1.0 was not quite bug-free, and in fact was barely different from the versions that preceeded it. The biggest difference was one of philsophy.
Previous to version 1.0, a new version of Linux would come out every week or two, and many people would upgrade right away. No one would know whether or not some new feature in the new version had a bug, and some people were bitten by new bugs. This caused many others to avoid upgrading to the most recent kernel for fear of encountering a bug.
Eventually, Linux versions were discussed as if they were vintages of wine. "I have never trusted any of the patch level 13 kernels." "Version 0.99 patch level 12 was a nice, stable kernel." "Ah, I remember version 0.95. It was so simple, clean, elegant..." A folklore sprang up about which versions should be run by which people.
With Linux 1.0, Linus Torvalds (the main author of Linux) devised a new scheme to avoid this confusion. Linux 1.0 was the start of a line of Linux versions with no new features, and hopefully no new bugs. Linux 1.0.1 fixed a few bugs, Linux 1.0.2 fixed more, and so on. However, no Linux version 1.0.<anything> contains any new features that Linux version 1.0 did not have, and therefore hopefully will contain no new bugs. These versions are intended to be safe.
New features are now being introduced in a new set of different versions: Linux versions 1.1.x. Linux 1.1 contained many new features that were not in 1.0. Each new version 1.1.x may or may not contain new features, and is likely to contain new bugs. These new versions are intended for people who like to play with new features and are willing to risk bugs. These versions are intended to be fun.
In general, any "even" version of Linux, like 1.0.x or 1.2.x, will be intended to be safe, and any "odd" version, like 1.1.x or 1.3.x, will be intended to be fun.
Linux 1.0 is still free; still licensed under the GNU public license. It is perhaps worth explaining what "free" means here. It does not mean free of monetary cost, although you do not have to pay to get a copy of Linux unless you want to. It means that the source code is available to anyone, and that anyone who makes changes to Linux must release the source code for their changes if they release their changes at all. This will ensure that Linux remains a completely open system.
More and more commercial software is being made available for Linux. As Linux users keep asking software companies when they are going to release a Linux version of their product, more and more companies have decided that they need to port their product to Linux. Linux Journal's presence at trade shows has increased interest among some vendors to port their applications to Linux.
However, other companies are scared away by the "counter-culture" appearance of Linux, confused about whether or not it is legal for them to release commercial software that runs on Linux, or simply not interested in porting their software.
Eric Youngdale, Al Longyear, and several other hard-working Linux hackers have been working on "iBCS2 compliance" for Linux. This weird-looking acronym (It stands for intel Binary Compatibility Standard) means that it is now possible to run at least some binaries intended for SCO Unix or for any of the many versions of SVR4, including UnixWare.
The SCO version of WordPerfect, for instance, runs, and even works under X. The SCO version of Xess, an X-based spreadsheet, is also known to work, and there have been reports of databases running as well. The iBCS2 team is interested in hearing success reports, which they will post in a list so that users can know whether software works under Linux before puchasing the software.
The iBCS2 patches have now been publicly released. The product isn't perfect yet—in fact, it is still in alpha testing as I write—but if you want to, you can play with it. Statically linked binaries seem to work fairly well, but the shared libraries for SCO and SVR4 are still being written, and do not work as well.
The IBCS2 patches are available from tsx-11.mit.edu in /pub/linux/ALPHA/ibcs2/ as I write, but may well have moved to /pub/linux/BETA/ibcs2/ by the time you read this. There is a README file in that directory which explains the rest of the files. The more people that try these patches, the better this iBCS2 support can get.
Linux Journal is still committed to giving its readers what they want and need. We really want your feedback. Please tell us what you like and dislike about Linux Journal, and what you think we should do to improve. You can send e-mail to ljeditor@ssc.com, or send paper mail to Linux Journal, P.O. Box 85867, Seattle, WA, 98145-1867, USA. You can phone us at (206)524-8338 or fax us at (206)782-7191, if you like.