Christopher B. Browne's Home Page
cbbrowne@acm.org

Footnotes

Christopher Browne

$Id: footnote.sgml,v 1.18 2005-12-28 14:06:08 cbbrowne Exp $

Table of Contents
1. Linux and Standards Compliance
2. Availability of Source Code
3. OS Jokes
4. The BOGUS Distribution
5. MicroChannel Architecture
6. Slurping news
7. Charismatics
8. Liars, **** Liars, Statisticians, and Benchmarks
9. What is the Moscow of New York?
10. Hosting
11. Belling the Cat
12. Apple: Software or Hardware Vendor?
13. IBM: Don't Cannibalize My Market!
14. WINE Configuration
15. Transmeta
16. Client/Server Computing
17. 32 Bit Computing
18. Software "Piracy": Abuse of a Word
19. Massive Transformations
20. The Holmesian Fallacy
21. Godwin's Law
22. What's GNU?

Here are some footnotes to other web pages in my hierarchy. Rather random stuff.

1. Linux and Standards Compliance

Work is ongoing to get Linux to comply more formally with Unix-related standards. Caldera acquired the makers of the Linux-FT distribution which had been validated to comply with the POSIX.1 standard and FIPS151-2.

The Linux-FT development group had joined XOpen, the organization formerly responsible for Unix and its associated standards. (Now it's The Open Group .)

Two areas where they had to make significant change was in the areas of:

C libraries in particular appear to be an area of problems. The Linux C libraries that are typically used were at one point directly based on the FSF/GNU C libraries (aka GLIBC), but have been hacked on rather a lot. There is a distinct need to clean up the changes to them and bring them into conformance with something.

Whether that be to synchronize with FSF libraries (modulo some platform-specific patches, of course) or something else is somewhat arguable, but the need for some commonality is clear. With the proliferation of architectures, the IA-32-orientation of the "traditional" Linux libraries is impeding efforts to have a common set of libraries available across different Linux architectures.

A major "flame war" of Spring 1997 was about this very matter; developers at Cygnus are working towards making the FSF GLIBC 2.0 into the "official" Linux libc. It is already the case for Linux/Alpha; the major rework is in retaining maximum compatibility when doing the same for the x86 architecture. Dissenters are either have a bad attitude towards the FSF (whether valid or invalid is a whole other issue), or feel that any breakage of existing code is a bad thing. (Unfortunately, there already is breakage as soon as you try to cross the boundary between IA-32 and other CPU architectures, as the "traditional" Linux LIBC is quite IA-32-dependent...)

Red Hat Software's "5.0" release now uses GLIBC 2.0, which has, as expected, resulted in a degree of controversy as it sometimes is injurious to the execution of LIBC5-based binaries. Other distributions (Debian, S.u.S.E., Caldera) are also soon adopting GLIBC for IA-32, which should lessen the "roar."

Google
Contact me at cbbrowne@acm.org