In the July 2000 issue, Moshe Bar's article “Linux System Calls” contained information that I would like to correct. I suspect Moshe noticed by now that he was temporarily caught in a pre-i386 time warp when he stated, “This routine sets up an IDT (Interrupt Descriptor Table) with 256 entries, each 4 bytes long. It should be noted that the IDT contains vectors...” This is more a description of the pre-i386 IVT (interrupt vector table) than the IA32's IDT.
The IA32 uses another level of indirection. In particular, the IDT contains 8 byte descriptors which are either interrupt, trap or task gates. I suspect Linux uses the first two varieties. These gates contain the interrupt handler's 32-bit offset, some attribute bits and a 16-bit selector which references a code segment descriptor. In this case, that code segment descriptor is in the GDT (global descriptor table). This descriptor also consists of 8 bytes and vectors to the code segment containing the interrupt handler. The entry point for the handler is derived from the 32-bit offset lurking in the original IDT entry.
—Richard Sevenich rsevenic@netscape.net
I would like to express my support of more articles such as the one on WAP in the July issue. Recently, I was working on a WAP proof of concept for my organization. While I was motivated to move on the project, I was having difficulties taking the relatively new WAP standard and applying it in a straightforward manner to a commercial-quality proof of concept my organization would support. Mr. Mikal's article got to the meat of the issue and demonstrated how to get productive in the WAP space, while leveraging the skill sets that I, and other engineers here, already had. This one-page article got the ball rolling on producing my organization's wireless presence. In doing so, another organization has embraced open-source and free software. It is articles like this, and the knowledge sharing of the Linux community, that make magazines such as Linux Journal welcome in my cubicle.
Thanks to Mr. Mikal and Linux Journal.
—Brad Wheat bradford.wheat@moai.com
When Linux began to grow in popularity, I jumped in, only to be put down by fellow Linux users. I used Linux on a Macintosh. As time went on and Linux distributions for the Mac grew, it became obvious that Linux users would have to acknowledge the Mac. But, rejection continued.
As a Mac user, I was aware of Linux before my PC pals. Apple had MKLinux, “an Apple thing”, I was told. Today, SuSE, the #1 distribution of Linux, is available for the PPC. I'll be merciful and not list all the others. After seeing the ads for PPC and Yellow Dog, it's nice to see a couple of articles, unlike the one degrading MKLinux in LJ a few months ago. Thank you.
—Calvin Bowen ctbowen@mindspring.com
Two things on your Python supplement:
First, the uproar over the naked man. I honestly didn't notice he was naked. After reading the Letters column, I went back and looked; and sure enough, nothing but a bowtie. Considering how much I've seen Laura Croft (and I've never seen or played whatever video game she's in), it only seems fair that some computer magazine would have a naked guy.
As for Guido's pet language (haha! ha! oh never mind...), I must admit I think it's a bit lacking compared to Perl in one way. A fixable way, mind you; this isn't a fatal flaw type of thing. The problem is cpan—Python seems to be missing it. Perl has cpan; pretty much all Perl modules can be found on cpan and installed with the following steps:
perl Makefile.PL make make test sudo make install
Now, I've been looking at Python for a specific problem: I need to take data in a mySQL database and put it into a PostScript printer. Python seems to be able to communicate with mySQL, and its piddle module is both delightfully named and PostScript capable.
However, I still can't get mySQL connections to work, and the piddle install file was bordering on useless (but I got it working). I'll keep at it because I have to, but I think the Python development community should attempt to recreate (and perhaps improve) cpan. As a developer, I've found cpan to be one of the best practical systems to enable code reuse. In fact, the C, C++ and pretty much all development communities would be wise to create their own versions of cpan.
—Kevin Lyda kevin@suberic.net
I just got through reading all of the letters in the July 2000 issue from people complaining about the cover of the supplement. Big deal! I own a computer store smack down in the middle of the bible belt. I put all of my old magazines out front for customers to look at while I am working on their computers, and not one person has complained about the cover. I service people from all walks of life, parents, teachers, clergy, children, etc., and no one has taken offense to the publication...in fact, it has caused Windows-only kind of people to ask about Linux. Have even had a couple of converts because of it.
Regardless of what anyone else says...keep up the good work.
—Rodney Rees rodney@crass-enterprises.com
I read the July 2000 LJ Letters, which was dedicated to reader responses to your Python supplement cover in May. I could well remember that supplement which motivated me to add Python to my language collection. It never occurred to me that the cover picture was offensive.
Mind you, I am a university lecturer, and my pigeonhole is transparent. This picture which annoyed so many in the Western world went here, in an Asian, predominantly Buddhist country, unnoticed!
—Visvanath Ratnaweera ratna@cs.pdn.ac.lk
While flipping through the latest LJ, I saw a small section at the beginning, labeled “Important Linux Web Sites”. What surprised me the most was to see http://slashdot.org/ as the top site listed. This may be a result simply of a sort when putting the list together, but I think many readers would agree that slashdot.org is not a good representation of the Linux operating system or community. Opinions by both the maintainers and visitors of that site are usually overly biased, ignorant or flame bait. If someone were to open an issue of LJ for the first time and wanted to learn more about Linux, I would much rather see them go to a more professionally run site, such as www.linux.com/ or www.gnu.org, than sites that sometimes make me embarrassed to be involved with Linux, like slashdot.org.
—Brian M Dial brian@flylife.org
In the August issue of LJ, Linley Gwennap reported that the Delphi Automotive Systems Palm docking station (or MPC, short for Mobile Productivity Center) uses Windows CE. While WinCE plays a prominent role in Delphi's overall product strategy, it was not an appropriate choice for the MPC. Instead, we chose to use the eCos real-time kernel from Red Hat. I would appreciate it if you would publish this correction in the next issue of LJ.
—Brad Coon MPC Product ArchitectDelphi Automotive Systems bradley.s.coon@delphiauto.com
I was reading my fresh new copy of issue 75 of LJ. In particular, I was reading “Collecting RFCs” and found this first statement: “Requests for Comment (RFCs) are the standards of the Internet”.
It's not really true, because RFCs are what they claim to be, requests for comment from the Internet community. RFCs are the second stage of the life cycle of the “official” Internet community documents. They start as “drafts” proposed by someone. After a while, they become RFCs and then, after having received some comment from someone else, they become STD (Standard), FYI (For Your Information) or BCP (Best Current Practice). So, RFCs are not standard, but only a mature stage of a proposal for a standard.
You can find the same document with two different names, one for the last RFC status and one for the definitive status; e.g., we have RFC-2026 (modified from RFC-1602) and the BCP-9 that is titled “The Internet Standards Process—Revision 3”. Or STD-1 that collects (and obsoletes) RFCs 2500, 2400, 2300, 2200, 2000, 1920, 1880, 1800, 1780, 1720, 1610, 1600, 1540, 1500, 1410, 1360, 1280, 1250, 1200, 1140, 1130, 1100, 1083. For more information about this, try www.rfc-editor.org/rfc.html.
Anyway, I found your article very interesting and love LJ. Go on this way!
—Vincenzo Romano vromano@mail.com
