Our policy is to print email addresses of respondents unless they specificaly request anonymity. If you submit a point of view by post please let me know a contact address/phone number if you would like to receive any feedback from our readers. Letters are edited for publication to tidy them up and remove extraneous text.
Thank you for the first issue of Linux@UK. Generally, I though it was a good read, but felt it has a fair way to go before it catches up with the standard of `Linux Journal'. The articles by Zbigniew Chamski and Simon Allen were particularly good.
One of the most common questions recently in the Linux newsgroups has been `When I rebuild the kernel, why is my Linux image so big and why doesn't it work?!' It would have been useful to add to `Alec's Linux Slackware Notes' that zImage no longer ends up in the top directory of the source tree, but in the arch/i386/boot/ directory.
The Lasermoon article entitles `Linux Kernelitus?' was entertaining, but, I thought, potentially damaging to Linux. Their major criticism of Linux seemed to be the rate at which new kernels appear. They state that:
"[Linux will never be taken seriously] when the underlying operating system kernel is changing so fast that accurate testing is both impossible and pointless."This is simply not true. Kernel version 1.2 has just been released. Before that, there has not been an official release kernel since V1.0.9 which was many many months ago; I forget the actual date. All the 1.1.x kernels are development kernels. They are not meant to be used by anyone who wants to use Linux for anything serious and should not be touched by ISVs (Independent Software Vendors). If a customer changes to a development kernel, then he/she should be prepared for the fact that commercial software might not work. The only people using development kernels should be those who are interested in Linux for its own sake and they are unlikely to be purchasing commercial software. If they are, then they should also be competent enough to know how to set up Lilo to boot from alternative versions of the kernel.
The absence of bigname software for Linux is nothing to do with the evolution of the kernel. As always with the big commercial companies, it's a chicken and egg situation getting them to relase software for a new platform. The Commodore Amiga was a wonderful example; despite having an operating system vastly superior to MS-DROS/Windoze, it never took off in the commercial market place because there was no software from the big companies (with the exception of Wordperfect, which, at the time, has a vastly inferior interface compared with wordprocessors developed especially for the Amiga). The big companies wouldn't port there software because there weren't any Amigas in the commercial marketplace.
Linux users need to pester the companies which produce those big applications like Frame, Informix and Oracle. Tell them about Linux; tell them there is a market for their software; even send them old Linux CDs! Most of these companies have heard of Linux by now, but they won't release software until they know there's a marketplace. Some companies, like NAG who have release their Fortran-90 compiler for Linux, are seeing the light. Let's help the spread of Linux by asking companies for the software we want to see.
Dr. Andrew C.R. Martin
First the idea of local user groups seems a very good idea. Perhaps a letter should be sent to linux-uk-discuss asking who would be interested in going to these club meetings and in what area of the country they live. I would be interested and I live in Portsmouth in Hampshire.
I would also be interested in the CD you mentioned. I expect most users purchase copies of the Sunsite archives etc and therefore there would be little use to include all this software on the disk. What would be usefull would be to include all the updates to the software since the last issue of the CD on the CD. The end user could then look for the piece of software they wanted in the ls-lR file and from the file data they would know which of your CD's the file is on.
The exception to this would be perhaps the .../slackware/NOT-FOR-COMMERCIAL-USE directory as I have found that this tends to get omitted from most CD distributions probably because there is no restriction on who can buy the CD. As the SIG is a user club then this problem should not arrise. This above mentioned directory holds many good pieces of software including the latest releases of Mosaic.
Therefore the contents of the CD I would like to see are:-
Gareth Blades (email@example.com)
After much persuasion here is a "letter to the editor" for your newsletter. If it's too late for the first edition then I might like to replace it with something else for the next. I assume you will edit it. Could you send me a membership/subscription form?
I wish to have facilities available under Linux to perform company accounts, financial planning, time management and mail merge (using TeX and a client address database). I have all these under DOS.
I would very much prefer GPL software since I can then read and modify the source code. I am a computer contractor and want these for my personal and company use, though it is possible that in future I may charge to install such software for clients.
I wonder whether any of your readers have similar needs or know of existing
free software? I am thinking of using Tcl/Tk for any front ends I write.
Could anyone recommend a free database package?
The maintainer of the serial code is called Ted Ts'o (apostrophe AFTER the S) rather than Ted T'so.
If you didn't spot this (it's probably too late now to change the
article) perhaps you could put an addendum in a following news letter.
For a little more money you can get an external tape drive that plugs into the printer port. Sorry, mate, not with Linux; it turns out that ftape does not support writing via the parallel port. So I posted a query to the newsgroup comp.os.linux.misc to ask for suggestions. One person said I should get a tower case and transfer all the works. A neat idea, but my motherboard has the cards fitting horizontally so to get more slots I would need to buy another motherboard. Ah, but my video chips are built in to the motherboard so I would have to buy a video card as well and by now the solution is not sounding so good, even without mentioning my three-year warranty.
Give away the scanner to a good home and use the slot for a SCSI card. This from the US where they take SCSI in their rangy stride and regard 1GB as a small disk. I found this idea quite attractive, and it would provide an easy expansion route for adding extra storage in the future, but the cheapest SCSI tape drive is about 900 zlotys so it is an expensive way to solve the present problem.
At work it's all networks nowadays so a colleague suggested a network solution; get another PC and link it up with an ethernet. All I need is a simple machine with lots of bays and slots, two ethercards and 2m of coax and I get to keep the scanner and the warranty. An unwanted and unloved 386SX would do nicely.
Right - I'm off to my local computer auction on Sunday.
Editors Note:One of the beauties of Linux is that if one machine gets too small for your needs you can buy another and another until you have a network setup that does all the jobs you need! With Linux there is no worry of having to buy more software for the new machine (could be more than half the cost of adding another machine in some situations). All the normal thorny legal issues of networking and sharing software simply do not arise!