Hurd Traffic #25 For 1 Dec 1999

By Zack Brown

Do you pine for the nice days of minix-1.1, when men were men and wrote their own device drivers?
Are you without a nice project and just dying to cut your teeth on an OS you can try to modify for your needs?
Are you finding it frustrating when everything works on minix? No more all-nighters to get a nifty program working?
Then this post might be just for you :-)
 
-- Linus Torvalds, 1991

Table Of Contents

Mailing List Stats For This Week

We looked at 81 posts in 245K.

There were 38 different contributors. 19 posted more than once. 12 posted last week too.

The top posters of the week were:

 

1. The Hurd In The Press
7 Nov 1999 - 15 Nov 1999 (15 posts) Archive Link: "discussion on fr.comp.os.linux.debats"
Topics: Bootloaders
People: David MadoreOKUJI YoshinoriMarcus BrinkmannDaniel De KokJim FranklinGordon Matzigkeit

Marcus Brinkmann noticed a thread about the Hurd in the usenet newsgroup fr.comp.os.linux.debats; unfortunately for him it was in French. He asked if someone would care to translate it and post a summary.

He also knew of a Hurd article in the French "Linux Magazine", and asked who had been interviewed. He also noticed that a search on deja.com revealed a surprising number of Hurd references.

Yann Pouillon, who had discovered the Hurd through "Linux Magazine", posted his own translations of the abstracts of several articles appearing in the Issue Marcus had mentioned. Among them was the interview, which turned out to have been with Gordon Matzigkeit, the original english text of which was included last issue in Issue #24, Section #6  (10 Nov 1999: Gordon Matzigkeit Interviewed In French GNU Magazine) .

Elsewhere, David Madore read the entire usenet thread on deja.com, and summarized it:

Starting point is message 537926587 on deja.com: Encolpe Hugues Ascylte DEGOUTE asks for the readers' opinions on the Hurd ``kernel'' and whether they think Linux 2.4 development should be abandoned in favor of trying to obtain a stable version of the Hurd. He explains that he would like to write an article in the bulletin of the ALDIL association (http://www.aldil.linux.eu.org/). Message 537967819, Hugues Marilleau points out that Hurd is not a kernel and makes fun of the suggestion to stop developing Linux. In message 538205151, News Claude explains that the Hurd is a collection of servers running over Mach, and says that the an article in issue number 8 of LinuxMag explains it all, and that this article is to be continued.

In message 539216982, Hugues Marilleau points out that a micro-kernel (sic) can hardly be as fast as a monolithic kernel. In message 539292309, Laurent FAVART points out that the difference in speed might not be all that important, and that microkernels have some other advantages that monolithic kernels can never have, and he gives the URL of the Mach home page. In message 541216245, Jerome Kalifa claims that Mach has been a failure all the way (NeXtStep, MkLinux being replaced by classical Linux and, soon, MacOS); he claims that ``RMS himself did admit that they made a mistake in choosing to host Hurd on Mach because it is too complex and not technically manageable''. In 541233403, Laurent FAVART points out that the failure of Mach may not be the failure of microkernels altogether, and mentions the existence of ``exokernels''.

In message 539477908, Guylhem Aznar compares Hurd with NT2000 and calls them vaporware. This led to a subthread of various sarcasms.

In 540609706, Thierry Pinelli claims that Grub is unable to boot the FreeBSD ELF kernels.

Against the claim that GRUB couldn't boot FreeBSD ELF kernels, OKUJI Yoshinori replied, "This is clearly inaccurate; GRUB can boot FreeBSD ELF kernels, even though FreeBSD's boot loader cannot do."

Marcus thanked David for his summary, and replied whistfully that "it shows at least that we need do to a lot more PR work, as there are various misconceptions floating around."

He went through David's summary, pointing out disagreements along the way. To Hugues Marilleau's point that microkernels could not be as fast as monolithic kernels, Marcus replied, "This has to be determined. Much research has been done on this topic, especiaaly how to speed up RPC's, and there is opinion among people who design distributed or microkernel systems that the performance doesn't need to be worse."

In response to Jerome Kalifa's examples of Mach's failures, Marcus replied, "This doesn't take into account that those are single server OS's, as opposed to the multi server Hurd system."

He concluded that the Hurd was not receiving very good publicity, and suggested that folks try to remedy this by creating a 'beginners' document. Jim Franklin agreed violently, and added that a centralized web repository for all Hurd documentation could do much toward recruiting new developers. Marcus pointed him toward http://www.debian.org/ports/hurd and asked what was missing. Daniel De Kok felt that the most important thing to do was to convince Open Source developers that microkernels were better than monolithic kernels. Jim Franklin asked which came first, the chicken or the egg, and Marcus replied, "The farmer who builds the barn. Get on with the work, guys :) A devils circle can only be cracked by applying brute force at any location, it doesn't matter where you start."

 

2. X Partial Success
15 Nov 1999 - 16 Nov 1999 (4 posts) Archive Link: "Installation feedback,X problems"
People: Marcus BrinkmannMichail Issakov

Michail Issakov reported partial success with X 3.3.3.1; amazingly, it would actually start, although he couldn't use it for anything because he could only get it running if he disabled the mouse. Marcus Brinkmann replied, "Wow. That it works at all is surprising for me :)"

Under the subject, X works (http://www.debian.org/Lists-Archives/debian-hurd-9911/msg00127.html) , Michail said that the mouse problem had been his fault. He now had X up and running, though it would crash if he tried to actually do anything with it.

 

3. New Version Of GRUB
16 Nov 1999 - 18 Nov 1999 (9 posts) Archive Link: "Newbie to Hurd"
Topics: Bootloaders, FS: FAT32
People: Gordon MatzigkeitOKUJI YoshinoriJim Franklin

Someone had a problem with the Hurd recognizing their partitions, and in the course of discussion, Jim Franklin forwarded an announcement from Gordon Matzigkeit:

I've released version 0.5.93.1 of the GRand Unified Bootloader. You can get it from:

ftp://alpha.gnu.org/gnu/grub/grub-0.5.93.1.tar.gz

Check out http://www.gnu.org/software/grub/grub.html for more information. OKUJI Yoshinori has been doing the vast majority of the development work on GRUB. I'm acting mostly as the Debian maintainer for GRUB, as well as making releases.

Debian packages of 0.5.93.1 are already available in `potato' for the i386 architecture, and should soon be built for hurd-i386 as well.

Here is the relevant NEWS (for 0.5.93, since 0.5.93.1 only adds a few bugfixes):

New in 0.5.93 - 1999-10-30:

  • ELF format of FreeBSD kernel is supported.
  • Support the partition ids for NetBSD and OpenBSD.
  • Exit from the grub shell just by pushing the key `q' in the menu.
  • New options for configure can disable some functions in Stage 2. See the output from `configure --help' for more information.
  • FAT32 support is added.
  • Minix fs support is added.
  • New commands "hide" and "unhide".
  • The character `=' after a command is not necessary any longer, but it is supported for backward compatibility.
  • The command "help" displays helpful information about builtin commands.
  • The command "geometry" displays the information of a drive specified and set the geometry to arbitrary C/H/S values if the optional arguments are used.
  • The command "configfile" loads a configuration file interactively.
  • The command "device" assigns a BIOS drive to an arbitrary filename in the grub shell.
  • The option `--no-floppy' force the grub shell to assume that there is no floppy, and the option `--probe-second-floppy' enables the probe of the second floppy drive.
  • Integrated the netboot support in the Dresden version of GRUB.
  • FreeBSD support in the grub shell is improved.
  • Killing (C-u and C-k), yanking (C-y) and manipulating the history (C-p and C-n) are supported.
  • The address argument for the command "install" is now optional.
  • Better completion support.
  • The command "cat" displays the contents of a file.

Have fun!

 

4. New GRUB 0.5.93.1 Boot Floppy
18 Nov 1999 - 20 Nov 1999 (7 posts) Archive Link: "GRUB 0.5.93.1 boot floppy released"
Topics: Bootloaders, FS: ext2
People: Gordon MatzigkeitMatthew VernonPaul Emsley

Gordon Matzigkeit pointed out that a lot of problems reported to the list were actually caused by people using a very old version of GRUB (v. 0.4). He explained:

I was puzzled about where people were getting it from. After some investigation, I found that it's because the Easy Guide recommends that people use gnu-0.2/grub-boot.image from their GNU mirror.

That file is sadly out-of-date (and won't be updated, since it belongs to a past release), so I've prepared a new ext2fs mini boot floppy with an updated version of GRUB on it:

ftp://alpha.gnu.org/gnu/grub/grub-boot-0.5.93.1.image

The image is just big enough to hold all the boottime GRUB files, and a config file. If there is demand for a full 1.44MB or 1.2MB image, then I can make one... I just thought people would appreciate the low bandwidth requirements of a mini image.

He asked Matthew Vernon to update the "Easy Guide", and Matthew replied that the official version had been updated for some time; it was only the version on the gnu site that was old. He added that until he got a password to upload new versions, that situation was unlikely to change. Paul Emsley asked what Matthew meant by the "official version", and Matthew replied with a pointer to the official version (http://www.pick.ucam.org/~mcv21/hurd.html) .

 

5. Synchronizing Clocks
18 Nov 1999 - 19 Nov 1999 (5 posts) Archive Link: "How to synchronize clocks?"
Topics: FS: NFS
People: Bill WhiteRoland McGrath

Bill White had two machines (one Linux, one Hurd), and found that doing native compiles on an NFS-mounted filesystem would cause errors, because their clocks were not synchronized. The answer seemed to be to set up an xntp server on the Linux box, with the Hurd box as its only client. He asked if this seemed like a good idea to folks.

Roland McGrath did not remember anyone else having tried porting xntp to the Hurd, and felt Bill was on the right track. Bill replied that instead of xntp, ntp4 was the proper package to port, since it was supposed to be xntp's successor. He'd gotten errors trying to compile it, but hadn't had a chance to investigate yet.

End of thread.

 

6. TeX Success
19 Nov 1999 - 24 Nov 1999 (12 posts) Archive Link: "My Hurd speaks PostScript"
People: Chris LingardMarcus Brinkmann

This issue came up in Issue #24, Section #1  (6 Nov 1999: Problems Building The Hurd By Hand) , when Chris Lingard had had trouble installing TeX. This time, Chris had managed to compile tar, and he summarized his efforts. Marcus Brinkmann was very happy about this, but cautioned Chris not to make changes to file locations unless they were absolutely necessary. Doing so would make merging his port into the official sources that much more difficult.

 

7. Booting From A Loopback FS
26 Nov 1999 (3 posts) Archive Link: "Booting the Hurd from a loopback FS?"
People: Nolan DarilekRoland McGrath

Nolan Darilek said, "I created a 200 meg file and ran mke2fs on it. I then mounted it and ran the cross-install script, so I now have a Hurd disk image ready to boot. I'm assuming I can just use dd and write this to a disk partition if I like what I see." He asked if it would be possible to boot from this image. Roland McGrath said yes it would, and suggested he search the mailing list archives. Nolan did this and found http://www.debian.org/Lists-Archives/debian-hurd-9907/msg00092.html, which he posted to the list. He had some problems with the procedures described in the post, but the thread ended there.

 

 

 

 

 

 

Sharon And Joy
 

Kernel Traffic is grateful to be developed on a computer donated by Professor Greg Benson and Professor Allan Cruse in the Department of Computer Science at the University of San Francisco. This is the same department that invented FlashMob Computing. Kernel Traffic is hosted by the generous folks at kernel.org. All pages on this site are copyright their original authors, and distributed under the terms of the GNU General Public License, version 2.0.