From: Digestifier To: Linux-Development@senator-bedfellow.mit.edu Reply-To: Linux-Development@senator-bedfellow.mit.edu Date: Sat, 15 Oct 94 21:13:07 EDT Subject: Linux-Development Digest #314 Linux-Development Digest #314, Volume #2 Sat, 15 Oct 94 21:13:07 EDT Contents: Re: Linux 1.1.52 (Lies, Damned Lies, and Benchmarks) (Jeff Kuehn) Re: Kernel 1.1.53 - no BOOM (Larry Butler) Re: LINUX Logical volumes (Dennis Heltzel) Re: We a FAQ: Linux vs. *BSD!!! (Jesus Monroy Jr) Re: We a FAQ: Linux vs. *BSD!!! (Jesus Monroy Jr) Re: Anyone working on improving NFS? (Amrik Thethi) Re: We a FAQ: Linux vs. *BSD!!! (Jesus Monroy Jr) Re: ext2fs vs. Berkeley FFS (Basile STARYNKEVITCH) Re: ext2fs vs. Berkeley FFS (ron house) Re: New Motif lib's for use with XFree 3.1 ? (Daniel Supernaw-Issen) ---------------------------------------------------------------------------- From: kuehn@citadel.scd.ucar.edu (Jeff Kuehn) Subject: Re: Linux 1.1.52 (Lies, Damned Lies, and Benchmarks) Date: 14 Oct 1994 21:46:08 GMT Dominik Kubla (kubla@Uni-Mainz.DE) wrote: : In article <37k9ss$dha@ncar.ucar.edu> Jeff Kuehn writes: : Shaune Beattie (sdgb1@cus.cam.ac.uk) wrote: : : ok, first off, obviously most of the tests are faster by ~2 times... : : (would have expected slightly more... but thats benchmarks for you :) : : the sole reason I'm posting this is to draw your attention to the Shell : : scripts 2,4,8... a factor of 2 is to be expected... but there is *NO* way : : my machine is 80 times faster than yours... I really think something : : might have gone astray there. Just that if you are spending time : : comparing kernel speeds (a task I don't envy after only running the : : benchmark 3/4 times) then it might be wise to look into this. : [...] : Which version of the libraries/compiler/ld.so/shell are you using? : ^^^^^ : That is the point. Usually under linux /bin/sh is linked to /bin/bash. : But imagine if it was either pdksh,ash or a stripped down bash (that is : using config.h.mini resulting in a bash w/o history and readline) or : even a zsh. That will make a lot of difference. : I think that the library or the compiler are only do little to the : performance (but i might be wrong). I am not so sure about ld.so ... : Thanks to all the folks who are running the benchmark as well. This is a : time consuming process and you all deserve a share of the credit! : What about using the SPEC benchmark ? v1.2 is available on FTP. I will try to : run the suite over the weekend. : Dominik The problem has been found. The version of time(1) on my slackware 2.0 system produces output which is not compatible with one of the awk(1) scripts used to do the benchmark's data reduction (loopm.awk). Because this awk script wasn't correctly parsing the timing, it was incorrectly reporting the number of loops per minute as 1.0. The problem should only have affected the shell tests as only those are reduced with this particular awk script. I'll be rerunning the tests this weekend so I can put out corrected results early next week. My apologies for any wild goose chases I've started w.r.t. scheduler problems. (see why the subject is, "Lies,..." :-) Thanks to David Niemi and Shaune Beattie for checking results. Hopefully this is the only error. --Jeff Kuehn, NCAR/SCD ------------------------------ From: butler@cs.tulane.edu (Larry Butler) Subject: Re: Kernel 1.1.53 - no BOOM Date: 14 Oct 1994 07:10:19 GMT In article , Steven M. Doyle wrote: >In <1994Oct11.171749.2385@ka4ybr.com> mah@ka4ybr.com (Mark A. Horton KA4YBR) writes: >>Console: colour EGA+ 132x44, 24 virtual consoles >>Serial driver version 4.00 with no serial options enabled >>tty00 at 0x03f8 (irq = 4) is a 16550A >>tty01 at 0x02f8 (irq = 3) is a 16550A >>tty02 at 0x03e8 (irq = 4) is a 16550A >>tty03 at 0x02e8 (irq = 3) is a 16550A > >One interesting point is the sharing of IRQ's between tty0/2 and tty1/3. This may >be causing part of your problem (only thing I can suggest not knowing exactly how >your link is set up) > This may be stupid, but I was wondering why the interrupt handler can't check for received data in both UARTs. If it is possible to poll the UARTS why can't this be done? Would it make the interrupt handler too time consuming? Just a thought, Larry ------------------------------ From: dheltzel@crl.com (Dennis Heltzel) Subject: Re: LINUX Logical volumes Date: 14 Oct 1994 00:39:04 -0000 Richelo Killian (killianr@beldin.sun.ac.za) wrote: : Is it posible to create logigal volumes across drives and/or partitions and then mount a single filesystem on that volume? I know it can be done on HP-UX, but I want to do it on my LINUX box? I've used logical volumes on IRIX a little. Worked pretty nice. I don't know if it would be worth the trouble to implement. Perhaps someone will write it as an optional add-on to an existing FS, so only people who really want it need to incur the overhead & risk. On a related note, I am S/A on a UNIX system that implements "transaction logging" (I'm sure it is called something else on other unices). This is really great for data integrity and actually improves performance. The basic system is that all FS writes are made to a special log partition as change records. A daemon uses these records to update the FS and then mark the record as deleted. This works in a sort of circular queue, with the daemon monitoring disk activity to only do FS writes when the drives are quiet or the queue is near capacity. The system boot sequence checks this partition and performs an update to the FS for all records in the partition. This means that a power failure only endangeres those changes not written to the log partition. Writes to the log partition are not buffered and are rather small and require little head movement. Obviously, all reads must go through this system also to detect and correct reads from blocks affected by logged records. I know this is not trivial to implement, but it works great, fsck doesn't even need to be run on a FS like this after a crash. It requires minimal extra disk space and greatly enhances the logical integrity of the FS. Just my $.02 Dennis ------------------------------ Crossposted-To: comp.os.386bsd.development,comp.unix.bsd From: jmonroy@netcom.com (Jesus Monroy Jr) Subject: Re: We a FAQ: Linux vs. *BSD!!! Date: Sat, 15 Oct 1994 20:58:49 GMT Ken Hughes (hughes@napa.eng.uop.edu) wrote: : Jordan K. Hubbard (jkh@freefall.cdrom.com) wrote: : : In article jmonroy@netcom.com (Jesus Monroy Jr) writes: : : This is a weekly question. : : More often than not, we get into a flame war : : on this. Let's stop this silliness!!! : : The only way we're going to stop this silliness is to simply start : : ignoring the querants. If someone asks "Which is better? Which is : : better?", jumping up and down all the while, and everybody just flat : : out _ignores_ the question and goes about their business as if nothing : : happened, folks will eventually get the point and stop asking. : : Consider carefully: It's not the questions that start the bloody flame : : wars, it's everyone's pathetic attempts to answer! "Well, xxx is : : better because of yyy.." "No it's not!" "Yes it is, you moron! Just : : look at blah blah blah!" "Well, you're a complete idiot who obviously : : wouldn't know an operating system if it bit you - yyy is _obviously_ : : better because bleh bleh bleh!". And so downhill it goes from there. : : :: [deleted counter arguement] :: : : 386BSD-1.0", so please don't start THAT thread again either). IMHO, the : only chance for resolving this issue would be for someone to sit down and : compile a list of features that shows which OS has what and which doesn't. : This is another good idea. A list of features.. is always asked for. : However, there's little hope of keeping this list up to date give the state : of change of each OS, unless someone is willing to form a team for this : purpose. : I disagree, Dave Burgess did a great job on the "NET/2 *BSD" FAQ, so I believe an appropriate person should be nearby. I'm not saying Dave should do it (he's to busy), but I can bet there such a person available.... a neutral to alll parties. : I wonder how the Windows-DOS-OS/2 people react to these questions? Does : anyone ever ask over in those groups? Or is everyone there resigned to the : fact that OS/2 is best or Windows is best or Chicago/Windows 95 will be best : or that they're all SOL? :-) : Good point. Should we extend the discussion to the other groups? Do I hear a Yeah on this call? -- Jesus Monroy Jr jmonroy@netcom.com Zebra Research /386BSD/device-drivers /fd /qic /clock /documentation ___________________________________________________________________________ ------------------------------ Crossposted-To: comp.os.386bsd.development,comp.unix.bsd From: jmonroy@netcom.com (Jesus Monroy Jr) Subject: Re: We a FAQ: Linux vs. *BSD!!! Date: Sat, 15 Oct 1994 21:03:34 GMT Andreas Helke (andreas@orion.mgen.uni-heidelberg.de) wrote: : Ken Hughes (hughes@napa.eng.uop.edu) wrote: : : I wonder how the Windows-DOS-OS/2 people react to these questions? Does : : anyone ever ask over in those groups? Or is everyone there resigned to the : : fact that OS/2 is best or Windows is best or Chicago/Windows 95 will be best : : or that they're all SOL? :-) : You only have to look at the multiple 100 post flamewars that regulary come : up in comp.os.ms-windows.advocacy. The Windows solution is simply to assign : a special newsgroup for this discussion. : The "Windows solution"... it sounds like an oxymoron. Sorry, that's a low shot, but we need to go beyond the convention, because it is efforts (and good technical threads -- like this one is becoming) that will lead people back to *BSD and LINUX and not the a solution that does not benifit ourselves. -- Jesus Monroy Jr jmonroy@netcom.com Zebra Research /386BSD/device-drivers /fd /qic /clock /documentation ___________________________________________________________________________ ------------------------------ From: at@setanta.demon.co.uk (Amrik Thethi) Subject: Re: Anyone working on improving NFS? Date: Thu, 13 Oct 1994 12:38:43 GMT In article stock@cs.buffalo.edu (Matthew D Stock) writes: >Hi. Is anyone currently working on improving NFS under Linux? If I >remember correctly, one of the big reasons it has performance problems is >because the caching is done using disk geometry information, which NFS >doesn't have. > >Is my information out of date? In any case, I'm interested in wokring on >the problem, so if you have information on where I should start, please let >me know. The latest kernels (> 1.1.50 I think) have a simple static 5 block cache in the NFS client. This improves average performance significantly (~50%). It is still not multi-threaded in anyway, so multiple processes using NFS filesystems will be seriously hit. I think the plan is to modify the system cache so that it can be used with NFS, but this wont happen before 1.2. Also the client code must be made muli-threading if you want decent muli-user performance, I don't know if there are any plans for that. -- Amrik Thethi. Tel. +223 421 008 Fax. +223 421 024 Setanta Software Ltd. Internet: at@setanta.demon.co.uk Cambridge, UK. ------------------------------ Crossposted-To: comp.os.386bsd.development,comp.unix.bsd From: jmonroy@netcom.com (Jesus Monroy Jr) Subject: Re: We a FAQ: Linux vs. *BSD!!! Date: Sat, 15 Oct 1994 21:14:28 GMT Jordan K. Hubbard (jkh@freefall.cdrom.com) wrote: : In article <37mflh$f6s@unix1.cc.uop.edu> hughes@napa.eng.uop.edu (Ken Hughes) writes: : Normally I would agree, but in this case I doubt that ignoring the : questions will stop them from being asked. New people come into these : Read what I said more carefully, please. I didn't propose to stop the : questions, I simply proposed to stop the answers. The questions will : keep coming in, but will "quench" themselves far more quickly if : people simply ignore them. We're always going to have this question, : there's no doubt, so the best we can do is try to minimize the amount : of damage it does. : Jordan, please consider the damage that a newbie could do, if they answered the question. That alone should convince you of the unsoundness of your idea. Yes, there is a big problem with this idea, but we have a bigger problem. That is, I'm tired of agruing with you (and a hunderd) other converts (sp?). I know for a fact I can't get you to do the 386bsd thing with me, so why should I bother. By the same token, I am not going to allow you to "bad-mouth" 386bsd. Given the oppertunity (sp?) I will continue. You know my staying power in this issue and we won't have any winners in an on going "flame-war". This is a good solution so that we my go back to work. Mind you, next week I'm interviewing for a job of about $70K. I won't fight you, but I can sure make it so you can't win.... resort to reason... rethink your strategy. : 386BSD-1.0", so please don't start THAT thread again either). IMHO, the : only chance for resolving this issue would be for someone to sit down and : compile a list of features that shows which OS has what and which doesn't. : I don't agree. As you seem to concur, the FAQ will be out of date and : essentially useless far too rapidly to make the effort of writing it : anything but a wasted one. : If you re-read my original statement, I said each appropriate group would maintain their respective portion of the FAQ. If you're going to have problems updating a simple FAQ list, how in the hell to plan to manage 4 terabytes of data ( the virtual space of a Pentium). -- Jesus Monroy Jr jmonroy@netcom.com Zebra Research /386BSD/device-drivers /fd /qic /clock /documentation ___________________________________________________________________________ ------------------------------ From: basile@rosser.serma.cea.fr (Basile STARYNKEVITCH) Subject: Re: ext2fs vs. Berkeley FFS Date: 11 Oct 1994 18:10:55 GMT >>>>> "Albert" == Albert D Cahalan writes: In article adc@zeta.coe.neu.edu (Albert D. Cahalan) writes: (about structured files with forks -making them dirs.) Albert> 1: File operations do not work the same. Try gzipping a Albert> directory without tar. Albert> 2: There is no way to recognize these directories as Albert> complete units. Albert> 3: File managers will open them as directory trees, Albert> because that is what they are, NOT record type files. -- I agree on that. However, i would like to say (with other people) that this is the Unix way of doing. This is not neccesarily a good thing. The Unix plain byte-stream file paradigm is outdated for todays needs. But a more structured file paradigm requires IMHO a complete rewrite of all the operating system: 1) Obviously, the kernel must change -not only the filesystem layers, but also new (or extended) system calls to manipulate file structure. 2) Reasonably, the executable format of programs should take advantage of such structured files (and looking into recent executable or object formats such as ELF or perhaps COFF easily advocates the need of structured files). 3) Hence, the execution model (ie the execve call or its successors) should also change. 4) all IO libraries should support the better IO model. 5) perhaps the languages should better support a structured IO model. 6) Probably, there won't be a universally agreed upon file (meta) structure. Hence, there should be several kind of files; this advocates file servers and microkernel techniques. (perhaps TRON model of files -which contains stream of either bytes or filetables like directories- might be interesting). This won't be Unix anymore. this won't be Linux neither. It is time to wait for Hurd or VSTa... -- Basile STARYNKEVITCH ---- Commissariat a l Energie Atomique (basile@soleil.serma.cea.fr) -only my opinions- --- -- Basile STARYNKEVITCH ---- Commissariat a l Energie Atomique DRN/DMT/SERMA * C.E. Saclay bat.470 * 91191 GIF/YVETTE CEDEX * France fax: (33) 1- 69.08.23.81; phone: (33) 1- 69.08.40.66 email: basile@soleil.serma.cea.fr; homephone: (33) 1- 46.65.45.53 N.B. Any opinions expressed here are solely mine, and not of my organization. N.B. Les opinions exprimees ici me sont personnelles et n engagent pas le CEA. Please cite a small part of my mail in all answers Veuillez citer une petite partie de mon courrier dans vos reponses ------------------------------ From: house@helios.usq.EDU.AU (ron house) Subject: Re: ext2fs vs. Berkeley FFS Date: Fri, 14 Oct 1994 07:29:17 GMT plm@atcmp.nl (Peter Mutsaers) writes: >Unix's idea of plain byte-streams has been very successful and >provides a lot of flexibility because assumptions on the contents of >files are put into user space. I see no recent developments whatsoever >that have changed such needs and have outdated this paradigm. I agree. I once calculated how many distinct file types were available on the HP3000 under MPE. I don't remember the number, but it was in the millions of trillions when you combined all possible parameter values. The most obvious characteristic of that OS was its bugginess when you wanted to put something in a file that HP hadn't anticipated. -- Ron House. USQ (house@usq.edu.au) Toowoomba, Australia. ------------------------------ Crossposted-To: comp.windows.x.i386unix From: daniel@austin.ibm.com (Daniel Supernaw-Issen) Subject: Re: New Motif lib's for use with XFree 3.1 ? Date: Thu, 13 Oct 1994 15:20:30 GMT In article <37hegv$4ql@freya.yggdrasil.com>, adam@yggdrasil.com (Adam J. Richter) writes: > In article <37c1gb$lo@tartan.metrolink.com>, > Craig Groeschel wrote: > >In article <374nup$aap@freya.yggdrasil.com>, > >Adam J. Richter wrote: > >>We had an X11R6 > >>beta release that used a downward compatible version version number for > >>its shared libraries and seemed to work fine with the R5 binaries that > >>we tried. > > [much deleted] Ok, Question: I understand that the yggdrasil xfree3.1-beta was built with the old version number - is yggdrasil making available xfree3.1 (non-beta) built the same way? And if so, is it available via anonymous ftp? It would be much appreciated as I'm one of those motif users who's gotten a bit burned by all this. Daniel Supernaw-Issen i speak for nobody but myself - I certainly don't speak for IBM. -- My other life is worth living. ------------------------------ ** FOR YOUR REFERENCE ** The service address, to which questions about the list itself and requests to be added to or deleted from it should be directed, is: Internet: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.development) via: Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU Linux may be obtained via one of these FTP sites: nic.funet.fi pub/OS/Linux tsx-11.mit.edu pub/linux sunsite.unc.edu pub/Linux End of Linux-Development Digest ****************************** .