From: Digestifier To: Linux-Misc@senator-bedfellow.mit.edu Reply-To: Linux-Misc@senator-bedfellow.mit.edu Date: Fri, 3 Sep 93 04:13:08 EDT Subject: Linux-Misc Digest #87 Linux-Misc Digest #87, Volume #1 Fri, 3 Sep 93 04:13:08 EDT Contents: Re: SLS considered harmful (wasRe: Bashing Peter MacDonald) (James A Robinson) NFS performance under OS/2, Unix-Intel and its competitors (Dan Ts'o) Re: Windows Pop Quiz Re: NT versus Linux (james shoemaker) Re: Linux and Corporate America (Peter Mutsaers) Re: Low Cost SLS 1.03 on Diskette (Gold Silver Soup and Silk) Re: Stacker-like Compression? (Stephen Sprunk) Re: help with SIGTSTP (more) (Simon J Ferrett) Re: Linux Systems Laboratories ??? (Pauline Price) Re: Stacker-like Compression? 16 Responses (steveo) Re: Wordprocessor under X (steveo) Re: Checker, a gnu Purify subset (Philip Brown) Re: Wordprocessor under X (Philip Brown) ---------------------------------------------------------------------------- From: jcg@world.std.com (James A Robinson) Subject: Re: SLS considered harmful (wasRe: Bashing Peter MacDonald) Date: Fri, 3 Sep 1993 01:09:51 GMT mdiehl@unm.edu (J. Michael Diehl) writes: >Ok, here is the the question that has always burned in my heart. I figured that >as soon as I had a machine to run Linux on, I'd ask, but now seems as good a >time as any. Just what is meant by a "distribution?" Why can't we just dload >related programs as a "package?" We could get the boot disk, the root disk, and >the compiler package, the X package, network package....etc. Then we could >simply dload our favorite install script for each package and people would be >supporting their own install-script-religion. ;^) This would be much simpler >than fighting over entire directories of packages. You already CAN do this. That is what the sunsite.unc.edu and tsx-11.mit.edu archives are for. The distributions are meant for people who either do not have the resources (my case), ability (also WAS my case), or time to put together *everything* one really needs to call their system a fairly complete setup. Can you imagine trying to get and install all 500+ utilities that come with SLS and propbably MCC? Right there is a compatiblity and standards mess i.e., /usr/bin or /usr/sbin, /usr/ucb, etc... Jim jimr@world.std.com ------------------------------ From: tso@cephalo.neusc.bcm.tmc.edu (Dan Ts'o) Crossposted-To: comp.os.os2.networking,comp.protocols.nfs,comp.unix.xenix.sco,comp.unix.solaris,comp.os.ms-windows.nt.misc Subject: NFS performance under OS/2, Unix-Intel and its competitors Date: 3 Sep 1993 01:40:00 GMT I am looking for a 32-bit OS/platform for an application of mine that will be running large programs and read large data sets from NFS servers. I know that I should consider getting a cheap Sun or the like, but at the moment am considering an Intel box (e.g. Gateway 486/66 or similar) running some PC OS. Actually the programs that I have run fine with a 32-bit DOS Extender. BUT Sun's PCNFS is too slow: a factor of three slower than the same operation on a Sparcstation 2, i.e. large NFS reads. I would appreciate it if some kind souls would run the simple program I have below on OS/2 running NFS, any of the UNIXes on Intel (Solaris, SCO, Linux, Next, etc.), Windows NT if possible, etc. I would like to know the relative performance of large NFS reads on OS's available for PC's. Here are the numbers I have so far. The program is invoke as $ time nfsread BIG_NFS_FILE 40 seconds (time for 10 reads) where BIG_NFS_FILE is the pathname of a large (>12Mb) file that sits on some NFS server. I realize that there will be some effective depending on the NFS server itself. I used a MVAXIII running Ultrix 4.3, and an SS2. for a 1200000 byte read: Server: MVAX III Server: SS2 MVAXIII, Ultrix 4.3 7.0 sec 4 sec Sparcstation 2 3.2 sec 2.2 sec 486DX2/66, 3C509, DOS/Watcom 9.2 sec 4 sec Note that I have found that for many CPU-intensive operations, a 486DX/66 is similar in speed to the SS2. But the PCNFS speed is very slow. So I'm hoping that its the PCNFS implement and that a 32-bit implementation like OS/2 or UNIX for Intel will be a lot faster. Please email responses and results. Thanks. Cheers, Dan Ts'o Div. Neuroscience 713-798-3331 Baylor College of Medicine 1 Baylor Plaza S603 dan@dna.neusc.bcm.tmc.edu Houston, TX 77030 tso@cephalo.neusc.bcm.tmc.edu ---- nfsread.c #include #include #define NFSFILE "A_BIG_NFS_FILE" #define NREAD 1200000 #define COUNT 10 char *malloc(); main(c, v) char **v; { char *p; char *file; int fd, i, count, f; int nread; long t, t1; if (c > 1) file = v[1]; else file = NFSFILE; #ifdef DOS fd = open(file, O_RDONLY|O_BINARY); #else fd = open(file, 0); #endif if (fd < 0) { fprintf(stderr, "%s: Bad open\n", file); exit(1); } if (c > 2) nread = atoi(v[2]); else nread = NREAD; if (c > 3) count = atoi(v[2]); else count = COUNT; p = malloc(nread); if (p == NULL) { fprintf(stderr, "%d: Bad malloc\n", nread); exit(1); } time (&t); for (i = 0; i < count; i++) { f = read(fd, p, nread); if (f != nread) { fprintf(stderr, "%d: Bad read\n", f); exit(1); } printf("%d\r", i); fflush(stdout); } time (&t1); printf("%d seconds\n", (t1 - t)); close(fd); exit(0); } ------------------------------ From: james@george.UUCP (james shoemaker) Crossposted-To: comp.os.ms-windows.advocacy Subject: Re: Windows Pop Quiz Re: NT versus Linux Date: Tue, 31 Aug 1993 08:57:10 CST tlilley@muddcs.claremont.edu (Ted) writes: > In article <25ve6n$j8q@news.u.washington.edu> tzs@hardy.u.washington.edu (Tim > Smith) writes: > > Mark A. Davis wrote: > >That presupposes that MS-"Windows" is an operating system, which by all > >definitions I have known, it is not. Current temperature: 68 degrees? > > What are your definitions? Back each one up with a cite to a major operat > ing > systems text, please. > > --Tim Smith > > > Well, here's one I did a while ago for your reading pleasure: > [....................] > > What is an OS? > ------------- > > Coffman and Denning: "A computer system may be defined in terms of the > various supervisory and control functions it provides for the processes > created by its users: > 1. Creating and removing processes. > 2. Controlling the progress of processes, i.e., ensuring that each > logically enabled process makes progress and that no process can block > indefinitely the progress of others. > 3. Acting on exceptional conditions arising during the execution > of a process, e.g., arithmetic or machine errors, interrupts, addressing > errors, illegal or privileged instructions, or protection violations. > 4. Allocating hardware resources among processes. > 5. Providing access to software resources, e.g., files, editors, > compilers, assemblers, subroutine libraries, and programming systems. > 6. Providing protection, access control, and security for > information. > 7. Providing interprocess communication and synchronization. > [.................] > ----------------------- > > Basically, the salient aspects of operating systems that I can pull > out of this is that an OS is a collection of programs that: > 1. Manages resources at the hardware level. > 2. Provides access to I/O devices and I/O time. > 3. Allocates primary and secondary storage. > 4. Provides a file system and command language. > (whether keyboard or mouse based). > 5. Schedules and protects processes from one > another. > > With regards to this definition, is Windows an OS? No. Is DOS > an OS? Probably not, unless we take into account the fact that > none save one of these books dealt with the concept of a single > user system, or even regards the PC as a computer (if it existed > at the time of writing). The one that does deal with the PC > sets the "disc operating system" as a classification unto itself. > If you take the simplistic view that Dos schedules processes one > at a time, and hence doesn't need to protect processes from one > another, then you satisfy criteria #6, and DOS is an OS. Few > of these books would view DOS as an OS, however, because most > deem multitasking as a basic element of operating systems. > "The most important characteristics include Concurrency...", > Coffman and Denning, p. 2 > "2.2 OPERATING SYSTEM CHARACTERISTICS > (1) Concurrency" > Lister, p. 9 > "Section 3.1 THE NEED FOR CONCURRENCY" > Barron, p. 29 > > According to the "experts", Windows and > DOS are an operating system together (as would be Desqview and > DOS, or Geoworks and DOS), but neither apart is an operating > system. Everyone happy? No, by your qualifications DOS+Windows != OS. Windows does not do #5 if you take scheduling to be based on #2 in the first list. ANY process can block ALL other processes under Windows if it wants to (I know this as a developer of windows apps). JWS ------------------------------ From: muts@compi.hobby.nl (Peter Mutsaers) Subject: Re: Linux and Corporate America Date: Thu, 2 Sep 1993 05:53:27 GMT >> On Tue, 31 Aug 1993 16:27:06 GMT, teffta@cs690-3.erie.ge.com >> (Andrew R. Tefft) said: ART> Look at it this way: since Linus (or anyone else) is not making ART> money off of Linux, what's he got to lose from someone ART> plagarizing it? Prestige? (ask him if he wants it.) The time ART> he has spent on it? Not really lost, since presumably Linux ART> would not die completely just from someone plagarizing it. In Don't you have any feeling for justice? It's just not fair that someelse makes money with something they did not make but just took from someone else. I couldn't bear the thought at least. -- _______________________________________________________________ Peter Mutsaers, Bunnik (Ut), the Netherlands. ------------------------------ From: jmadison@etsun.tech.iupui.edu (Gold Silver Soup and Silk) Subject: Re: Low Cost SLS 1.03 on Diskette Date: Thu, 2 Sep 1993 21:18:56 GMT In article <263p41$4i9@samba.oit.unc.edu> mjohnsto@ditdah.morse.net (Michael R. Johnston) writes: >[The question should be raised: Should c.o.l.a be used for "advertisements" >of Linux distributions (strictly distrubutions, not releases). Obviously, >many of these folks are making money off of software which they don't >write or even support. Or even package! Downloading SLS and putting it on >diskette hardly qualifies as "development". While I understand that this is >covered in the GPL, I don't have to allow it on c.o.l.announce. What do people >think? Good? Bad? I don't mind covering announcements of new SLS releases: >i.e., a distribution which Peter et. al. have actually package and distribute, >but when someone is selling someone else's distribution, what should be >done? Please comment. --mdw] (i didn't know where else to send this...) i don't like the idea, becuase SLS already does it's own distribution. I don't think it's fair that someone else should make money off of someone else's distribution...that's pretty cheesy IMHO. Especially if he's making a profit on it. but on the other hand, they _did_ take the time to put it on diskette for you; and people would like to have the cheapest way to get something w/out doing it themselves sometimes, so, i guess, although i think it's cheesy, you should go ahead & let them. (did that make sense?) -jonM -- jmadison@etsun.tech.iupui.edu <>< DJ AllStar of tha foolz(?)...comin' soon (we hope!) get Linux OS, it's dope! it's free! it's UNIX! ------------------------------ From: ssprunk@nyx.cs.du.edu (Stephen Sprunk) Subject: Re: Stacker-like Compression? Date: Fri, 3 Sep 93 03:08:24 GMT I realize that someone is going to have a field day flaming me for this idea, but... How about creating a device (ie /dev/chda1) that is compressed instead of a file system? Why bother with inodes and blocks and sectors? Let the user decide what file system to lay on the compressed drive and the file system has no clue. If the compressed drive isn't achieving the expected ratio (set of course by a /bin/ctune or similar), then it just maps part of the "drive" that the fs sees as bad. If the ratio gets better, then the chd device marks it good again. This assumes that the standard fs's can accept this kind of erratic "drive" behavior. I'm also assuming that the fs's aren't trying to optimize head movement across the platter (some wired fs's on other platforms do). SUMMARY: Compressed "drive" device (ie /dev/chda1) to replace regular drive device (ie /dev/hda1) and allow fs's on top of it. Blocks marked good/bad to fit expected compression ratio. Stephen -- Stephen Sprunk PGP 2.2 Public Key by finger/keyserver Printed on 100% recycled bits. Minimum 35% post-consumer waste. ------------------------------ Crossposted-To: comp.os.linux.help From: c9108932@peach.newcastle.edu.au (Simon J Ferrett) Subject: Re: help with SIGTSTP (more) Date: Fri, 3 Sep 1993 02:54:09 GMT jmorriso@rflab.ee.ubc.ca (John Paul Morrison) writes: >Can someone explain the correct way to handle SIGTSTP in Linux? >I have run into a number of programs which do not handle it >correctly: > talk > gopher >When I type ^Z, both programs go berserk; they go into a loop >and the continuously update the screen. I dont know why talk hasnt >been fixed by someone else already; gopher hasnt (as far as I know) >been ported to Linux, so I can understand why I've run into this. >another program which doesnt seem to handle ^Z properly is elvis; >it doesnt get into a loop, but vi is sure in a weird state when you >type fg. a friewnd of mine and I modified ytalk so that it would compile properly under linux, and it too suffers the same symptoms... any ideas/thingoes would be appreciated... Simon. -- c9108932@cs.newcastle.edu.au - Simon Ferrett Due to technical difficuties, we are unable to bring you your regularly scheduled .signature - normal transmission will resume as soon as possible... ------------------------------ From: pprice@panix.com (Pauline Price) Subject: Re: Linux Systems Laboratories ??? Date: 2 Sep 1993 23:58:27 -0400 In <1993Sep2.062924.3296@surfcty.surfcty.com> chris@surfcty.surfcty.com (Chris D. Johnston) writes: >Well, they have an 800 number. I've called it twice asking for >information. I am very interested in their $150 motif. >I have connected to a very nice answering service person. >No return calls, no postcards or literature as requested. Their motif is a resell of Micro-Link's Motif for linux. They send 4 1.44 3.5" diskettes, two with the run-time stuff, and two with the dev- elopment stuff. I guess it is available on 5.25" too. With that you get a few pages of installation guide. Other than that you are on your own! The ML package includes the OReilley and Assoc. Motif book for their price. Keep that in mind. I also already had the book, so I didn't need it. >I figure the owner has a full time job and this LSL is his >part time hobby... OR, he's so swamped with requests, he >really needs to hire some help. I'm sure he's getting one >heck of a phone bill publishing an 800 number that noone >follows up on. I'm sorry to hear that your calls were not returned. My many calls have been returned, and I also got Dan on the phone, rather than the service a number of times. I don't know what he may do on a more full time basis. But he has been generally available and reliable for my need, except, as previously mentioned with the back version of the kernal in the SLS dist. But his response was to offer me the next version when he gets it made up at no additional cost. (I will return the first set of disks, when I get the new ones.) --Pauli ------------------------------ From: steveo@beyond.demon.co.uk (steveo) Subject: Re: Stacker-like Compression? 16 Responses Date: Wed, 1 Sep 1993 19:19:51 +0000 I won't bother quoting, by now the original question has been repeated too many times I think :-) However a suggestion, instead of trying to make a compressed filesystem why not a compressed block device, divide each block into quarters, and allocate 1-4 quarters to a block on write out depending on how good the compression turns out. Of course it would need a logical to physical translation table somewhere. This approach should support ANY filesystem that can be mounted under Linux. Am I mad, or is this sort of thing the easiest answer ?? ------------------------------ From: steveo@beyond.demon.co.uk (steveo) Subject: Re: Wordprocessor under X Date: Wed, 1 Sep 1993 19:25:40 +0000 Edward v/d Jagt (cspas152@sus.edu.eur.nl) wrote: : Does anyone have or know about a Wordprocessor for X (under Linux) with : capabilities like WordPerfect has (something which has graphical and layout : capabilities). This raises a question in my mind, has anyone ported the Andrew Toolkit from X11R5 contrib. It is supposed to provide a truly wonderful wordprocessing environment, but I have never been able to compile it not enuff spare disc :-( ------------------------------ From: philb@cats.ucsc.edu (Philip Brown) Subject: Re: Checker, a gnu Purify subset Date: 3 Sep 1993 07:36:35 GMT In imp@boulder.parcplace.com (Warner Losh) writes: [check for uninitialized data] >then it will be purify like. Otherwise, it is just a very good >malloc debugger. Heck, gcc should do that. -- -------------------------------------------------------------------- "I want World Peace" "We're the CIA. We don't DO that sort of thing" philb@cats.ucsc.edu philb@soda.berkeley.edu ------------------------------ From: philb@cats.ucsc.edu (Philip Brown) Subject: Re: Wordprocessor under X Date: 3 Sep 1993 07:40:01 GMT In <1993Sep1.192540.609@beyond.demon.co.uk> steveo@beyond.demon.co.uk (steveo) writes: > This raises a question in my mind, has anyone ported the Andrew Toolkit >from X11R5 contrib. It is supposed to provide a truly wonderful wordprocessing >environment, but I have never been able to compile it not enuff spare disc :-( Well, then the operative word is just "compiled", not ported, eh? I learned some tricks form the andrew newsgroup. If you have "almost" enough space, you can go for a compile by just compiling some key things. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * [included article from somwehers...] Someting like an EZLite[An Andrew ToolKit view (a footnote) was included here, but could not be displayed.] is in the works. On the build front, you can shrink what's installed by #undef'ing these tokens in config/site.h: #undef AMS_ENV #undef METAMAIL_ENV #undef CONTRIB_ENV #undef MK_HELP #undef MK_BASIC_INSETS #undef MK_BLD_BLKS #undef MK_TEXT_EXT ... and rebuilding into a clean destination directory. If you do this, let me know the final size. So, you bring over all 60 Meg of sources but you really only want a text editor. What's a guy to do? After running imake at the top-level of the sources, make all the Makefiles and remove everthing under the directories that you don't need: % imake ... % make Makefiles % foreach trashit (ams atkams contrib atk/help atk/ness atk/table atk/console atk/apt atk/bush atk/value atk/preview atk/org atk/eq atk/chart atk/calc atk/toez) > echo ... trashing $trashit > rm -rf $trashit/* > echo depend world World Install dependInstall:: > $trashit/Makefile > cp $trashit/Makefile $trashit/Imakefile > end ... then resume the build process. -Gary Keim Andrew Consortium -- -------------------------------------------------------------------- "I want World Peace" "We're the CIA. We don't DO that sort of thing" philb@cats.ucsc.edu philb@soda.berkeley.edu ------------------------------ ** 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-Misc-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.misc) via: Internet: Linux-Misc@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-Misc Digest ******************************