Sun at least has an xserver that supports overlay visuals. Show me that one linux/BSD/whatever (assuming you are onw of those).
Lets assume that you had to use software that required a pseudocolor visual (thats 8 bit color, and there are lots of such commercial apps) and you run native in 24 bit color. What do you do?
Log the fuck out, configure the damn Xserver for pseudo, and restart. What a waste.
On sun what do you do? Pop up the damn window, no wait. Granted sgi does overlays, but sgi is dead already they just dont know it yet.
... where scientists publish for quotes of their publications...
If your view of the motivation of scientific publication is solely as quote whoring, then you have been exposed to the seedy underbelly of science. I will concede that in academia, where most of scientific journal articles originate, publishing is the measure of success. But I do believe, and not naively, that much of scientific publication is motivated by something as quaint as the "joy" of the work and the desire to make it available to peers.
I think this is a great idea. There are several journals in my field (geophysics) that all provide web search engines for journal papers. I have long wanted search results to point to the pdfs of the articles.
I think to be fair, the six month wait is a minimum. I would be probably be OK with one year during which the article is available only to subscribers, but after that year, the article becomes public domain.
Did you actually ever use the machine for something?
...it was a nice little package... had nice compilers, and was a pleasure to use...
Back in those days, you pretty much were stuck with xmkmf (imake) to compile anything interesting. I remember hacking macII.cf if you do not. Of course, if you were particularly tough, you could hack the individual makefiles (Ugh!), but I wouldnt call that a pleasure to use. I won't even start on the A/UX implementation of X.
"Nice compilers?" WAYSACIHS !! (What are you smoking and can I have some).
... ran existing Max apps very well...
Sure, with N Mb of RAM, where N is a large and prohibitively expensive number. Maybe in the first boot mode, *before* launching A/UX. Afterwards you had the sluggish random crash mode.
Maybe you, unlike many others I have talked with, did not experience the look and feel of A/UX: random and unprovocated hangs and much (all but unbearably lengthy) rebooting -- especially if you attempted to do exactly what was touted as the biggest advantage -- mesh use of the unix side of A/UX with the more traditional "Mac apps".
Re: What's wrong with this title?
on
OS X
·
· Score: 1
Ah you youngsters...
A/UX was a painful abortion that should never have been.
The IIfx was pretty much like the IIci only was almost twice as fast (40 Mhz instead of 25-30 for the IIci). I had a IIci because I couldnt scrape up the extra scrip for the IIfx.
Just to clarify, A/UX was an "Operating System" and IIfx was a "computer". For the most part, you can run several several different "Operating Systems" on a single "computer". The IIfx did NOT ship with A/UX by default -- you had to get A/UX and install yourself (at least I did, maybe there were other ways).
Re:More accurately, the reincarnaton of A/UX?
on
OS X
·
· Score: 1
Are you out of your mind?
A/UX sucked so bad I can still smell it 10 years later. I used it on a IIci back in 90-91 finishing up a masters thesis. I have mercifully forgotten most of that experience.
More reboots required even than NT 3.5.
The ars article (link) discusses boot time a little -- claims that it is not so important. But if you have EVER waited for the mac OS to boot so you could boot the ridiculously slow A/UX kernel than you might agree that boot time is important. I can still remember that slowly creeping status bar from hell. I shudder to think...
Not only that, but A/UX was a very much frontal lobotmized unix. Of course that was back in the days before autoconf, so extensive hacking was required to get A/UX to compile *anything*. OS/X can at least handle a symbolic link.
On the other hand, if we let our memory get a little fuzzy we can say that it was a good thing, and maybe OS/X will take where it needed to go.
Although this is way OT, I had a similar experience. I finished my thesis in 1998 in Engineering Geoscience and did not dig in enough to learn Latex (a mistake I am now fixing). I had 100 pages without figures, and all kinds of stringent demands regarding layouts of table of contents, lists of figures, running heads, etc etc etc ad spewnitum. And two appendices of formulae.
After proofs by my advisors I tried to make the few "small" changes to the text suggested and watched the whole document self-destruct. I had returned to the campus from another state just to submit the thesis to the graduate department, and I spent the next 10 hours feverishly trying to get Word to cooperate for a final printing before my flight. Unfortunately I had a single long text document - but the lovely tools in Word are designed to automate all those tasks, right?
MicroSloth's rules for auto-formatting were devised by some netherworld slug with horns and a pitchfork.
I still dont have the latex proficiency to run off journal papers but I am working past the odd memo with equations. As far as references on latex, get the "dog book" -- the rescue st bernard with the whiskey onboard in the mini-keg. (The Latex Companion, goossens et al).
Word is actually great for something less than a couple pages without inline figures.
Anyway, my condolences and regards,
Katz does not "Features" post
on
Me-Commerce
·
· Score: 2
Every time I see "Features" I expect another Katz post, and not necessarily anything properly called a feature.
Honestly, I have nothing personally against Katz, but it does bother me to see "Features" tagged for 90% of his stories.
"My guess is it would be way more than 1750 lines per bug..."
Im just guessing here from context, but I think you mean "bugs per line".
Though I suspect from long experience with unixes and routine installation/use/reboot of NT, that the way you have it now, more lines of code per bug, would be accurate for just about any flavor of UNIX (barring A/UX or AIX, ofcourse).
"... Utilities that are far and away better than their counterparts on other UNIXes..."
What? Ever benchmarked gcc against a commercial compiler? granted the GNU compilers are nice and most free software compiles out-of-box with gcc, but show me the auto-par directives, support for the SGI or CRAY style paralellization, or even performance tuned/vectorized libs like BLAS or LAPACK.
When compiling games and window managers GNU stuff works great... but when you have to go to work you need beefier tools.
"I honestly don't know why people buy C/C++ compilers anymore, except to maybe get the snazzy IDEs that come with them".
Ill tell you why I bought workshop 4.0. Not for IDE (I use vi), but for optimization. Gcc cannot come even remotely close to optimization levels that SUN's compiler can. What about CRAY or SGI style parallelizing (sp?) directives and/or cmdline opts?
BTW, its the same story with gcc vs. pgcc, they (portland group) can concentrate more on the wicked optimization than on the portability- and I mean wicked. Do a little benchmark of your favorite app and you will quickly realize that even though the free gcc is nice and portable that it is worth the money for production apps to use a faster (read better optimizing) compiler. Incidentally, pgcc beats the pants off suns cc.
Hardly. We have a dozen or so x86 and a couple ultra 60s.
The problem --> Little/Big endian. No bin compat.
Which sucks real bad if you do any binary IO (e.g. SEGY).
As for the server pack, for the ultras its worth any extra money cause you can mirror disks, but (I think) this is NA for the x86. So cost for x86 is much less than $1150. Certainly it aint free.
Maybe off topic and this is posted a bit down below but can I get the Xfree position?
One of my biggest gripes about x86 (whatever distro) is that you can't get overlay visuals as far as I know. What I mean is 24 bit root and simultaneously open up a pseudocolor window. Why is because I use 24 bit root but some of the software we use requires psuedocolor display (LandMark -- Ugh!).
I emailed xfree86 a while ago and was told that the problem is the cards, and that maybe the high-end cards (eg matrox) might soon be capable.
BTW, I agree 100% about the 1920x1200. The aspect ratio screwed up OGL (for me). I traded my wide 24" for the standard big monitor and am happy at 1280x1024 and the higher refresh and standard aspect.
Though crazy me I also traded the silly "Elite3D" for the regular creator3D cause it doesnt waste four UPA slots like the older turbo ZX (two for fans, believe it or not).
Anyway- once Xfree86 solves the BIG PROBLEM of no overlay graphics I will rethink the whole issue. As far as I know (with linux, BSD, x86 solaris, etc. etc. youre-favorite-distro-here) you cant have a 24 bit root and pop up a pseudo on top. How shitty is that? The limitiations are mostly in the hardware I guess (some cards like high end matroxs might be able to overlay, as I hear from the xfree86 techs).
There seems to be a lot of rug beating going on here. People who use single or even dual cpu x86 machines but have never sat down in front of an ultra may not be aware of some of the issues re linux. When I first came to work my company was dominated by x86 solaris. I brought linux and bought the portland group compilers (for a fraction of suns workshop fees) and blew away the solaris x86 machines. Boot in a small fraction of the time, comile quicker, better uptime, better memory handling, etc etc and best of all customize, customize, customize. BTW CDE sucks. I tried hard to get them to switch to linux but no go. Too much legacy code linked against libs you cant get for linux (binary only dists from INT- Interactive Network Tech. for example). So they still use the x86 slowlaris. However! I am sitting in front of a twin ultra60 with 400s and a Gig of ram. What makes it really sweet is the creator3D graphics (and the dual heads). Find a linux driver for that, or show me double-buffered hardware accelerated 3D graphics on linux. I use geomview and do a lot of 3D work. On linux it is merely passable, because you must use software rendering. But on this ultra, baby, it simply flies. Until linux gets caught up at that end (and I dont want bleeding edge solutions, this is a production environment), I am quite happy with sparc solaris. just my 2 c.
Slashdot Mirror
Lets assume that you had to use software that required a pseudocolor visual (thats 8 bit color, and there are lots of such commercial apps) and you run native in 24 bit color. What do you do?
Log the fuck out, configure the damn Xserver for pseudo, and restart. What a waste.
On sun what do you do? Pop up the damn window, no wait. Granted sgi does overlays, but sgi is dead already they just dont know it yet.
If your view of the motivation of scientific publication is solely as quote whoring, then you have been exposed to the seedy underbelly of science. I will concede that in academia, where most of scientific journal articles originate, publishing is the measure of success. But I do believe, and not naively, that much of scientific publication is motivated by something as quaint as the "joy" of the work and the desire to make it available to peers.
I think to be fair, the six month wait is a minimum. I would be probably be OK with one year during which the article is available only to subscribers, but after that year, the article becomes public domain.
Did you actually ever use the machine for something?
Back in those days, you pretty much were stuck with xmkmf (imake) to compile anything interesting. I remember hacking macII.cf if you do not. Of course, if you were particularly tough, you could hack the individual makefiles (Ugh!), but I wouldnt call that a pleasure to use. I won't even start on the A/UX implementation of X.
"Nice compilers?" WAYSACIHS !! (What are you smoking and can I have some).
Sure, with N Mb of RAM, where N is a large and prohibitively expensive number. Maybe in the first boot mode, *before* launching A/UX. Afterwards you had the sluggish random crash mode.
Maybe you, unlike many others I have talked with, did not experience the look and feel of A/UX: random and unprovocated hangs and much (all but unbearably lengthy) rebooting -- especially if you attempted to do exactly what was touted as the biggest advantage -- mesh use of the unix side of A/UX with the more traditional "Mac apps".
A/UX was a painful abortion that should never have been.
The IIfx was pretty much like the IIci only was almost twice as fast (40 Mhz instead of 25-30 for the IIci). I had a IIci because I couldnt scrape up the extra scrip for the IIfx.
Just to clarify, A/UX was an "Operating System" and IIfx was a "computer". For the most part, you can run several several different "Operating Systems" on a single "computer". The IIfx did NOT ship with A/UX by default -- you had to get A/UX and install yourself (at least I did, maybe there were other ways).
A/UX sucked so bad I can still smell it 10 years later. I used it on a IIci back in 90-91 finishing up a masters thesis. I have mercifully forgotten most of that experience.
More reboots required even than NT 3.5.
The ars article (link) discusses boot time a little -- claims that it is not so important. But if you have EVER waited for the mac OS to boot so you could boot the ridiculously slow A/UX kernel than you might agree that boot time is important. I can still remember that slowly creeping status bar from hell. I shudder to think ...
Not only that, but A/UX was a very much frontal lobotmized unix. Of course that was back in the days before autoconf, so extensive hacking was required to get A/UX to compile *anything*. OS/X can at least handle a symbolic link.
On the other hand, if we let our memory get a little fuzzy we can say that it was a good thing, and maybe OS/X will take where it needed to go.
My Condolences.
Although this is way OT, I had a similar experience. I finished my thesis in 1998 in Engineering Geoscience and did not dig in enough to learn Latex (a mistake I am now fixing). I had 100 pages without figures, and all kinds of stringent demands regarding layouts of table of contents, lists of figures, running heads, etc etc etc ad spewnitum. And two appendices of formulae.
After proofs by my advisors I tried to make the few "small" changes to the text suggested and watched the whole document self-destruct. I had returned to the campus from another state just to submit the thesis to the graduate department, and I spent the next 10 hours feverishly trying to get Word to cooperate for a final printing before my flight. Unfortunately I had a single long text document - but the lovely tools in Word are designed to automate all those tasks, right?
MicroSloth's rules for auto-formatting were devised by some netherworld slug with horns and a pitchfork.
I still dont have the latex proficiency to run off journal papers but I am working past the odd memo with equations. As far as references on latex, get the "dog book" -- the rescue st bernard with the whiskey onboard in the mini-keg. (The Latex Companion, goossens et al).
Word is actually great for something less than a couple pages without inline figures.
Anyway, my condolences and regards,
Honestly, I have nothing personally against Katz, but it does bother me to see "Features" tagged for 90% of his stories.
Is this a gimmick?
Im just guessing here from context, but I think you mean "bugs per line".
Though I suspect from long experience with unixes and routine installation/use/reboot of NT, that the way you have it now, more lines of code per bug, would be accurate for just about any flavor of UNIX (barring A/UX or AIX, ofcourse).
You now have my respect. My attention is another matter ...
I dont hate everything by Katz just because it is by Katz, but I hate that everything by Katz is a "Feature".
FWIW.
Why in heck did you create an OS that you have to REBOOT in order to change the IP?
Why do you need to REBOOT to change the hostname?
Why in god's name must you REBOOT five gazillion times to install NT?
Do you expect to get out of the bathroom soon?
Multi-headed display can't keep the slashdot effect at bay.
This is standard on "high-end" workstations. Any feel for when we'll get it on x86 hardware?
I know this probably gets kicked up the tree to X-level rather than window manager level coding, but you got your hands in that, eh?
"You need to give them a problem or a set of objectives, provide them with a large amount of hardware, and then ask them to solve the problem."
If only everyone thought like this! Fortunately the people I work for don't micromanage, and they just pumped in 35K+ for some really nice hardware.
The real important freedom is the ability to use whichever software you like. In general the free stuff is the best, but not always.
"... Utilities that are far and away better than their counterparts on other UNIXes ..."
... but when you have to go to work you need beefier tools.
What? Ever benchmarked gcc against a commercial compiler? granted the GNU compilers are nice and most free software compiles out-of-box with gcc, but show me the auto-par directives, support for the SGI or CRAY style paralellization, or even performance tuned/vectorized libs like BLAS or LAPACK.
When compiling games and window managers GNU stuff works great
just my $0.02
"I honestly don't know why people buy C/C++ compilers anymore, except to maybe get the snazzy IDEs that come with them".
Ill tell you why I bought workshop 4.0. Not for IDE (I use vi), but for optimization. Gcc cannot come even remotely close to optimization levels that SUN's compiler can. What about CRAY or SGI style parallelizing (sp?) directives and/or cmdline opts?
BTW, its the same story with gcc vs. pgcc, they (portland group) can concentrate more on the wicked optimization than on the portability- and I mean wicked. Do a little benchmark of your favorite app and you will quickly realize that even though the free gcc is nice and portable that it is worth the money for production apps to use a faster (read better optimizing) compiler. Incidentally, pgcc beats the pants off suns cc.
I do agree that non-standard cmdline sucks.
Homogenous operating environment?
Hardly. We have a dozen or so x86 and a couple ultra 60s.
The problem --> Little/Big endian. No bin compat.
Which sucks real bad if you do any binary IO (e.g. SEGY).
As for the server pack, for the ultras its worth any extra money cause you can mirror disks, but (I think) this is NA for the x86. So cost for x86 is much less than $1150. Certainly it aint free.
No wonder we have this shitty isdn and flaky ISP instead of frac T.
...
Shouldnt have left school. Although we do have cat5 100-T for our internal net which really is 10 times faster than the net at UC.
Or, move back to Frisco, get the juicy bandwidth and pay 10 times the rent
Maybe off topic and this is posted a bit down below but can I get the Xfree position?
One of my biggest gripes about x86 (whatever distro) is that you can't get overlay visuals as far as I know. What I mean is 24 bit root and simultaneously open up a pseudocolor window. Why is because I use 24 bit root but some of the software we use requires psuedocolor display (LandMark -- Ugh!).
I emailed xfree86 a while ago and was told that the problem is the cards, and that maybe the high-end cards (eg matrox) might soon be capable.
Any news on this?
Solaris DOES in fact offer 24 bit root depth: check this (running 2.6):
% version
Machine hardware: sun4u
OS version: 5.6
Processor type: sparc
Hardware: SUNW,Ultra-60
% xwininfo
xwininfo: Window id: 0x3a (the root window) (has no name)
Width: 1280
Height: 1024
Depth: 24
Visual Class: TrueColor
-geometry 1280x1024+0+0
BTW, I agree 100% about the 1920x1200. The aspect ratio screwed up OGL (for me). I traded my wide 24" for the standard big monitor and am happy at 1280x1024 and the higher refresh and standard aspect.
Though crazy me I also traded the silly "Elite3D" for the regular creator3D cause it doesnt waste four UPA slots like the older turbo ZX (two for fans, believe it or not).
Anyway- once Xfree86 solves the BIG PROBLEM of no overlay graphics I will rethink the whole issue. As far as I know (with linux, BSD, x86 solaris, etc. etc. youre-favorite-distro-here) you cant have a 24 bit root and pop up a pseudo on top. How shitty is that? The limitiations are mostly in the hardware I guess (some cards like high end matroxs might be able to overlay, as I hear from the xfree86 techs).
There seems to be a lot of rug beating going on here. People who use single or even dual cpu x86 machines but have never sat down in front of an ultra may not be aware of some of the issues re linux. When I first came to work my company was dominated by x86 solaris. I brought linux and bought the portland group compilers (for a fraction of suns workshop fees) and blew away the solaris x86 machines. Boot in a small fraction of the time, comile quicker, better uptime, better memory handling, etc etc and best of all customize, customize, customize. BTW CDE sucks. I tried hard to get them to switch to linux but no go. Too much legacy code linked against libs you cant get for linux (binary only dists from INT- Interactive Network Tech. for example). So they still use the x86 slowlaris. However! I am sitting in front of a twin ultra60 with 400s and a Gig of ram. What makes it really sweet is the creator3D graphics (and the dual heads). Find a linux driver for that, or show me double-buffered hardware accelerated 3D graphics on linux. I use geomview and do a lot of 3D work. On linux it is merely passable, because you must use software rendering. But on this ultra, baby, it simply flies. Until linux gets caught up at that end (and I dont want bleeding edge solutions, this is a production environment), I am quite happy with sparc solaris. just my 2 c.
Without the "click here to get plugin" crap? Or know the trick to circumvent that. thanx