>This is false. It is not possible to convert >automatically ("mechanically"? sweet God!) from >HTML to TeX in a reasonable way, since TeX >contains information that is entirely absent in >HTML. Ditto HTML to SGML. Ditto HTML to >PostScript and vice versa. Ditto plaintext to >much of anything.
Who said anything about gaining information in the conversion? You can make postscript out of plain ascii text. And when you print it, guess what, it looks like plain ascii. (WHY you'd want to do this is an open question.)
That "much more info" isn't what people read the documentation for, is it? Will multiple fonts help me beat "chat" into submission? (I still dial the modem with minicom and quit out to run the PPP daemon. That sucker's evil.)
The main problem with plain text is that the wordwrapping is fixed, and even then why is that a bad minimum acceptable format if the documentation writer has something interesting to say? If somebody has information I want, I want them to write it up and get it to me, not to be deterred by what format it has to be posted in. Converting formats is the archive's job, not the author's.
And the archive CAN convert. Mechanically if they just MUST have it in format X, and no that doesn't add markup information that the original format didn't have. But you don't have to be a subject matter expert to add markup. A maintainer with time on their hands can take a plain ascii email and re-paragraph it and add boldface and underlining and everyhing, without being an expert in the content. I don't have to know anything about the history of sweedish currency to add markup to a document on the subject, do I? And if I screw something up, I'm sure I'll hear about it in an open environment
Sheesh, with half of the documentation I've seen we're lucky if the author can express their thoughts in english. Rejecting their submission because they didn't chose a format that allows the broadest possible range of fonts to express themselves with just doesn't make sense.
If they pick a format like Word where we wind up LOSING markup information converting to a format we're willing to post... Well, that's the breaks from them using a non-recommended format. But being unwilling to accept submissions that don't have ENOUGH markup capability... Why?
This guy wanted to rattle of a quick and dirty article about today's hot topic (Linux) by phoning up the inventor and getting a personal quote.
Linus wouldn't take his call.
So he wrote an entire article about the fact that Linus wouldn't take his call, full of speculation and allegory and other forms of hamburger helper to get the word count up.
There's a certain amount of peverse ingenuity in doing that much to avoid actually having to do any kind of work. I can't figure out if I should be amused or disgusted. Let's see...
1) Did the fact Linus wouldn't take his call actually suprise him? There IS a pending feature freeze on the 2.3 series, he's kind of swamped right now in a BIG WAY.
2) Why is he writing a column this long about something he calls inevitable within the column?
3) Did it ever occur to him that Linus's employers at Transmeta might be behind it? It is a company phone after all, he theoretically does some kind of actual work for them. PR is nice but it's not ALL he does there.
4) He needs an appointment weeks in advance to see his dentist, but won't send the fax to get on Linus's calendar.
5) He scolded somebody for asking a question about patenting genomes, but never said what great new original questions HE was going to ask Linus.
If this was a post a mailing list, this guy would probably have just found a home in my twit-filter... Must have been one heck of a tight deadline to result in an article like that...
As far as the myriad formats go, it is possible for the maintainers to convert from one to another. This doesn't require any knowledge of the topic at hand, and with most formats it could probably be done mechanically. So this may not be as much of a problem as they're making it out to be.
As an end user, though, I want simple. Ascii is OK, and HTML provides miscelaneous improvements like boldface, bullet point lists, and the often underestimated ability to rewrap the sucker to a different window size without winding up with wasted space or scrollbars, or even printing it out intelligibly on the first try.
Just about anything else is probably getting too fancy. Postscript, PDF, or Tex may provide professional publishing capabilities, but will they provide legible text if I've booted from a rescue disk to a text console on somebody else's system and only have "cat" (or if I'm lucky, "less") to view the file with? I can ignore paragraph tags in that case, but I'd better have that postscript printed out somewhere...
I have one machine with onboard SIS 6326 AGP video took a week of fiddling to get X to run at all, and it still simply DOES NOT WORK. The graphics literally smear when you drag the window, the mouse pointer leaves droppings, all text in X is illegible... Runs fine in text mode, but the graphics just won't work right even with X config files other people swear work for them.
It did NOT detect the graphics card, and I've NEVER had it detect the monitor, and yes that's made a difference on several occasions. (Especially with another box that had an SiS 5595, another piece of junk embedded video chip. With Red Hat 6 it worked fine, but only after I found the right monitor from the "nameless Dell 14 inch" list. The wrong monitor options included "refuse to run because it can't find a mode" and "loud screeching noise from the back so I shut the monitor off even before I hit ctrl-alt-backspace because I'm afraid smoke's about to come out of it". Oh configuring X is just a JOY...
Also, the modem I pulled out of the drawer for one machine recently had a jumper setting that insisted it was Com 3 according to the little sticker, but it turned out to be Com 11. (Extra jumper to shift the IRQ to 16 bit mode that was not documented on the sticker. Fun.) No Linux install has EVER found a modem on a nonstandard setting automatically (although OS/2 never had any trouble, you'd think that the INSTALL could scan for IRQ's...) This particular joy took 2 days worth of free time (I.E. most of an hour sitting down and scratching head) to figure out. Eventually scrubbed through the man pages and figured out how to run "setserial auto_irq autoconfig", which is remarkably non-obvious when you can't remember what "setserial" is called... But only after a lot of red herrings in the "plug and play" section of the Bios telling it to reserve IRQ 3 for the ISA bus...
I've been banging my head on computers since I was 12, so I've known where to stick the hot poker to get the system to confess most of the time. But a newbie would have given up, and I've wasted a LOT of time setting up boxes.
I still haven't gotten Red Hat 6 to connect to my ISP, the ISP-HOWTO that ships with this version is a JOKE, sentences cut off in the middle and it tells you to make changes to files it doesn't name. It's been a year since I last configured networking, I don't remember this stuff... I remembered where to look it up and it wasn't there, I feel cheated.:P
(I went out and bought a "Linux for dummies quick reference" after that experience with the Howtos... May not have the info I need but at least someone will have read it through before shipping it.)
Oh, another fun little detail: getting to the power down message on shutdown is no guarantee that it won't fsck on the way back up. On any Red Hat 6 system I've seen so far. Dunno why...
You can tell I detoured through OS/2 for a couple of years, can't you... Their hardware detection positively spoiled me...
Still, can't complain. A friend wanted me to explain why Windows 95 was booting into "safe mode". That'll put things right into perspective, won't it? It turned out to be a missing font. Now that's just pure evil... (Strangely enough, Red Hat 6 installed without a hitch on that box, apart from guessing the monitor type again.:) Rob
Who's trying to "beat" Gates/microsoft? We're simply rendering them irrelevant. He may hold all the cards, but we're playing chess. Open Source isn't about people who sell software, it's about people who use software. When microsoft's toll booths and potholes get in our way, we go around them, like the internet we created. We have work to do: writing papers, simulating wind tunnels, or playing doom. Microsoft is not involved. Who cares about them? Despite Gates's ego and money, the computer industry is not about him and never was. It's about us, the programmers and users who do stuff with computers. If MS is no longer an asset to us, discard it and move on. Replace the broken tools with ones that work and get back to using them, not arguing about them. All this talk of Gates is a distraction at best. Rob
The anti-usability comments are interesting. I don't know half of what goes on under the hood of my car, and if it breaks I take it to somebody who does. I can't drive a stick shift. Once a little light came on and I had to stop and figure out how to put more water in it, and I was glad the stuff under the hood was clearly labeled, but I still pay somebody to change my oil for me. Computers are like cars, phones, and toasters. 99% of the people using them don't care what makes it go, they just want it to work with as little effort on their part as possible. If you force them to learn the details, they'll go elsewhere. Period. My car takes me to work/grocery store/movie theater, I don't tinker with it in the garage and paint flames down the side. I only think about it if it breaks. The hobbyists can create a market, and guide the market, but if we don't provide what the market wants somebody else WILL. And we'll wind up having to deal with it. The point of Linux is it's a better way of doing things. But if it's only a better way of doing things for techies and not for all the regular people who use computers like toasters (stick this in here and press that button and then it comes out finished) and hire techies for the same reason I hire auto mechanics, then we're doomed to having to support Windows or Macintoshes, and that's SERIOUSLY unpleasant. (And even the techies, deep down, want the computer to work like on Star Trek. "Computer, do this". It does it. End of story.) Rob
I'd like to point out that a Journaling filesystem is a high priority on NT for the same reason that parachutes and fire extinguishers were high priorities on the Hindenburg. If the primary cause of system crashes is a backhoe taking out the power, quickly recovering from crashes isn't a primary development priority. On a system that crashes in the presence of large flowers or brightly colored wallpaper, rapid crash recovery is a big time development priority. On Linux, JFS is primarily for bragging rights, isn't it? We're focused on not crashing in the first place. Rob
Now you're talking keeping geographically distributed servers synchronized, which doesn't sound like fun, especially if a server goes down (Not necessarily Linux, how about the connection to the site? Backhoe, power, coffee, etc.) For all I know it could involve rewriting half the site's infrastructure to do properly...
This isn't necessarily arguing against doing it at all. If the demand warrants it, certainly, but is it a good first step to contemplate? I can't shake the price/performance argument, even with andover throwing money at the problem... A beefier server with raid 5 sitting on a couple of T1's struck me as a better first step...
Rob
Re:How long can a company guard a product's featur
on
iMac Clone Gets Sued
·
· Score: 1
It's got the engine in the front! It costs over $20 grand! It's not a bug! It just LOOKS like one!
>Certainly not! I use Debian, and I don't want >some RedHat shit on my system.
If I remember correctly, this is why Bruce Perens left. The reason -I- left debian was the general "I don't care if anyone other than me uses it, although I'd prefer it if they didn't" attitude of the mailing list. I could only take about six months of that, then I went for the distribution that was most interested in bringing open source to the largest number of people. Turned out to be Red Hat.
>Plus RPM is a proprietary binary package that
How can you have a proprietary GPL'd toolset?
>can't be extracted with standard tools, so when >you want to read the README on a >Sun/Alpha/Windows machine you're completly dead.
Only by your definition of "standard tools". Considering you have to download gzip for most platforms other than Linux anyway, where's the problem? (Other unixen come with "compress", remember?) On linux (including debian), a tool to use the suckers is part of the base package.
And why isn't the readme on the web page you download it from?
>The Debian package format has the advantage to be >an 'ar' archive (+ tar.gz inside)
It has the disadvantage of nobody using it.
Nobody replaces an established standard without a darn good reason, and even then it's an uphill battle. That's the whole reason micros~1 has survived this far. GIF is patented 8 bit with no transparency and clearly superior PNG -STILL- is having an uphill battle.
There an art to knowing what can and cannot work, and one big trick is not to confuse the means with the ends (such as Stallman telling people not to call the system "Linux").
Don't stand at the bottom of hoover dam and push. It won't work. Go around, get dynamite, do something useful. But standing there and pushing is just sad.
>Proposing RPM to be the standard on Linux >machines is the worse idea I've ever seen
You should get out more. Try visiting Washington DC, it's an eye-opener.
>Haven't you ever downloaded the tools from >prep.ai? Didn't you ever compile GCC, or egcs, >from scratch to optimize the compiler and make it >run faster? Haven't played with GCC since college and I get the tool sources off of CD usually, but I am in the process of installing a system from scratch starting with kernel source code and the "bootdisk howto". Pretty straightforward going so far, although getting X to recognize an onboard SIS graphics card is a bit of a pain, I'm to the point where everything works except the pixels on the screen are out of order, I might muck around in its source to see if I can fix that...
Might make an "Install Linux yourself from source code with tweezers and a magnifying glass HOWTO" when I'm done... Right now it's just so I'll know what it all does.
>Yes, this is meant as a kind of a troll, but >think about it. What developer with any self >respect would complain about a non-RPM'd >developer tool? Anyone, such as me, who doesn't think solely and myopically of their own needs.
>I can't believe I'm hearing a Linux >"developer" complain about not knowing how to >handle a tar file! Of course I know how to handle a tar file. But this is precisely the point. The developer community is extremely self-centered, that's why the system's been around for 8 years and we're only JUST starting to write a desktop for the thing.
Think of someone LEARNING to program on linux. Everything had darn well better come pre-installed or they're screwed. And there's more to install than unzipping, there's uninstalling, inventory/versioning, and fun configuration things like enabling desktop icons and making them go away again during uninstall....
Trying to get X working on my graphics card I've installed and uninstalled the server at least three times. (Uninstall the old X server, install X-SIS from suse, uninstall that and try the new 3.3.3.1 SVGA server which has exactly the same pixel ordering problem X-SIS did. The mouse leaves droppings, dragging windows smudges the contents, text is illegible, it's sad...)
It would have been a real pain to figure out which files were XBASE and which were specific to the card without a built-in uninstall. (Cleaning up the sumlinks would have been nice too, but that's at least partly a Red Hat specific problem...)
Getting places like IBM -USED- to providing uniform install packages is important. Like it or not, Red Hat's Package manager is the standard we've got. It is the most widely used install program, and I believe every single distribution except Debian and Slackware now uses it exclusively. And both Debian and slackware provide it as an option.
>The artist analogy made earlier really works if >you look at it from this perspective. I remember >one project where I did put in the heavy hours, >not because of a deadline or any management >induced requirement, but because I needed to do >it. Afterward, several coworkers and superiors >expressed appreciation, but there was no huge >payoff/bonus/promotion in it for me. And that's >OK, because now the code is there, and its still >in production use.
I did the exact same thing for a large nameless company (ok, IBM). I inherited a piece of junk that was technically complete, but not working, and spent evenings and weekends virtually rewriting it.
I didn't even get a "thank you" from management, just "too big a change, we can't risk it" until the test department came back and said "profoundly unacceptable in all categories" and the boss saved face with "wait, we have a fix" and took credit for my six months work as a band-aid fix. (And gave me a bad performance review for staying in my office coding instead of attending his truly pointless daily meetings.)
The reward? The code shipped, as part of a larger overall product with a declining market share (OS/2). Three years later, it's an archaeological relic.
This is another big push for open source, freeing code from context. So the program itself is no longer useful, there are a half-dozen valuable subsystems in that thing that could easily have been salvaged and re-used. I wrote an object oriented GUI management system they wanted to patent, my own (fairly optimized) grep implementation, a configuration file parsing mechanism I was outright proud of, and even some new (generally applicable) build tools.
All of the above is under lock and key somewhere, collecting dust. It's closed source, the program is no longer being developed, that's the end of it.
I made my own happy ending to the story: I left the company. Programmers are a lot like cats: as long as we're happy you can't get rid of us, make us unhappy you can't keep us.
The problem isn't a shortage of warm bodies with university-imparted C++/Java/HTML on their resume. The problem is a shortage of good, experienced programmers who produce good code and enjoy doing it.
If you're doing this for the money, YOU WILL BURN OUT. Fast. People who get CS degrees with dollar signs in their eyes don't last long. They're also generally really bad programmers (see point 1).
THIS is why free/open source is winning. The ones who care about doing it right so bad it hurts are the ones who start and run all these projects because it has to be done RIGHT, darn it! The hordes of wannabes and apathetic check cashers send us complaints about it not working right, and if we're lucky they're good enough to spot the bug. Ocasionally, they pick up enough as they go along to become real hackers.
You can't hire that, because money isn't what motivates it. "I can do better" ego and a raging sence of injustice that some idiot was getting paid for doing it wrong in the first place started Richard Stallman with the GNU project. Yearning for a decent intellectual challenge (a specialized kind of boredom), and that nasty urge to take the refrigerator apart and see how it works prodded Torvalds to complete it with the Linux kernel. Apache was just a bunch of people tinkering with a defective tool they used regularly and a "wow, cool, can I borrow that" attitude, who were lazy enough to want to make things easier on themselves by sharing the work.
Then pride, fan mail, and a general sense of accomplishment kick in and look out, 200+% annual growth.
And if we can all make money off of this kind of thing, even do it as a day job, well that's a bonus.
Rob
If you've ever had to say "no, not like that" to somebody, you understand open source.
1) You need a copy of windows to run. To do it legally costs $$$, especially NT.
2) Running a whole second OS is a serious resource hog.
3) It's effectively running on a second (virtual) computer, in its own little sealed box. Why not just get a second computer and a monitor/keyboard/rat switch?
Wine provides the Win32 system calls to a Linux process, allowing things like a windows CGI program to do credit card validation to be spawned from Linux' Apache. It may never run every windows program in existence, but:
1) Neither does any one version of Windows.
2) I don't own every windows program in existence. I only care about the ones I have (which these days, are mostly games, half of which actually run under DOS.)
3) This is legacy support. 50% of the legacy windows programs out there aren't Y2K compliant anyway, and an amazing number of people are limping along with "good enough for now" 3.1 installs left over from the 1980's for their daily word processing and checkbook balancing/payroll. (Sheesh, last year I helped a friend of a friend copy his comic book store inventory system from an old 386 SX with a 100 meg hard drive to an old 386 DX with a 200 meg drive. Only reason he left the old system was he'd tried Dos 6 doublespace and the drive started to eat itself.)
We don't HAVE to support the latest and greatest Windows apps, those companies are still around and we can lobby for a native version as we penetrate farther and farther into "grandma" land and our usage numbers go up with drool-proof interfaces like Gnome and automatic install/configuration and pre-installs. And we ALREADY support a lot of the old stuff, and creep farther every day.
The Wine people are adding new APIs faster than Microsoft is. They're better at it. Someday, they'll catch up.
IBM's selection ofthe 8088 was a little more complicated than that:
Motorola couldn't guarantee the volumes of chips IBM wanted with the newer 16 bit technology. Intel's 8088 using 8 bit chips pretending to be 16 bit (yup, the 386 SX was not the first) meant they could use older more proven fabrication techniques and existing (dirt-cheap) 8 bit support chips.
I was under the impression that the 8086 existed FIRST, and the 8088 was a hack to get the volume and cheapness IBM wanted. I guess we'd have to ask an Intel veteran. Rob
Re:How long can a company guard a product's featur
on
iMac Clone Gets Sued
·
· Score: 1
Define "deviating from the standard beige case". IBM made some remarkably stylish solid black computers a year or two back. I still have the keyboard from one somewhere. Okay it wasn't bright green with a translucent case, but it certainly looked unique, and it was pre-imac. The original model T was produced in "any color the customer wants, so long as it's black". Later, they diversified into bright red sports cars and psychadelic VW bugs (or the new "Volkswagen Feature"; there's no such thing as a fuel injected bug). Look-and-feel differentiation is normal for an industry. It didn't occur to most of US for the first decade or two because we're not exactly normal, are we? Rob
>A jock gets a scholarship because they ACCOMPLISH >something. I am not saying I find it all that >meaningful, but they DID SOMETHING. ie. practiced >enough to get good at a sport. >Giving a person a scholarship due to their gender >is hardly the same. The person in question did >not ACCOMPLISH anything, other than being born >that gender. Ok, so give out a computer science scholarship to beauty pageant winners. That's no more arbitrary than giving some a college scholarship for playing football, and you'd be amazed how much work goes into winning a beauty pageant.
And it's less arbitrary than "hot chick". It's an empirically "hot chick". Much more legal.:)
Why server redundancy? ftp.cdrom.com maxes out a T3 with one PC running BSD. (Admittedly it's a dual cpu P3 with a gig of ram or something similarly insane, but still, when's the last time it crashed?) Didn't the whole "mindcraft fiasco" show that from a sheer bandwidth perspective Linux on fairly medium hardware can keep about 5 T1 lines busy without breaking a sweat?
If (once the ISDN bottleneck is removed) CPU to run the perl scripts starts becoming a bottleneck, go with a one of those nice penguin paralell servers. You can afford it now.:)
Of course having a second box to test code changes on before putting them out "live" (I.E. crashdot) sounds like a good idea, though.:)
Define "free". Anarchy isn't necessarily what most people think of when they say freedom. This is the old "yelling 'movie' in a crowded firehouse" argument, certain limitations on the freedom to hit random strangers in the head with an axe have turned out to be a Good Idea.
The first amendment actually takes freedom away from people: the freedom to supress other people's opinions. "Freedoms to prevent" are part of the whole domain of freedom as a concept, but they're generally the ones people in a pragmatic situation restrict. The freedom to interfere with other people's freedom (with an axe, jail cell, or proprietary source code) often needs to be regulated outside of a utopian environment.
Strangely enough, RMS of all people came out with the pragmatic, real-world solution in this case. The GPL arguably protects positive freedoms by restricting the kind of "freedom to prevent" that general leads to tyrrany, anarchy, and monopoly. If the man was the foaming wild-eyed radical people keep casting him as, he'd be pro-BSD and defending people's rights to steal code from anywhere and make it proprietary.
The GPL is one of the real-world compromises, like the bill of rights, that applies a little duct-tape to the theory until you get something that actually works. It prevents code forking, and makes projects un-FUDdable.
Rob
Re:Static page requests, BAH! So what?!!
on
NT vs. Linux: Again
·
· Score: 1
It's not just a question of static pages and images, which are fairly common. It's the fact they're serving thousands of connections that complete instantly without a single pause or dropped packet. Oh yeah that's a real world scenario, the majority of users out there are surfing on a 56k modem or a T1 they share with a ton of co-workers. The majority of users are around ten hops away! A real world connection is GOING to take a few seconds to complete. THAT'S why boxes go to their knees when they get flooded, they're serving thousands of stalled connections, fresh requests come in and there are no resources (threads, I/O ports, memory, CPU time, or bandwidth with the zillions of retransmitted packets) to service them.
This benchmark is about as valid as a for loop that stays entirely within the CPU cache. Not likely.
(That said, we need to fix it just for bragging rights. But anybody who bases a deployment decision on this kind of nonsense deserves what they get.)
As long as it's a binary release, it really should use Red Hat's Package Manager. We should let all application developers know this. A gzipped tar file can't check for prerequisites, select a default place to install itself, or make simple configuration changes. They're not designed to. That's not what they're for.
RPM's aren't distribution-specific anymore (even Debian and Slackware now handle them pretty naturally with the toys in their respective base install). They're common across the whole of Linux now. Application developers who want to make packages easily installable on Linux by newbies (and, like it or not, each of us was a newbie at one point, how else do you gain users but by attracting and converting newbies) should use RPM any place they'd use "installshield" on a windoze.
How "professional" is it to release a ".zip" file for a windows program? A configuration readme is fine for us techies, but even I have been known to put off installing new software indefinitely if I know it's going to take more than 2 minutes to get it up and I'm just evaluating it, so I won't know if it's worth the effort until I've seen it!
RPM's provide a competitive advantage to Linux application software vendors that use them. It's as much a part of making Linux newbie-friendly (or "career secretary who needs help finding the on switch" friendly) as drool-and-click desktops.
Speaking of which, a desktop can do a much better job figuring out what to do when you click on an RPM than it can when you click on a.tgz file.
First we have "flash bios" that a virus can blank and render the hardware useless, and now we have chips in development which can be physically shorted out in software. Oh yeah, there's a step forward... Infinite monkey attacks happen in real life: if it can go wrong some clueless newbie will stumble across it. Count on it. (That's how we debug Linux, isn't it?:)
Ownership of creative content is a long-running debate (look into the whole litigation over the pen name "Robert Cringely" between the copyright owner infoworld & the guy who made it famous writing columns in their employ).
The people who are responsible for the creative content generate the real value, and are the geese that lay the golden eggs. The existing backlog of creative content is always a finite resource, no matter how big a backlog that is. It doesn't matter if we're talking about programmers, authors, musicians, painters, actors...
If kernelnotes now has the people who made LinuxHQ great, then kernelnotes will prosper and LinuxHQ will fade unless they have their own talent (which is, as yet, unproven). In the mean time, if the original creative team that won the award (even if it's only for collating data, there's a lot of valid creative effort in good organization and an intelligible presentation. Ask Tim O'Reilly...) is at kernelnotes, then they should get the "cool site" award. The new linuxhq can earn its own, if it's up to it.
Nah, the coca-cola thing is based on a century of established brand name, one heck of a distribution and marketing infrastructure, and a fairly effective caffiene delivery system. As "new Coke" proved, the century of marketing has been successful enough that what it actually tastes like is almost completely irrelevant by now, (unless of course you try to change it). Modern lab techniques can easily figure out what's in it pretty accurately, and the formula's been published anyway (in the back of the book "For God, Country, and Coca-Cola", check your library). The trade secret is for marketing reasons, so nobody else can claim to be identical to Coke. (Couldn't back it up in court without a certified copy of the original formula, and you haven't got the formula, have you? Lab analysis is +/- a few percent on the quantity of each ingredient, and is it chinese cinnamon or...? Not good enough to hold up in court, all you could get was "very similar to Coke" which Pepsi is anyway with a vastly different formula. They both taste like malted battery acid, actually... (I drink Dr. Pepper, although I own stock in Coke.))
Slashdot Mirror
>automatically ("mechanically"? sweet God!) from
>HTML to TeX in a reasonable way, since TeX
>contains information that is entirely absent in
>HTML. Ditto HTML to SGML. Ditto HTML to
>PostScript and vice versa. Ditto plaintext to
>much of anything.
Who said anything about gaining information in the conversion? You can make postscript out of plain ascii text. And when you print it, guess what, it looks like plain ascii. (WHY you'd want to do this is an open question.)
That "much more info" isn't what people read the documentation for, is it? Will multiple fonts help me beat "chat" into submission? (I still dial the modem with minicom and quit out to run the PPP daemon. That sucker's evil.)
The main problem with plain text is that the wordwrapping is fixed, and even then why is that a bad minimum acceptable format if the documentation writer has something interesting to say? If somebody has information I want, I want them to write it up and get it to me, not to be deterred by what format it has to be posted in. Converting formats is the archive's job, not the author's.
And the archive CAN convert. Mechanically if they just MUST have it in format X, and no that doesn't add markup information that the original format didn't have. But you don't have to be a subject matter expert to add markup. A maintainer with time on their hands can take a plain ascii email and re-paragraph it and add boldface and underlining and everyhing, without being an expert in the content. I don't have to know anything about the history of sweedish currency to add markup to a document on the subject, do I? And if I screw something up, I'm sure I'll hear about it in an open environment
Sheesh, with half of the documentation I've seen we're lucky if the author can express their thoughts in english. Rejecting their submission because they didn't chose a format that allows the broadest possible range of fonts to express themselves with just doesn't make sense.
If they pick a format like Word where we wind up LOSING markup information converting to a format we're willing to post... Well, that's the breaks from them using a non-recommended format. But being unwilling to accept submissions that don't have ENOUGH markup capability... Why?
Rob
This guy wanted to rattle of a quick and dirty article about today's hot topic (Linux) by phoning up the inventor and getting a personal quote.
Linus wouldn't take his call.
So he wrote an entire article about the fact that Linus wouldn't take his call, full of speculation and allegory and other forms of hamburger helper to get the word count up.
There's a certain amount of peverse ingenuity in doing that much to avoid actually having to do any kind of work. I can't figure out if I should be amused or disgusted. Let's see...
1) Did the fact Linus wouldn't take his call actually suprise him? There IS a pending feature freeze on the 2.3 series, he's kind of swamped right now in a BIG WAY.
2) Why is he writing a column this long about something he calls inevitable within the column?
3) Did it ever occur to him that Linus's employers at Transmeta might be behind it? It is a company phone after all, he theoretically does some kind of actual work for them. PR is nice but it's not ALL he does there.
4) He needs an appointment weeks in advance to see his dentist, but won't send the fax to get on Linus's calendar.
5) He scolded somebody for asking a question about patenting genomes, but never said what great new original questions HE was going to ask Linus.
If this was a post a mailing list, this guy would probably have just found a home in my twit-filter... Must have been one heck of a tight deadline to result in an article like that...
Rob
As an end user, though, I want simple. Ascii is OK, and HTML provides miscelaneous improvements like boldface, bullet point lists, and the often underestimated ability to rewrap the sucker to a different window size without winding up with wasted space or scrollbars, or even printing it out intelligibly on the first try.
Just about anything else is probably getting too fancy. Postscript, PDF, or Tex may provide professional publishing capabilities, but will they provide legible text if I've booted from a rescue disk to a text console on somebody else's system and only have "cat" (or if I'm lucky, "less") to view the file with? I can ignore paragraph tags in that case, but I'd better have that postscript printed out somewhere...
I'm all for keep it simple.
It did NOT detect the graphics card, and I've NEVER had it detect the monitor, and yes that's made a difference on several occasions. (Especially with another box that had an SiS 5595, another piece of junk embedded video chip. With Red Hat 6 it worked fine, but only after I found the right monitor from the "nameless Dell 14 inch" list. The wrong monitor options included "refuse to run because it can't find a mode" and "loud screeching noise from the back so I shut the monitor off even before I hit ctrl-alt-backspace because I'm afraid smoke's about to come out of it". Oh configuring X is just a JOY...
Also, the modem I pulled out of the drawer for one machine recently had a jumper setting that insisted it was Com 3 according to the little sticker, but it turned out to be Com 11. (Extra jumper to shift the IRQ to 16 bit mode that was not documented on the sticker. Fun.) No Linux install has EVER found a modem on a nonstandard setting automatically (although OS/2 never had any trouble, you'd think that the INSTALL could scan for IRQ's...) This particular joy took 2 days worth of free time (I.E. most of an hour sitting down and scratching head) to figure out. Eventually scrubbed through the man pages and figured out how to run "setserial auto_irq autoconfig", which is remarkably non-obvious when you can't remember what "setserial" is called... But only after a lot of red herrings in the "plug and play" section of the Bios telling it to reserve IRQ 3 for the ISA bus...
I've been banging my head on computers since I was 12, so I've known where to stick the hot poker to get the system to confess most of the time. But a newbie would have given up, and I've wasted a LOT of time setting up boxes.
I still haven't gotten Red Hat 6 to connect to my ISP, the ISP-HOWTO that ships with this version is a JOKE, sentences cut off in the middle and it tells you to make changes to files it doesn't name. It's been a year since I last configured networking, I don't remember this stuff... I remembered where to look it up and it wasn't there, I feel cheated. :P
(I went out and bought a "Linux for dummies quick reference" after that experience with the Howtos... May not have the info I need but at least someone will have read it through before shipping it.)
Oh, another fun little detail: getting to the power down message on shutdown is no guarantee that it won't fsck on the way back up. On any Red Hat 6 system I've seen so far. Dunno why...
You can tell I detoured through OS/2 for a couple of years, can't you... Their hardware detection positively spoiled me...
Still, can't complain. A friend wanted me to explain why Windows 95 was booting into "safe mode". That'll put things right into perspective, won't it? It turned out to be a missing font. Now that's just pure evil... (Strangely enough, Red Hat 6 installed without a hitch on that box, apart from guessing the monitor type again. :) Rob
Who's trying to "beat" Gates/microsoft? We're simply rendering them irrelevant. He may hold all the cards, but we're playing chess. Open Source isn't about people who sell software, it's about people who use software. When microsoft's toll booths and potholes get in our way, we go around them, like the internet we created. We have work to do: writing papers, simulating wind tunnels, or playing doom. Microsoft is not involved. Who cares about them? Despite Gates's ego and money, the computer industry is not about him and never was. It's about us, the programmers and users who do stuff with computers. If MS is no longer an asset to us, discard it and move on. Replace the broken tools with ones that work and get back to using them, not arguing about them. All this talk of Gates is a distraction at best. Rob
The anti-usability comments are interesting. I don't know half of what goes on under the hood of my car, and if it breaks I take it to somebody who does. I can't drive a stick shift. Once a little light came on and I had to stop and figure out how to put more water in it, and I was glad the stuff under the hood was clearly labeled, but I still pay somebody to change my oil for me. Computers are like cars, phones, and toasters. 99% of the people using them don't care what makes it go, they just want it to work with as little effort on their part as possible. If you force them to learn the details, they'll go elsewhere. Period. My car takes me to work/grocery store/movie theater, I don't tinker with it in the garage and paint flames down the side. I only think about it if it breaks. The hobbyists can create a market, and guide the market, but if we don't provide what the market wants somebody else WILL. And we'll wind up having to deal with it. The point of Linux is it's a better way of doing things. But if it's only a better way of doing things for techies and not for all the regular people who use computers like toasters (stick this in here and press that button and then it comes out finished) and hire techies for the same reason I hire auto mechanics, then we're doomed to having to support Windows or Macintoshes, and that's SERIOUSLY unpleasant. (And even the techies, deep down, want the computer to work like on Star Trek. "Computer, do this". It does it. End of story.) Rob
I'd like to point out that a Journaling filesystem is a high priority on NT for the same reason that parachutes and fire extinguishers were high priorities on the Hindenburg. If the primary cause of system crashes is a backhoe taking out the power, quickly recovering from crashes isn't a primary development priority. On a system that crashes in the presence of large flowers or brightly colored wallpaper, rapid crash recovery is a big time development priority. On Linux, JFS is primarily for bragging rights, isn't it? We're focused on not crashing in the first place. Rob
This isn't necessarily arguing against doing it at all. If the demand warrants it, certainly, but is it a good first step to contemplate? I can't shake the price/performance argument, even with andover throwing money at the problem... A beefier server with raid 5 sitting on a couple of T1's struck me as a better first step...
Rob
Logically, it must be a feature. :P
Rob
>some RedHat shit on my system.
If I remember correctly, this is why Bruce Perens left. The reason -I- left debian was the general "I don't care if anyone other than me uses it, although I'd prefer it if they didn't" attitude of the mailing list. I could only take about six months of that, then I went for the distribution that was most interested in bringing open source to the largest number of people. Turned out to be Red Hat.
>Plus RPM is a proprietary binary package that
How can you have a proprietary GPL'd toolset?
>can't be extracted with standard tools, so when
>you want to read the README on a
>Sun/Alpha/Windows machine you're completly dead.
Only by your definition of "standard tools". Considering you have to download gzip for most platforms other than Linux anyway, where's the problem? (Other unixen come with "compress", remember?) On linux (including debian), a tool to use the suckers is part of the base package.
And why isn't the readme on the web page you download it from?
>The Debian package format has the advantage to be
>an 'ar' archive (+ tar.gz inside)
It has the disadvantage of nobody using it.
Nobody replaces an established standard without a darn good reason, and even then it's an uphill battle. That's the whole reason micros~1 has survived this far. GIF is patented 8 bit with no transparency and clearly superior PNG -STILL- is having an uphill battle.
There an art to knowing what can and cannot work, and one big trick is not to confuse the means with the ends (such as Stallman telling people not to call the system "Linux").
Don't stand at the bottom of hoover dam and push. It won't work. Go around, get dynamite, do something useful. But standing there and pushing is just sad.
>Proposing RPM to be the standard on Linux
>machines is the worse idea I've ever seen
You should get out more. Try visiting Washington DC, it's an eye-opener.
Rob
>prep.ai? Didn't you ever compile GCC, or egcs,
>from scratch to optimize the compiler and make it
>run faster?
Haven't played with GCC since college and I get the tool sources off of CD usually, but I am in the process of installing a system from scratch starting with kernel source code and the "bootdisk howto". Pretty straightforward going so far, although getting X to recognize an onboard SIS graphics card is a bit of a pain, I'm to the point where everything works except the pixels on the screen are out of order, I might muck around in its source to see if I can fix that...
Might make an "Install Linux yourself from source code with tweezers and a magnifying glass HOWTO" when I'm done... Right now it's just so I'll know what it all does.
>Yes, this is meant as a kind of a troll, but
>think about it. What developer with any self
>respect would complain about a non-RPM'd
>developer tool?
Anyone, such as me, who doesn't think solely and myopically of their own needs.
>I can't believe I'm hearing a Linux
>"developer" complain about not knowing how to
>handle a tar file!
Of course I know how to handle a tar file. But this is precisely the point. The developer community is extremely self-centered, that's why the system's been around for 8 years and we're only JUST starting to write a desktop for the thing.
Think of someone LEARNING to program on linux. Everything had darn well better come pre-installed or they're screwed. And there's more to install than unzipping, there's uninstalling, inventory/versioning, and fun configuration things like enabling desktop icons and making them go away again during uninstall....
Trying to get X working on my graphics card I've installed and uninstalled the server at least three times. (Uninstall the old X server, install X-SIS from suse, uninstall that and try the new 3.3.3.1 SVGA server which has exactly the same pixel ordering problem X-SIS did. The mouse leaves droppings, dragging windows smudges the contents, text is illegible, it's sad...)
It would have been a real pain to figure out which files were XBASE and which were specific to the card without a built-in uninstall. (Cleaning up the sumlinks would have been nice too, but that's at least partly a Red Hat specific problem...)
Getting places like IBM -USED- to providing uniform install packages is important. Like it or not, Red Hat's Package manager is the standard we've got. It is the most widely used install program, and I believe every single distribution except Debian and Slackware now uses it exclusively. And both Debian and slackware provide it as an option.
Rob
>you look at it from this perspective. I remember
>one project where I did put in the heavy hours,
>not because of a deadline or any management
>induced requirement, but because I needed to do
>it. Afterward, several coworkers and superiors
>expressed appreciation, but there was no huge
>payoff/bonus/promotion in it for me. And that's
>OK, because now the code is there, and its still
>in production use.
I did the exact same thing for a large nameless company (ok, IBM). I inherited a piece of junk that was technically complete, but not working, and spent evenings and weekends virtually rewriting it.
I didn't even get a "thank you" from management, just "too big a change, we can't risk it" until the test department came back and said "profoundly unacceptable in all categories" and the boss saved face with "wait, we have a fix" and took credit for my six months work as a band-aid fix. (And gave me a bad performance review for staying in my office coding instead of attending his truly pointless daily meetings.)
The reward? The code shipped, as part of a larger overall product with a declining market share (OS/2). Three years later, it's an archaeological relic.
This is another big push for open source, freeing code from context. So the program itself is no longer useful, there are a half-dozen valuable subsystems in that thing that could easily have been salvaged and re-used. I wrote an object oriented GUI management system they wanted to patent, my own (fairly optimized) grep implementation, a configuration file parsing mechanism I was outright proud of, and even some new (generally applicable) build tools.
All of the above is under lock and key somewhere, collecting dust. It's closed source, the program is no longer being developed, that's the end of it.
I made my own happy ending to the story: I left the company. Programmers are a lot like cats: as long as we're happy you can't get rid of us, make us unhappy you can't keep us.
Rob
If you're doing this for the money, YOU WILL BURN OUT. Fast. People who get CS degrees with dollar signs in their eyes don't last long. They're also generally really bad programmers (see point 1).
THIS is why free/open source is winning. The ones who care about doing it right so bad it hurts are the ones who start and run all these projects because it has to be done RIGHT, darn it! The hordes of wannabes and apathetic check cashers send us complaints about it not working right, and if we're lucky they're good enough to spot the bug. Ocasionally, they pick up enough as they go along to become real hackers.
You can't hire that, because money isn't what motivates it. "I can do better" ego and a raging sence of injustice that some idiot was getting paid for doing it wrong in the first place started Richard Stallman with the GNU project. Yearning for a decent intellectual challenge (a specialized kind of boredom), and that nasty urge to take the refrigerator apart and see how it works prodded Torvalds to complete it with the Linux kernel. Apache was just a bunch of people tinkering with a defective tool they used regularly and a "wow, cool, can I borrow that" attitude, who were lazy enough to want to make things easier on themselves by sharing the work.
Then pride, fan mail, and a general sense of accomplishment kick in and look out, 200+% annual growth.
And if we can all make money off of this kind of thing, even do it as a day job, well that's a bonus.
Rob
If you've ever had to say "no, not like that" to somebody, you understand open source.
1) You need a copy of windows to run. To do it legally costs $$$, especially NT.
2) Running a whole second OS is a serious resource hog.
3) It's effectively running on a second (virtual) computer, in its own little sealed box. Why not just get a second computer and a monitor/keyboard/rat switch?
Wine provides the Win32 system calls to a Linux process, allowing things like a windows CGI program to do credit card validation to be spawned from Linux' Apache. It may never run every windows program in existence, but:
1) Neither does any one version of Windows.
2) I don't own every windows program in existence. I only care about the ones I have (which these days, are mostly games, half of which actually run under DOS.)
3) This is legacy support. 50% of the legacy windows programs out there aren't Y2K compliant anyway, and an amazing number of people are limping along with "good enough for now" 3.1 installs left over from the 1980's for their daily word processing and checkbook balancing/payroll. (Sheesh, last year I helped a friend of a friend copy his comic book store inventory system from an old 386 SX with a 100 meg hard drive to an old 386 DX with a 200 meg drive. Only reason he left the old system was he'd tried Dos 6 doublespace and the drive started to eat itself.)
We don't HAVE to support the latest and greatest Windows apps, those companies are still around and we can lobby for a native version as we penetrate farther and farther into "grandma" land and our usage numbers go up with drool-proof interfaces like Gnome and automatic install/configuration and pre-installs. And we ALREADY support a lot of the old stuff, and creep farther every day.
The Wine people are adding new APIs faster than Microsoft is. They're better at it. Someday, they'll catch up.
Rob
Motorola couldn't guarantee the volumes of chips IBM wanted with the newer 16 bit technology. Intel's 8088 using 8 bit chips pretending to be 16 bit (yup, the 386 SX was not the first) meant they could use older more proven fabrication techniques and existing (dirt-cheap) 8 bit support chips.
I was under the impression that the 8086 existed FIRST, and the 8088 was a hack to get the volume and cheapness IBM wanted. I guess we'd have to ask an Intel veteran. Rob
Define "deviating from the standard beige case". IBM made some remarkably stylish solid black computers a year or two back. I still have the keyboard from one somewhere. Okay it wasn't bright green with a translucent case, but it certainly looked unique, and it was pre-imac. The original model T was produced in "any color the customer wants, so long as it's black". Later, they diversified into bright red sports cars and psychadelic VW bugs (or the new "Volkswagen Feature"; there's no such thing as a fuel injected bug). Look-and-feel differentiation is normal for an industry. It didn't occur to most of US for the first decade or two because we're not exactly normal, are we? Rob
And it's less arbitrary than "hot chick". It's an empirically "hot chick". Much more legal. :)
If (once the ISDN bottleneck is removed) CPU to run the perl scripts starts becoming a bottleneck, go with a one of those nice penguin paralell servers. You can afford it now. :)
Of course having a second box to test code changes on before putting them out "live" (I.E. crashdot) sounds like a good idea, though. :)
The first amendment actually takes freedom away from people: the freedom to supress other people's opinions. "Freedoms to prevent" are part of the whole domain of freedom as a concept, but they're generally the ones people in a pragmatic situation restrict. The freedom to interfere with other people's freedom (with an axe, jail cell, or proprietary source code) often needs to be regulated outside of a utopian environment.
Strangely enough, RMS of all people came out with the pragmatic, real-world solution in this case. The GPL arguably protects positive freedoms by restricting the kind of "freedom to prevent" that general leads to tyrrany, anarchy, and monopoly. If the man was the foaming wild-eyed radical people keep casting him as, he'd be pro-BSD and defending people's rights to steal code from anywhere and make it proprietary.
The GPL is one of the real-world compromises, like the bill of rights, that applies a little duct-tape to the theory until you get something that actually works. It prevents code forking, and makes projects un-FUDdable.
Rob
This benchmark is about as valid as a for loop that stays entirely within the CPU cache. Not likely.
(That said, we need to fix it just for bragging rights. But anybody who bases a deployment decision on this kind of nonsense deserves what they get.)
Rob
RPM's aren't distribution-specific anymore (even Debian and Slackware now handle them pretty naturally with the toys in their respective base install). They're common across the whole of Linux now. Application developers who want to make packages easily installable on Linux by newbies (and, like it or not, each of us was a newbie at one point, how else do you gain users but by attracting and converting newbies) should use RPM any place they'd use "installshield" on a windoze.
How "professional" is it to release a ".zip" file for a windows program? A configuration readme is fine for us techies, but even I have been known to put off installing new software indefinitely if I know it's going to take more than 2 minutes to get it up and I'm just evaluating it, so I won't know if it's worth the effort until I've seen it!
RPM's provide a competitive advantage to Linux application software vendors that use them. It's as much a part of making Linux newbie-friendly (or "career secretary who needs help finding the on switch" friendly) as drool-and-click desktops.
Speaking of which, a desktop can do a much better job figuring out what to do when you click on an RPM than it can when you click on a .tgz file.
Ok, spleen vented, moving on...
First we have "flash bios" that a virus can blank and render the hardware useless, and now we have chips in development which can be physically shorted out in software. Oh yeah, there's a step forward... Infinite monkey attacks happen in real life: if it can go wrong some clueless newbie will stumble across it. Count on it. (That's how we debug Linux, isn't it? :)
The people who are responsible for the creative content generate the real value, and are the geese that lay the golden eggs. The existing backlog of creative content is always a finite resource, no matter how big a backlog that is. It doesn't matter if we're talking about programmers, authors, musicians, painters, actors...
If kernelnotes now has the people who made LinuxHQ great, then kernelnotes will prosper and LinuxHQ will fade unless they have their own talent (which is, as yet, unproven). In the mean time, if the original creative team that won the award (even if it's only for collating data, there's a lot of valid creative effort in good organization and an intelligible presentation. Ask Tim O'Reilly...) is at kernelnotes, then they should get the "cool site" award. The new linuxhq can earn its own, if it's up to it.
I wonder if they'd like my button that says "Stop Plate Techtonics"... They could go have a protest rally against El Nino, too...
Nah, the coca-cola thing is based on a century of established brand name, one heck of a distribution and marketing infrastructure, and a fairly effective caffiene delivery system. As "new Coke" proved, the century of marketing has been successful enough that what it actually tastes like is almost completely irrelevant by now, (unless of course you try to change it). Modern lab techniques can easily figure out what's in it pretty accurately, and the formula's been published anyway (in the back of the book "For God, Country, and Coca-Cola", check your library). The trade secret is for marketing reasons, so nobody else can claim to be identical to Coke. (Couldn't back it up in court without a certified copy of the original formula, and you haven't got the formula, have you? Lab analysis is +/- a few percent on the quantity of each ingredient, and is it chinese cinnamon or...? Not good enough to hold up in court, all you could get was "very similar to Coke" which Pepsi is anyway with a vastly different formula. They both taste like malted battery acid, actually... (I drink Dr. Pepper, although I own stock in Coke.))