Any organizations or companies behind open source--Mozilla, GNOME, KDE, Apache, Red Hat, Novell, and probably many others--would be protected by and thus have an incentive to back such an organization.
I would like to see a patent license in addition to coverage of the actual source under copyright and an existing open source license.
If the patent isn't necessary, why has Red Hat done it? Also, the Eolas browser plug-in patent was issued Nov. 17, 1998; I'm quite sure Microsoft had implemented IE before then, yet Eolas still won at least the first round of battle. (See also news.com reporting on the issue.)
Finally, you're saying that defending a patent infringement lawsuit is cheaper than filing a patent? I still find that hard to believe.
Virtually nobody is writing open-source software to place it in the public
domain. Rather, much of it is licensed under the GNU GPL, which embraces the
property rights of copyright and uses them to ensure that the code remains
open. I propose a parallel license for patents: a perpetual, irrevocable
license for open-source software[1] to implement, use, and improve the patented
concepts and inventions free of charge.
If we patent our patentable work, instead of merely copyrighting our code,
we can build a defensive patent portfolio. This would give us some leverage
against patent infringement suits, as well as being good business sense in the
current climate.
What is the harm in not adopting such a license? Besides the possibility of
open-source ultimately being crushed by patents, there is the risk of our work
becoming a de facto Microsoft R&D lab. We are already seeing that
future with XUL (or libglade) and Microsoft's XAML.
In addition, this license would give Red Hat a graceful way to keep their
promise that they
will never charge licensing fees on their patents.
And now, IBM has patented something very much like the Open Source model
itself. Can we afford to continue ignoring patents? IBM was once greatly
despised, and there is nothing to say that if Microsoft falls, they won't
become a new tyrant.
Of course, open-source developers would still need to apply and pay for the
patent, but it is much cheaper to apply than retroactively fight one.
[1] Rather than "open-source software", the patent license would have to
define which software licenses are considered open source. If the patent
license relies on an external definition like "OSI-approved", then the OSI
could change the license after the fact by changing their approvals. Since the
proposed license is irrevocable, the patent couldn't be withdrawn from it if
OSI added a license the patent-holder objected to.
(This post is based on the ideas of someone else. I'll drop them a line so
they can take credit or elaborate as they see fit.)
If you've ever looked up the benchmarks, it will kick the C3's ass. The 1GHz C3 is about half (or a bit less) of the power of an equivalently clocked Celeron.
The Amiga GUI--in OS1.3--is still better than Windows 9x. Death to auto-raising! However, the superior configurability of FVWM2 makes it better than the Amiga, because now I can determine size, position, and depth from any point along the window border.
Windows always makes me feel like I'm trying to use a brick designed as an answering machine for a calculator. It's just inconceivably stupid.
Then again, all that's purely subjective, so both of these posts are a complete waste of time.
The inet is not necessarily the be-all end-all of tech support; some people will have a standalone box for one reason or another. Please don't take tools away from others simply because you have no immediate use for them.
"Maybe the business where I work just got unlucky, but every LCD that we've gotten looks good at ONE resolution (around 1280x1024 or something like that). Want to go higher or lower? Guess that's too bad for you."
Yes, that's how LCDs work. It has a "native resolution", say 1280x1024, and it actually has exactly that many pixels on it. Running it higher will "lose" some of the pixels, and running it lower will either duplicate some of them or just give you a black border.
With a CRT, you just adjust the timings, and let the pixels slop onto whatever phosphors happen to be there. If you want multiple resolutions or superior color adjustment, you're stuck with CRT. Just go for a nice high refresh rate, and you should be fine...
What's so different about FreeBSD that you're
willing to trust it?
About the only thing I can think of is the fact
that Linux distros generally install the kitchen
sink, and FreeBSD prefers to make you add
ports/packages yourself. But with a custom
install (and a script to duplicate it), that can
be taken care of.
Is it some sort of "stability" issue? What kind
of stability? How often are you going to be
playing with expect/send pairs in a chat script
to get PPP working on a production telephony box?
Maybe it's API "stability"... in which case I'm
obliged to point out that FreeBSD currently only
emulates the 4.3BSD syscalls (see COMPAT_43 in
your kernel config.)
It obviously can't be anything to do with
commercial support, because Linux gets far more
of that than all of the BSDs.
You could very well be right about that particular error. C != (3|4), one of which is the illegal instruction code. (Wow, I haven't done any assembly in years now. I can't even remember what the most common guru error I saw was!) Silly me for counting letters of the alphabet....
If the post-. part of the number was really two separate 16-bit codes, there was no visual element to separate them, and the first one was quite often 0000. The pre-. part was certainly an address; I could tell by where it crashed whether it was something I did (0020xxxx) directly or whether it was something I caused the ROM to do (00Fxxxxx). (0020xxxx because we had a whopping 2 MB of fast RAM. On a pair of 30-pin 1x8 SIMMs even, and those were the high-density modules...)
One last note... we only had two A500's, and it really wasn't until our second one with Workbench 1.3 that I got into the assembly programming. I got to see my share of "SOFTWARE ERROR-TASK HELD. Click Retry to try again or Cancel to reset/debug" requesters, in which clicking Retry did nothing and Cancel proceeded to the guru meditation.
This one occasionally bites me ever since I trashed the MBR a couple of years ago. (I am still grateful for gpart.) Because of the way I reinstalled things, Win98's file system is not quite as big as its partition, and occasionally it emits the error on booting, with no removable media in any drive:
The volume that was removed had open files on it. Next time be sure the volume can really be removed before ejecting it.
Attempting to continue causes a squall of SODs in both blue and black, proclaiming the death of everything multiple times. Eventually it locks hard and has to be reset in hardware.
Simply trying to give the three-finger salute causes the same hard lock.
Then, of course, it bitches that it wasn't shut down properly. Sheesh, it didn't even get started!
Your.sig is correct--you are wrong. The Amiga gave out 32-bit errors, with a period between them.
In case anyone cares, for the error number you give, that's an illegal instruction encountered at address 0x8001.
My favorite meditations were the equivalent of the UNIX bus error, when the given address was way outside the 24-bit space that our A500 used.... "87FFFFFC? What's it doing there?"
I didn't see this firsthand, but my father did, so it's not quite urban legend yet.
He was doing something on the mainframe one day at work (back in the day when 1200 bps was fast!) and got the following error:
Unusual use of parentheses accepted with some reservations as to intended meaning.
Very few people there had ever seen that particular error.
Re:Internet Explorer has amused me many times...
on
Gnarly Error Messages
·
· Score: 2
"Ignoring this, everything works fine, but clicking OK will of course close all instances of the browser."
I've seen this many a time on our old Pentium struggling with Netscape 4.7. If the browser is still operable, I just push the error dialog down under the Taskbar and go merrily on my way.
Another problem that NS likes to throw up on that machine is an infinite crash loop. The first box looks like a normal crash, but all the ones after Close is clicked say something about advert.dll.
That machine is so old, it came with Netscape 1.x. I have no idea when my mom will give it up, though.
"s00per-high resolution in text...framebuffer...arghh!"
Hmm. Are you referring to hi-res textmodes (132x50 is my favorite on my 17" monitor), or perhaps the raster textmode (800x600, requires SC_PIXEL_MODE defined in the kernel)? Both of these require the VESA module loaded. (LOAD_VESA="YES" in/boot/loader.conf, kldload vesa, or options VESA in the kernel conf.) I think this only applies to syscons and not the VT220 emulator. You can futz with the modes via vidcontrol on the command line, then stick the relevant flags in "allscreens_flags" in/etc/rc.conf when you find something you like.
Now if only syscons supported 1152x864 raster textmode....
Slashdot is not a homogenous herd of clones, despite what moderation may suggest. People are different: we have different opinions, and we disagree with each other. In fact, it's been observed throughout history that once a minority movement gains some success, they fall apart because they're no longer so strongly united by that common goal. Yet those differences were always present.
However, I'm willing to listen to your argument if you can find me some post pairs made by a single person that contradict themselves, and prove that the information they had at the time of the later post was identical to the information they had at the time of the earlier one. It is perfectly reasonable to adapt to changing reality. That is, after all, the nature of survival.
Sysadminning (playing with toys like ipfw, op, sshd, etc.) is NOT LAN design and implementation. Go take a sysadmin class and a networking class and see if they cover any of the same material if you don't believe me.
Furthermore, there is no mention of whether he's just kind of expanded into the netadmin job simply because he was there and good with computers. At my dad's company, the netadmins became netadmins because they were the previous datapriests, and the servers got connected to the "Internet". Now they're switching to NT because their Sendmail is an open relay and they have no idea of how to fix it.
Don't procrastinate. Asking someone to debug an hour before the project's due won't help you on the job. Have time to learn how to do it.
Do the design on paper, and test a few cases on paper. Compilers can only check syntax.
Do sanity checking, especially for types and loop bounds.
Use the compiler's warning lights (-Wall, 'use warnings'...) Sometimes, they'll show that what you did is not what you meant.
Use print statements. They'll show what you did, not what you meant.
Use the debugger. It shows what you did, not what you meant.
It's easy to write code that "should" work. It's much harder to make it really work. (One of my worst days was when I told someone I could "probably" write something so trivial with no bugs. We went through ~10 test&fix cycles on a program with about as many lines.)
Fyi: Lisp is interpreted, C# and the other 20 or so languages for.NET are not.
"20 or so other languages" are really just C# in syntactic disguise. For instance, they stripped multiple-inheritance from the.NET version of Eiffel, because.NET doesn't support it. But now that it's a ".NET version" it's no longer compatible with normal Eiffel, so it's not Eiffel anymore, ne?
"Linux doesn't crash. It can't; it's simply not possible... Slashdot told me so. They said that Linux crashing would defy the laws of physics, or something."
Obviously they've never tried to change VTs when X was busy starting. (2.4.17, stock Linus kernel)
Note that OSS uses code review extensively. "To enough eyeballs, all bugs are shallow." Furthermore, even OSS programmers make mistakes--I have read with my own eyes a line of a Linux 2.4 kernel patch (NOTE: "stable" kernel!) that changed "#eliif" to "#elif". Would not some sort of review find that? Someone could probably write automated tools for that kind of thing. "Lint" might be a good name.
What was your first kernel and what was so funny about it?
I didn't find it terribly funny, but here goes.
Kernel 2.2.16-22 of Red Hat 7.0. It didn't compile. I had about a 50% success rate compiling my own Linus kernels. (2.4.x, x=0,1,2,4,5,6,9,12,13,17.) Then, kernel 2.4.18-3 on RH7.3 could compile, if you beat on it hard enough, but wouldn't boot.
I was beginning to think that expecting a release to work out of the box was expecting too much. Then I met the FreeBSD ports system.
Slashdot Mirror
Doesn't look very alive, though. No news since being /.ed in March 2001....
Any organizations or companies behind open source--Mozilla, GNOME, KDE, Apache, Red Hat, Novell, and probably many others--would be protected by and thus have an incentive to back such an organization.
I would like to see a patent license in addition to coverage of the actual source under copyright and an existing open source license.
If the patent isn't necessary, why has Red Hat done it? Also, the Eolas browser plug-in patent was issued Nov. 17, 1998; I'm quite sure Microsoft had implemented IE before then, yet Eolas still won at least the first round of battle. (See also news.com reporting on the issue.)
Finally, you're saying that defending a patent infringement lawsuit is cheaper than filing a patent? I still find that hard to believe.
Virtually nobody is writing open-source software to place it in the public domain. Rather, much of it is licensed under the GNU GPL, which embraces the property rights of copyright and uses them to ensure that the code remains open. I propose a parallel license for patents: a perpetual, irrevocable license for open-source software[1] to implement, use, and improve the patented concepts and inventions free of charge.
If we patent our patentable work, instead of merely copyrighting our code, we can build a defensive patent portfolio. This would give us some leverage against patent infringement suits, as well as being good business sense in the current climate.
What is the harm in not adopting such a license? Besides the possibility of open-source ultimately being crushed by patents, there is the risk of our work becoming a de facto Microsoft R&D lab. We are already seeing that future with XUL (or libglade) and Microsoft's XAML.
In addition, this license would give Red Hat a graceful way to keep their promise that they will never charge licensing fees on their patents.
And now, IBM has patented something very much like the Open Source model itself. Can we afford to continue ignoring patents? IBM was once greatly despised, and there is nothing to say that if Microsoft falls, they won't become a new tyrant.
Of course, open-source developers would still need to apply and pay for the patent, but it is much cheaper to apply than retroactively fight one.
[1] Rather than "open-source software", the patent license would have to define which software licenses are considered open source. If the patent license relies on an external definition like "OSI-approved", then the OSI could change the license after the fact by changing their approvals. Since the proposed license is irrevocable, the patent couldn't be withdrawn from it if OSI added a license the patent-holder objected to.
(This post is based on the ideas of someone else. I'll drop them a line so they can take credit or elaborate as they see fit.)
If you've ever looked up the benchmarks, it will kick the C3's ass. The 1GHz C3 is about half (or a bit less) of the power of an equivalently clocked Celeron.
The Amiga GUI--in OS1.3--is still better than Windows 9x. Death to auto-raising! However, the superior configurability of FVWM2 makes it better than the Amiga, because now I can determine size, position, and depth from any point along the window border.
Windows always makes me feel like I'm trying to use a brick designed as an answering machine for a calculator. It's just inconceivably stupid.
Then again, all that's purely subjective, so both of these posts are a complete waste of time.
The inet is not necessarily the be-all end-all of tech support; some people will have a standalone box for one reason or another. Please don't take tools away from others simply because you have no immediate use for them.
"Maybe the business where I work just got
unlucky, but every LCD that we've gotten looks
good at ONE resolution (around 1280x1024 or
something like that). Want to go higher or lower?
Guess that's too bad for you."
Yes, that's how LCDs work. It has a
"native resolution", say 1280x1024, and it
actually has exactly that many pixels on it.
Running it higher will "lose" some of the pixels,
and running it lower will either duplicate some
of them or just give you a black border.
With a CRT, you just adjust the timings, and let
the pixels slop onto whatever phosphors happen to
be there. If you want multiple resolutions or
superior color adjustment, you're stuck with CRT.
Just go for a nice high refresh rate, and you
should be fine...
About the only thing I can think of is the fact that Linux distros generally install the kitchen sink, and FreeBSD prefers to make you add ports/packages yourself. But with a custom install (and a script to duplicate it), that can be taken care of.
Is it some sort of "stability" issue? What kind of stability? How often are you going to be playing with expect/send pairs in a chat script to get PPP working on a production telephony box? Maybe it's API "stability"... in which case I'm obliged to point out that FreeBSD currently only emulates the 4.3BSD syscalls (see COMPAT_43 in your kernel config.)
It obviously can't be anything to do with commercial support, because Linux gets far more of that than all of the BSDs.
Would you mind enlightening me?
You could very well be right about that particular error. C != (3|4), one of which is the illegal instruction code. (Wow, I haven't done any assembly in years now. I can't even remember what the most common guru error I saw was!) Silly me for counting letters of the alphabet....
If the post-. part of the number was really two separate 16-bit codes, there was no visual element to separate them, and the first one was quite often 0000. The pre-. part was certainly an address; I could tell by where it crashed whether it was something I did (0020xxxx) directly or whether it was something I caused the ROM to do (00Fxxxxx). (0020xxxx because we had a whopping 2 MB of fast RAM. On a pair of 30-pin 1x8 SIMMs even, and those were the high-density modules...)
One last note... we only had two A500's, and it really wasn't until our second one with Workbench 1.3 that I got into the assembly programming. I got to see my share of "SOFTWARE ERROR-TASK HELD. Click Retry to try again or Cancel to reset/debug" requesters, in which clicking Retry did nothing and Cancel proceeded to the guru meditation.
This one occasionally bites me ever since I trashed the MBR a couple of years ago. (I am still grateful for gpart.) Because of the way I reinstalled things, Win98's file system is not quite as big as its partition, and occasionally it emits the error on booting, with no removable media in any drive:
The volume that was removed had open files on it. Next time be sure the volume can really be removed before ejecting it.
Attempting to continue causes a squall of SODs in both blue and black, proclaiming the death of everything multiple times. Eventually it locks hard and has to be reset in hardware.
Simply trying to give the three-finger salute causes the same hard lock.
Then, of course, it bitches that it wasn't shut down properly. Sheesh, it didn't even get started!
Your .sig is correct--you are wrong. The Amiga gave out 32-bit errors, with a period between them.
In case anyone cares, for the error number you give, that's an illegal instruction encountered at address 0x8001.
My favorite meditations were the equivalent of the UNIX bus error, when the given address was way outside the 24-bit space that our A500 used.... "87FFFFFC? What's it doing there?"
I didn't see this firsthand, but my father did, so it's not quite urban legend yet.
He was doing something on the mainframe one day at work (back in the day when 1200 bps was fast!) and got the following error:
Unusual use of parentheses accepted with some reservations as to intended meaning.
Very few people there had ever seen that particular error.
"Ignoring this, everything works fine, but clicking OK will of course close all instances of the browser."
I've seen this many a time on our old Pentium struggling with Netscape 4.7. If the browser is still operable, I just push the error dialog down under the Taskbar and go merrily on my way.
Another problem that NS likes to throw up on that machine is an infinite crash loop. The first box looks like a normal crash, but all the ones after Close is clicked say something about advert.dll.
That machine is so old, it came with Netscape 1.x. I have no idea when my mom will give it up, though.
Also, sd = da if you happen to have a 1337 SCSI system.
"s00per-high resolution in text...framebuffer...arghh!"
/boot/loader.conf, kldload vesa, or options VESA in the kernel conf.) I think this only applies to syscons and not the VT220 emulator. You can futz with the modes via vidcontrol on the command line, then stick the relevant flags in "allscreens_flags" in /etc/rc.conf when you find something you like.
Hmm. Are you referring to hi-res textmodes (132x50 is my favorite on my 17" monitor), or perhaps the raster textmode (800x600, requires SC_PIXEL_MODE defined in the kernel)? Both of these require the VESA module loaded. (LOAD_VESA="YES" in
Now if only syscons supported 1152x864 raster textmode....
Slashdot is not a homogenous herd of clones, despite what moderation may suggest. People are different: we have different opinions, and we disagree with each other. In fact, it's been observed throughout history that once a minority movement gains some success, they fall apart because they're no longer so strongly united by that common goal. Yet those differences were always present.
However, I'm willing to listen to your argument if you can find me some post pairs made by a single person that contradict themselves, and prove that the information they had at the time of the later post was identical to the information they had at the time of the earlier one. It is perfectly reasonable to adapt to changing reality. That is, after all, the nature of survival.
Sysadminning (playing with toys like ipfw, op, sshd, etc.) is NOT LAN design and implementation. Go take a sysadmin class and a networking class and see if they cover any of the same material if you don't believe me.
Furthermore, there is no mention of whether he's just kind of expanded into the netadmin job simply because he was there and good with computers. At my dad's company, the netadmins became netadmins because they were the previous datapriests, and the servers got connected to the "Internet". Now they're switching to NT because their Sendmail is an open relay and they have no idea of how to fix it.
It's easy to write code that "should" work. It's much harder to make it really work. (One of my worst days was when I told someone I could "probably" write something so trivial with no bugs. We went through ~10 test&fix cycles on a program with about as many lines.)
The version of this exploit referenced from Larholm's unpatched IE vulnerabilities does not work in Moz 1.0-RC3. It fails with "connection refused".
Clothing doesn't make people gay. Try reading this book and see if you look at the world in the same way ever again.
"Linux doesn't crash. It can't; it's simply not possible... Slashdot told me so. They said that Linux crashing would defy the laws of physics, or something."
Obviously they've never tried to change VTs when X was busy starting. (2.4.17, stock Linus kernel)
Note that OSS uses code review extensively. "To enough eyeballs, all bugs are shallow." Furthermore, even OSS programmers make mistakes--I have read with my own eyes a line of a Linux 2.4 kernel patch (NOTE: "stable" kernel!) that changed "#eliif" to "#elif". Would not some sort of review find that? Someone could probably write automated tools for that kind of thing. "Lint" might be a good name.
Kernel 2.2.16-22 of Red Hat 7.0. It didn't compile. I had about a 50% success rate compiling my own Linus kernels. (2.4.x, x=0,1,2,4,5,6,9,12,13,17.) Then, kernel 2.4.18-3 on RH7.3 could compile, if you beat on it hard enough, but wouldn't boot.
I was beginning to think that expecting a release to work out of the box was expecting too much. Then I met the FreeBSD ports system.