For this specific application, scientific supercomputing with a blade architecture, would native VLIW offer any performance benefits?
If it would only be a few percent, it wouldn't be worth it. Transmeta has picked the game they want to play, and it would be a big deal at this point to engineer a special version of their chips that make it possible to run native VLIW code.
I'm guessing that typical scientific processing involves a lot of loops that run many iterations, which is the ideal situation for the code morphing engine; so I'll go out on a limb and predict that it is not worth it to make a special version of the chip.
Okay, fair enough. The cases aren't parallel. But I can still hope that they will add the feature. We're just talking some more ROM space for the Theora codec, and no licensing fees.
I am new to linux. One of the most *frustrating* things I've found when trying to work with linux is the fact that nothing seems to come in executable form.
I'm not sure which distribution of Linux you tried, but most distros offer ready-to-go software in "packages".
A package is a file with software bundled up and ready to install: a program, possibly some support programs, other support files (such as help), all ready to go.
I suggest you try out Fedora. I hear it has a nice package system and an easy installer. It doesn't have a very large number of packages yet, but more are arriving all the time.
UserLinux will be based on Debian GNU/Linux. Debian has over 11,000 packages (!) and is the most complete distribution of free software available. It's not all that easy to install, but it's easy to add software later, once you learn the basics of apt-get.
Basically you NEED to treat the internet as a hostile zone.
Okay, ClickApt needs to have a whitelist of trusted servers. By default, it only accepts packages from those servers. By default, all official Debian mirrors are trusted.
For a business situation, the IT department sets up their own APT server(s) on the corporate intranet. All packages there are automatically trusted.
As always, if you own a computer (i.e. have root access) you could tweak the settings to make ClickApt trust a new source. Or you could just use apt-get or aptitude as now, which just blindly accept any sources listed in the sources.list file.
ClickApt would also be a nice way to finish up the install process: instead of tasksel or some other bland package selector, you would be looking at a web page that can be browsed and that can link the home page of a project so you can read about it, look at screen shots, etc.
With all my Debian experience, I know which packages I want after I install, but ClickApt would be so much more newbie-friendly.
The top feature I would like to see added to Debian is an APT-based tool, designed to be integrated into web pages. The idea is that users can read a web page, decide they want a package, and click on the package; it will then be installed. I call this "ClickApt", a version of APT that can install a package with just a single click:
0) ClickApt makes sure that sources.list already has a source for this package. If it is not already there, ClickApt adds a source to sources.list.
1) If there are any dependencies, they will be installed. The dependencies will not, by default, be enumerated; users get a list of how many files and how many MB will be downloaded, and a progress bar.
2) ClickApt should check package signatures, and official Debian packages should simply be installed without any "Are you sure?" nattering. Unsigned, unknown packages should get a warning box and an "Are you sure?". ClickApt should also refuse to touch core system packages; if someone is persuaded to click on a web link and it wants to overwrite critical stuff, that should fail. A sysadmin can always use apt-get if ClickApt isn't willing to do something that actually needs to be done.
3) ClickApt should be customizable, for experts. A "More Details..." button on the download dialog will show all the packages being downloaded and their versions.
4) ClickApt should also have a system upgrading mode, that acts like "apt-get update; apt-get dist-upgrade". It should also work with a single click.
Lindows has a system something like this, but not based on APT. Lindows users browse through a web-based "warehouse" and click on applications they would like to have on their computers, and it Just Works. Lindows.com charges a minimum of $50 per year for people to use it, and people seem to be paying. There is no reason why Debian and UserLinux shouldn't have an equally easy-to-use tool that is based on APT.
With ClickApt, Debian package installation and updating would pass the "Grandma test" mentioned in the article.
My first thought was "I hope they are going to use Ogg Theora for this." Then in the article text it said that they have been "developed... using video-compression technologies licensed by On2 Technologies". Folks, Ogg Theora is based on the On2 compression technologies!
The Chinese market is huge. Many DVD players are made in China. It seems very likely to me that the EVD standard will at least carve out a niche for itself. Potentially, it will have sufficient impact that all future DVD players will be made EVD-compatible. It ought to just be a matter of putting some more stuff in the ROM of the DVD player. It this really is based on Ogg Theora, there will be no fees or royalties to pay.
Of course, the MPAA will probably drag their heels about releasing Hollywood movies in EVD format. But I would love it if there was a widespread standard based on Ogg Theora, so I could burn my own discs using nothing but free software and know that my friends have players that can watch the discs.
AMD has said that next year they will be shipping Opteron chips that only dissipate 30 Watts. An opteron runs 32-bit code faster than an Athlon, and totally owns if you run it in 64-bit mode. If you could buy a chip like that, for the same price as an Athlon, why would you want an Athlon?
(If there was a problem getting a good motherboard for the Opteron, that would be a good reason to still want Athlons, but there isn't a problem. There are plenty of good motherboards for Opteron and Athlon64 already.)
1 redundancy drive per drive set. If you have 3 drives in the set, 2 hold data. If you have 4 drives in the set, 3 hold data. And so on. That's RAID 5.
0) IDE peformance sucks when two drives both want to use the same controller.
1) According to a Linux software RAID web page I read (but I'm not sure where; lost the URL so I can't tell you) when an IDE drive fails, it can confuse and hang the controller it's connected to. If you only have one drive per controller you don't care, but if you have two drives on one controller, one drive can fail and it can "take out" the other drive (at least untily you reboot to un-hang the controller). Since RAID can survive the loss of one drive, but not two, you really want just one IDE drive per controller.
Note that if you want to do SCSI RAID, you can just hang all the SCSI drives on one controller. But with IDE it's one controller, one drive.
I understand your data is striped across 3 drives, and you can afford to lose one - what RAID "version" is that?
RAID 5. If you connect N drives in RAID 5, you get N - 1 drives that can store data and 1 drive that's "wasted" to the redundancy. So my RAID 5 with three 120 GB drives has 240 GB of usable space, the same amount as the gigaQube.
You could always just use RAID 1, with two drives in a "mirror" (both drives kept in perfect sync). Then the single PCI slot will still be available.
You could even do something wacky like building a Linux software RAID that includes an external drive plugged in to one of the high-speed USB connectors, or one of the 1394 (FireWire) connectors. As long as Linux can recognize the device, you should be able to RAID it.
I seriously considered putting a stack of external boxes next to my server: Linux software RAID with hot-swap ability! But you pay a lot more for a 120 GB drive in an external enclosure than you pay for just a 120 GB drive, and each external enclosure will have a cooling fan and I didn't want the noise.
By the way, I left the RF shield plates in place on the external bays. The hard drives are all inside and not visible. I plan to swap round IDE cables in (replacing the current flat ones) and measure the system temperature to see if the round cables make any difference.
I built a Mini-ITX file server. It has three 120 GB hard drives; they are running Linux software RAID, so I have the same amount of storage as the gigaQube... but I can have any one hard drive die and I'm okay.
The gigaQube is smaller, but my Mini-ITX file server is small enough for me. It's also extremely quiet.
Details:
It's a VIA EPIA-M motherboard, with a 1 GHz "Nehemiah" core. It has two IDE controllers onboard, and I used an IDE controller PCI card to get another available controller for the third drive. The case is a common Mini-ITX case, almost a cube shape, which I got at the Fry's Electronics in my area. One drive is mounted in the (only) hard drive holder in the case; one drive is mounted in the 3.5" external bay; and one drive is mounted in an adapter bracket which is mounted in one of the two 5.25" bays. I actually have one 5.25" bay free, but I don't need it for anything. I use the 100 Mbit Ethernet jack on the motherboard for hooking the server up to my net, and I have Debian GNU/Linux (stable branch) installed. It's a sweet little server.
If you support private space ventures such as X-Prize, you should also support OrbDev.
Doesn't follow. OrbDev opposes the Treaty but their "solution" would be an equally big problem. Instead of no one owning stuff in space, the first yammerhead to point his finger at the sky and scream "It's mine!!!" gets everything.
(Just in case: you know all that dark matter in the universe? It's mine!!! Heck, I'll claim all the dark matter in any universe we ever discover!!! I'm preparing a fee schedule for interacting gravitationally with my dark matter. I'll be sure to send OrbDev their bill.)
The problem is that there are no system menus at all in X.
The solution is another layer on top of X: as you said yourself, GTK or Qt.
There isn't only one, but this is the free software world. Who could prevent people from starting up new competing projects?
I realize that ultimately it's up to the developer to make apps that look aesthetically pleasing, are usable, AND conform with the look and feel expectations of the platform. In Windows you at least have a baseline for what that look and feel is, and though far from perfect, the platform has much better look and feel consistency than does the X Window System platform.
X itself was designed not to have any "look and feel consistency". But GNOME and KDE both have that. All GNOME apps should look and work in a similar way (see the HIG). Likewise with KDE.
I don't run "the X Window System platform". I run GNOME... and my apps are consistent and I'm comfortable with my desktop environment. What you want is available in Linux, it's just that no one forces all developers to code for just one platform.
I agree with you about shared libraries reducing bloat.
The GNOME guys care about bloat too; GNOME uses the standard GTK dialogs for file selection (file save and file open), instead of putting a GNOME layer over GTK for those, specifically to avoid bloat. (This means that GNOME 2.4 doesn't have cool file selection dialogs yet, because the new version of GTK with the cool dialogs hasn't shipped yet.)
I'm not sure what you mean about Nautilus being custom code; I think of Nautilus as being part of GNOME these days, and they have left certain jobs to Nautilus (such as drawing the desktop).
Note that if you have a modern machine with huge amounts of memory, you can run KDE and GNOME apps side by side and never see any slowdown. If you have a creaky old machine with limited RAM, you would be better off to pick either GNOME or KDE, and run all your apps from that environment. As soon as you run Scribus under GNOME, or GnomeMeeting under KDE, you need to have both GNOME and KDE libraries loaded at the same time, which will cause swapping on a lower-memory computer.
Some people decry the large number of dependencies of modern software (GNOME or KDE). The dependencies are mostly a good thing: the more shared library use, the better for memory footprint. And since Debian does such a good job of managing dependencies, I don't have any trouble with installing applications.
Can anyone tell me how to query the Spamhaus block list (SBL) from a Linux command line? I tried to use the "dig" utility to do this ("dig @sbl.spamhaus.org suspectedspammer.com any") but it doesn't work.
I read the "how to use SBL page" (here) and I understand that I can set my MTA to use it to block spam. But I'd like to test it out a bit before putting it into production, and ideally I'd like to be able to use this in scripts.
If they had ten times as many people working on the OS, how come it took them forever to get v2 out.
Lots of reasons. For example, IBM had promised that OS/2 would run on all their 286 computers, and making OS/2 run on a 286 wasted a lot of time. Also, cooperating with IBM was a tedious process, with lots of meetings back and forth making decisions. Beyond those two, I won't speculate; check a book on the history of OS/2.
Note that OS/2 2.x shipped from IBM only. Work on that was finished after the "divorce". IBM didn't tire of MS foot-dragging; MS was already completely out of the development process by then.
And, btw, added windows support pretty quickly too.
Note that MS refused to license Win32 for OS/2, so this was only Windows 3.x support. Maybe IBM really thought OS/2 native apps would take over the world, and Win32 support wasn't needed; but probably IBM would have licensed Win32 had MS been willing. OS/2 would have had a much better chance if it could have run Win32 apps.
If you want to hate MS for something, hate MS for refusing to license Win32.
Even then, I can put thousands of people to "work" on something, but make sure they never accomplish much.
My friend worked overtime for years, as one of hundreds, and he believes MS was 100% behind OS/2. Hundreds of people would even now be a large commitment from MS, but 14 years ago it was huge.
If OS/2 was just a scam, why did MS have so many developers (including me) running it? Why did MS replace all the computers in their library with ones running OS/2? Why did they schedule free lectures for MS employees on the architecture of OS/2?
You gloss over the fact that IBM left OS/2 in microsoft's hands, and nothing got done.
How can I gloss over something that isn't true?
The famous "divorce" between MS and IBM left all OS/2 development in IBM's hands. IBM had complete control of the work done on OS/2 from that day forward.
(OS/2 was years ahead of Windows in functionality and stability. Many of Windows 3.1 programmers used the OS/2 platform because it was more stable and productive to write apps on.)
OS/2 had three big problems: it didn't run well on the hardware of the day, and it wasn't easy to port Windows apps to OS/2, and OS/2 didn't run legacy apps well (until too much later).
Businesses bought Windows 3.x because it ran on the computers that they actually had, and they could run all their DOS apps side by side with any Windows apps. (OS/2 1.x would only run a single, well-behaved DOS app in the "compatibility box", and many DOS apps were not well-behaved enough and would lock up the computer.) Once they bought in on Windows, they were no longer interested in OS/2 because there was no easy migration path from Windows to OS/2.
You say that many Windows developers used OS/2 for development of Windows 3.1 apps. I don't remember that, and my friend who worked on OS/2 doesn't remember that either. Certainly no one ran Windows 3.1 apps on OS/2 because it didn't work. I worked on Microsoft Word for DOS, versions 5.5 and 6.0, and we built those on OS/2 (and tested on both OS/2 and DOS). But the Windows developers built on Windows, as far as I recall.
I think MS's plan became - hold OS/2, develop windows. Make sure we have Windows apps ready to go. Don't spend much time on OS/2. Clearly, MS bit the hand that fed them. Sadly, IBM was too distracted and uninterested in the PC market to actually do anything about it.
Dude, I was there and this is pure fantasy. Microsoft covered all the bets, and the Windows bets paid off and the OS/2 ones didn't. My friend who worked on OS/2 said that when Windows 3.0 launched, there were about 30 developers on the Windows 3.0 team... and about 300 developers on OS/2. That's ten times as many developers working on OS/2. And IBM was not uninterested; they had huge legions of developers working on OS/2 also.
Do you know how long the OS/2 v2 upgrade programming project was underway? I suspect it was long underway before Windows 3.0 shipped. I'd have to go back and look, but OS/2 2.0 was under MS development for more than five years before IBM yanked it.
My friend who actually worked on that project says he thinks it started in 1988, and that it shipped somewhere around 1992. And he confirmed what I told you: after the "divorce" IBM took over all development, and Microsoft put all the hundreds of OS/2 developers on other projects.
I'll say it again: Microsoft really thought OS/2 was the future, but the customers voted with their dollars for Windows, and MS decided to focus on Windows. IBM still wanted OS/2, so the companies "divorced" and IBM took on all OS/2 development. There was no secret plan by MS to trick everyone else. And if IBM, way back when, hadn't insisted on making OS/2 incompatible with Windows, there might have been an upgrade path and it might have been OS/2 that took over from Windows.
Everyone was convinced by Microsoft that OS/2 was going to be the next big thing.
Then MS stabbed IBM in the heart, dumped OS/2 development, poured it on Windows, and got Windows 3.0 out the door
Strange, revisionist history. The MS/IBM "divorce" happened after 3.0 had already shipped. MS didn't stab IBM; they abandoned OS/2 development, and IBM kept OS/2 for themselves.
I worked at Microsoft while this was going on. When I was hired, in 1990, Microsoft thought OS/2 was the next big thing. Windows was viewed as a toy that was a stepping-stone to OS/2. But customer reaction to Windows 3.0 was overwhelmingly positive, and Microsoft decided to go with what the customers actually wanted.
Looking further back, the whole reason for the Windows vs. OS/2 conflict was that IBM insisted on making the graphics system in OS/2 work very differently than the graphics system in Windows. IBM felt that the OS/2 way was better, but it made it much harder to write one application that could be natively built for both Windows and OS/2. If there had been a smooth, seamless transition available to move from Windows to OS/2, OS/2 would have worked out much better than it did.
So, WordPerfect and everyone else had apps ready to go on OS/2, and Microsoft had apps ready to run on Windows. Still, Word sucked, and Excel wasn't much better.
Microsoft isn't monolithic. The MS Windows guys were trying to get companies to write for Windows. The MS OS/2 guys were trying to get companies to write for OS/2. Microsoft, itself, covered all bases and supported everything on both Windows and OS/2. When it turned out that customers wanted Windows, it turned out that customers bought Windows apps from Microsoft. Most other companies placed bets only on OS/2 and had to scramble to release Windows software.
What's my mom going to touch up her photos in, gimp?
Why not? I'd suggest giving her GIMP 1.3, which is much nicer than the current stable GIMP. When it releases (soon) it will be GIMP 2.0.
I guess your point is that your mom won't know how to install the GIMP. I don't see why that's a problem.
0) If you set her up to use Debian, she could type "apt-get install gimp-1.3". Most people are smart enough to learn to do this. You could even talk her through it with a phone call.
1) Even if she isn't using Debian, you could ssh in to her system and set it up for her.
2) You could set her up with Lindows, as long as you configure it to not run as root all the time, and as long as she is paid up on her subscription to Click-N-Run, she won't ever call you to get new software. Click-N-Run is easy to use -- easier than installing software under Windows.
You said the failure of the ACLU to take on this is because it is biased for gun control.
More or less what I said. With examples to back up the assertion.
That's bullshit bashing since there's a better reason for the ACLU not to take the case
Actually, I never said that the ACLU's refusal to support Second Amendment rights was the only reason the ACLU won't touch this. Just that it's sufficient.
It seems reasonable to me that if the ACLU has never, ever supported the Second Amendment, that it is unlikely to start by taking this case.
And I submit that my post wasn't bashing, given that I stated facts and provided some evidence. If I had taken the next step and said "...and because they won't support the Second Amendment, they are worthless slime lords" or some such, that would be bashing.
It's more of a bill on which several rights are illustrated than an illustrated Bill of Rights.
Interesting. It is the ACLU, not me, that is calling that poster the "Illustrated Bill of Rights" (on the web page; the poster itself appears to say "The American Civil Liberties Union Bill of Rights" in all capital letters). My understanding is that the first ten Amendments are collectively called the "Bill of Rights". It's interesting that the ACLU would use that name while using a different assortment of Amendments.
if you have an explicit anti-2nd Ammendment blocking category
They don't. They have a "weapons" category, and NRA's political arm is filed there too. And the blocking is enabled by default, while anti-gun organizations are not blocked. That ain't right.
Please point out what in my original post is "mindless ACLU bashing".
Symatec's firewall is a private product made by a private company
The ACLU nonetheless has a history of turning a blind eye to Second Amendment issues, which is what the parent post was talking about and what I was talking about.
Yes, they also protect speech, many times speech you may not like.
And for that, at least, I give them some props. Free speech is of little value if it can only be used for non-controversial stuff. We don't need the ACLU to protect our right to comment cheerfully on good weather.
Slashdot Mirror
For this specific application, scientific supercomputing with a blade architecture, would native VLIW offer any performance benefits?
If it would only be a few percent, it wouldn't be worth it. Transmeta has picked the game they want to play, and it would be a big deal at this point to engineer a special version of their chips that make it possible to run native VLIW code.
I'm guessing that typical scientific processing involves a lot of loops that run many iterations, which is the ideal situation for the code morphing engine; so I'll go out on a limb and predict that it is not worth it to make a special version of the chip.
steveha
Okay, fair enough. The cases aren't parallel. But I can still hope that they will add the feature. We're just talking some more ROM space for the Theora codec, and no licensing fees.
steveha
I am new to linux. One of the most *frustrating* things I've found when trying to work with linux is the fact that nothing seems to come in executable form.
I'm not sure which distribution of Linux you tried, but most distros offer ready-to-go software in "packages".
A package is a file with software bundled up and ready to install: a program, possibly some support programs, other support files (such as help), all ready to go.
I suggest you try out Fedora. I hear it has a nice package system and an easy installer. It doesn't have a very large number of packages yet, but more are arriving all the time.
UserLinux will be based on Debian GNU/Linux. Debian has over 11,000 packages (!) and is the most complete distribution of free software available. It's not all that easy to install, but it's easy to add software later, once you learn the basics of apt-get.
steveha
Basically you NEED to treat the internet as a hostile zone.
Okay, ClickApt needs to have a whitelist of trusted servers. By default, it only accepts packages from those servers. By default, all official Debian mirrors are trusted.
For a business situation, the IT department sets up their own APT server(s) on the corporate intranet. All packages there are automatically trusted.
As always, if you own a computer (i.e. have root access) you could tweak the settings to make ClickApt trust a new source. Or you could just use apt-get or aptitude as now, which just blindly accept any sources listed in the sources.list file.
ClickApt would also be a nice way to finish up the install process: instead of tasksel or some other bland package selector, you would be looking at a web page that can be browsed and that can link the home page of a project so you can read about it, look at screen shots, etc.
With all my Debian experience, I know which packages I want after I install, but ClickApt would be so much more newbie-friendly.
steveha
The top feature I would like to see added to Debian is an APT-based tool, designed to be integrated into web pages. The idea is that users can read a web page, decide they want a package, and click on the package; it will then be installed. I call this "ClickApt", a version of APT that can install a package with just a single click:
0) ClickApt makes sure that sources.list already has a source for this package. If it is not already there, ClickApt adds a source to sources.list.
1) If there are any dependencies, they will be installed. The dependencies will not, by default, be enumerated; users get a list of how many files and how many MB will be downloaded, and a progress bar.
2) ClickApt should check package signatures, and official Debian packages should simply be installed without any "Are you sure?" nattering. Unsigned, unknown packages should get a warning box and an "Are you sure?". ClickApt should also refuse to touch core system packages; if someone is persuaded to click on a web link and it wants to overwrite critical stuff, that should fail. A sysadmin can always use apt-get if ClickApt isn't willing to do something that actually needs to be done.
3) ClickApt should be customizable, for experts. A "More Details..." button on the download dialog will show all the packages being downloaded and their versions.
4) ClickApt should also have a system upgrading mode, that acts like "apt-get update; apt-get dist-upgrade". It should also work with a single click.
Lindows has a system something like this, but not based on APT. Lindows users browse through a web-based "warehouse" and click on applications they would like to have on their computers, and it Just Works. Lindows.com charges a minimum of $50 per year for people to use it, and people seem to be paying. There is no reason why Debian and UserLinux shouldn't have an equally easy-to-use tool that is based on APT.
With ClickApt, Debian package installation and updating would pass the "Grandma test" mentioned in the article.
steveha
I didn't realize that On2 had other, improved codecs. VP3 (which Ogg Theora is based on) is an older one. VP5 and VP6 are newer ones.
Oh, well. I can still hope that the EVD standard will play Ogg Theora as a bonus. Most DVD players can play VCD, so that isn't too unlikely.
Thanks for the correction.
steveha
My first thought was "I hope they are going to use Ogg Theora for this." Then in the article text it said that they have been "developed... using video-compression technologies licensed by On2 Technologies". Folks, Ogg Theora is based on the On2 compression technologies!
The Chinese market is huge. Many DVD players are made in China. It seems very likely to me that the EVD standard will at least carve out a niche for itself. Potentially, it will have sufficient impact that all future DVD players will be made EVD-compatible. It ought to just be a matter of putting some more stuff in the ROM of the DVD player. It this really is based on Ogg Theora, there will be no fees or royalties to pay.
Of course, the MPAA will probably drag their heels about releasing Hollywood movies in EVD format. But I would love it if there was a widespread standard based on Ogg Theora, so I could burn my own discs using nothing but free software and know that my friends have players that can watch the discs.
steveha
AMD has said that next year they will be shipping Opteron chips that only dissipate 30 Watts. An opteron runs 32-bit code faster than an Athlon, and totally owns if you run it in 64-bit mode. If you could buy a chip like that, for the same price as an Athlon, why would you want an Athlon?
(If there was a problem getting a good motherboard for the Opteron, that would be a good reason to still want Athlons, but there isn't a problem. There are plenty of good motherboards for Opteron and Athlon64 already.)
steveha
1 redundancy drive per drive set. If you have 3 drives in the set, 2 hold data. If you have 4 drives in the set, 3 hold data. And so on. That's RAID 5.
steveha
why you didn't use 2 drives on one IDE channels?
0) IDE peformance sucks when two drives both want to use the same controller.
1) According to a Linux software RAID web page I read (but I'm not sure where; lost the URL so I can't tell you) when an IDE drive fails, it can confuse and hang the controller it's connected to. If you only have one drive per controller you don't care, but if you have two drives on one controller, one drive can fail and it can "take out" the other drive (at least untily you reboot to un-hang the controller). Since RAID can survive the loss of one drive, but not two, you really want just one IDE drive per controller.
Note that if you want to do SCSI RAID, you can just hang all the SCSI drives on one controller. But with IDE it's one controller, one drive.
I understand your data is striped across 3 drives, and you can afford to lose one - what RAID "version" is that?
RAID 5. If you connect N drives in RAID 5, you get N - 1 drives that can store data and 1 drive that's "wasted" to the redundancy. So my RAID 5 with three 120 GB drives has 240 GB of usable space, the same amount as the gigaQube.
You could always just use RAID 1, with two drives in a "mirror" (both drives kept in perfect sync). Then the single PCI slot will still be available.
You could even do something wacky like building a Linux software RAID that includes an external drive plugged in to one of the high-speed USB connectors, or one of the 1394 (FireWire) connectors. As long as Linux can recognize the device, you should be able to RAID it.
I seriously considered putting a stack of external boxes next to my server: Linux software RAID with hot-swap ability! But you pay a lot more for a 120 GB drive in an external enclosure than you pay for just a 120 GB drive, and each external enclosure will have a cooling fan and I didn't want the noise.
Good luck with your projects.
steveha
By the way, I left the RF shield plates in place on the external bays. The hard drives are all inside and not visible. I plan to swap round IDE cables in (replacing the current flat ones) and measure the system temperature to see if the round cables make any difference.
steveha
I built a Mini-ITX file server. It has three 120 GB hard drives; they are running Linux software RAID, so I have the same amount of storage as the gigaQube... but I can have any one hard drive die and I'm okay.
The gigaQube is smaller, but my Mini-ITX file server is small enough for me. It's also extremely quiet.
Details:
It's a VIA EPIA-M motherboard, with a 1 GHz "Nehemiah" core. It has two IDE controllers onboard, and I used an IDE controller PCI card to get another available controller for the third drive. The case is a common Mini-ITX case, almost a cube shape, which I got at the Fry's Electronics in my area. One drive is mounted in the (only) hard drive holder in the case; one drive is mounted in the 3.5" external bay; and one drive is mounted in an adapter bracket which is mounted in one of the two 5.25" bays. I actually have one 5.25" bay free, but I don't need it for anything. I use the 100 Mbit Ethernet jack on the motherboard for hooking the server up to my net, and I have Debian GNU/Linux (stable branch) installed. It's a sweet little server.
steveha
If you support private space ventures such as X-Prize, you should also support OrbDev.
Doesn't follow. OrbDev opposes the Treaty but their "solution" would be an equally big problem. Instead of no one owning stuff in space, the first yammerhead to point his finger at the sky and scream "It's mine!!!" gets everything.
(Just in case: you know all that dark matter in the universe? It's mine!!! Heck, I'll claim all the dark matter in any universe we ever discover!!! I'm preparing a fee schedule for interacting gravitationally with my dark matter. I'll be sure to send OrbDev their bill.)
steveha
The problem is that there are no system menus at all in X.
The solution is another layer on top of X: as you said yourself, GTK or Qt.
There isn't only one, but this is the free software world. Who could prevent people from starting up new competing projects?
I realize that ultimately it's up to the developer to make apps that look aesthetically pleasing, are usable, AND conform with the look and feel expectations of the platform. In Windows you at least have a baseline for what that look and feel is, and though far from perfect, the platform has much better look and feel consistency than does the X Window System platform.
X itself was designed not to have any "look and feel consistency". But GNOME and KDE both have that. All GNOME apps should look and work in a similar way (see the HIG). Likewise with KDE.
I don't run "the X Window System platform". I run GNOME... and my apps are consistent and I'm comfortable with my desktop environment. What you want is available in Linux, it's just that no one forces all developers to code for just one platform.
steveha
I agree with you about shared libraries reducing bloat.
The GNOME guys care about bloat too; GNOME uses the standard GTK dialogs for file selection (file save and file open), instead of putting a GNOME layer over GTK for those, specifically to avoid bloat. (This means that GNOME 2.4 doesn't have cool file selection dialogs yet, because the new version of GTK with the cool dialogs hasn't shipped yet.)
I'm not sure what you mean about Nautilus being custom code; I think of Nautilus as being part of GNOME these days, and they have left certain jobs to Nautilus (such as drawing the desktop).
Note that if you have a modern machine with huge amounts of memory, you can run KDE and GNOME apps side by side and never see any slowdown. If you have a creaky old machine with limited RAM, you would be better off to pick either GNOME or KDE, and run all your apps from that environment. As soon as you run Scribus under GNOME, or GnomeMeeting under KDE, you need to have both GNOME and KDE libraries loaded at the same time, which will cause swapping on a lower-memory computer.
Some people decry the large number of dependencies of modern software (GNOME or KDE). The dependencies are mostly a good thing: the more shared library use, the better for memory footprint. And since Debian does such a good job of managing dependencies, I don't have any trouble with installing applications.
steveha
Can anyone tell me how to query the Spamhaus block list (SBL) from a Linux command line? I tried to use the "dig" utility to do this ("dig @sbl.spamhaus.org suspectedspammer.com any") but it doesn't work.
I read the "how to use SBL page" (here) and I understand that I can set my MTA to use it to block spam. But I'd like to test it out a bit before putting it into production, and ideally I'd like to be able to use this in scripts.
steveha
re: 10X as many people
OS/2 the OS or OS/2 Apps?
OS/2 the OS.
If they had ten times as many people working on the OS, how come it took them forever to get v2 out.
Lots of reasons. For example, IBM had promised that OS/2 would run on all their 286 computers, and making OS/2 run on a 286 wasted a lot of time. Also, cooperating with IBM was a tedious process, with lots of meetings back and forth making decisions. Beyond those two, I won't speculate; check a book on the history of OS/2.
Note that OS/2 2.x shipped from IBM only. Work on that was finished after the "divorce". IBM didn't tire of MS foot-dragging; MS was already completely out of the development process by then.
And, btw, added windows support pretty quickly too.
Note that MS refused to license Win32 for OS/2, so this was only Windows 3.x support. Maybe IBM really thought OS/2 native apps would take over the world, and Win32 support wasn't needed; but probably IBM would have licensed Win32 had MS been willing. OS/2 would have had a much better chance if it could have run Win32 apps.
If you want to hate MS for something, hate MS for refusing to license Win32.
Even then, I can put thousands of people to "work" on something, but make sure they never accomplish much.
My friend worked overtime for years, as one of hundreds, and he believes MS was 100% behind OS/2. Hundreds of people would even now be a large commitment from MS, but 14 years ago it was huge.
If OS/2 was just a scam, why did MS have so many developers (including me) running it? Why did MS replace all the computers in their library with ones running OS/2? Why did they schedule free lectures for MS employees on the architecture of OS/2?
steveha
MS had ten times as many people working on OS/2 as Windows, and you insist that MS "didn't put much effort in making it fly"?
I guess you are going to believe what you want to believe, never mind any evidence. So there's not much more I can say.
steveha
You gloss over the fact that IBM left OS/2 in microsoft's hands, and nothing got done.
How can I gloss over something that isn't true?
The famous "divorce" between MS and IBM left all OS/2 development in IBM's hands. IBM had complete control of the work done on OS/2 from that day forward.
(OS/2 was years ahead of Windows in functionality and stability. Many of Windows 3.1 programmers used the OS/2 platform because it was more stable and productive to write apps on.)
OS/2 had three big problems: it didn't run well on the hardware of the day, and it wasn't easy to port Windows apps to OS/2, and OS/2 didn't run legacy apps well (until too much later).
Businesses bought Windows 3.x because it ran on the computers that they actually had, and they could run all their DOS apps side by side with any Windows apps. (OS/2 1.x would only run a single, well-behaved DOS app in the "compatibility box", and many DOS apps were not well-behaved enough and would lock up the computer.) Once they bought in on Windows, they were no longer interested in OS/2 because there was no easy migration path from Windows to OS/2.
You say that many Windows developers used OS/2 for development of Windows 3.1 apps. I don't remember that, and my friend who worked on OS/2 doesn't remember that either. Certainly no one ran Windows 3.1 apps on OS/2 because it didn't work. I worked on Microsoft Word for DOS, versions 5.5 and 6.0, and we built those on OS/2 (and tested on both OS/2 and DOS). But the Windows developers built on Windows, as far as I recall.
I think MS's plan became - hold OS/2, develop windows. Make sure we have Windows apps ready to go. Don't spend much time on OS/2. Clearly, MS bit the hand that fed them. Sadly, IBM was too distracted and uninterested in the PC market to actually do anything about it.
Dude, I was there and this is pure fantasy. Microsoft covered all the bets, and the Windows bets paid off and the OS/2 ones didn't. My friend who worked on OS/2 said that when Windows 3.0 launched, there were about 30 developers on the Windows 3.0 team... and about 300 developers on OS/2. That's ten times as many developers working on OS/2. And IBM was not uninterested; they had huge legions of developers working on OS/2 also.
Do you know how long the OS/2 v2 upgrade programming project was underway? I suspect it was long underway before Windows 3.0 shipped. I'd have to go back and look, but OS/2 2.0 was under MS development for more than five years before IBM yanked it.
My friend who actually worked on that project says he thinks it started in 1988, and that it shipped somewhere around 1992. And he confirmed what I told you: after the "divorce" IBM took over all development, and Microsoft put all the hundreds of OS/2 developers on other projects.
I'll say it again: Microsoft really thought OS/2 was the future, but the customers voted with their dollars for Windows, and MS decided to focus on Windows. IBM still wanted OS/2, so the companies "divorced" and IBM took on all OS/2 development. There was no secret plan by MS to trick everyone else. And if IBM, way back when, hadn't insisted on making OS/2 incompatible with Windows, there might have been an upgrade path and it might have been OS/2 that took over from Windows.
steveha
Everyone was convinced by Microsoft that OS/2 was going to be the next big thing.
Then MS stabbed IBM in the heart, dumped OS/2 development, poured it on Windows, and got Windows 3.0 out the door
Strange, revisionist history. The MS/IBM "divorce" happened after 3.0 had already shipped. MS didn't stab IBM; they abandoned OS/2 development, and IBM kept OS/2 for themselves.
I worked at Microsoft while this was going on. When I was hired, in 1990, Microsoft thought OS/2 was the next big thing. Windows was viewed as a toy that was a stepping-stone to OS/2. But customer reaction to Windows 3.0 was overwhelmingly positive, and Microsoft decided to go with what the customers actually wanted.
Looking further back, the whole reason for the Windows vs. OS/2 conflict was that IBM insisted on making the graphics system in OS/2 work very differently than the graphics system in Windows. IBM felt that the OS/2 way was better, but it made it much harder to write one application that could be natively built for both Windows and OS/2. If there had been a smooth, seamless transition available to move from Windows to OS/2, OS/2 would have worked out much better than it did.
So, WordPerfect and everyone else had apps ready to go on OS/2, and Microsoft had apps ready to run on Windows. Still, Word sucked, and Excel wasn't much better.
Microsoft isn't monolithic. The MS Windows guys were trying to get companies to write for Windows. The MS OS/2 guys were trying to get companies to write for OS/2. Microsoft, itself, covered all bases and supported everything on both Windows and OS/2. When it turned out that customers wanted Windows, it turned out that customers bought Windows apps from Microsoft. Most other companies placed bets only on OS/2 and had to scramble to release Windows software.
steveha
What's my mom going to touch up her photos in, gimp?
Why not? I'd suggest giving her GIMP 1.3, which is much nicer than the current stable GIMP. When it releases (soon) it will be GIMP 2.0.
I guess your point is that your mom won't know how to install the GIMP. I don't see why that's a problem.
0) If you set her up to use Debian, she could type "apt-get install gimp-1.3". Most people are smart enough to learn to do this. You could even talk her through it with a phone call.
1) Even if she isn't using Debian, you could ssh in to her system and set it up for her.
2) You could set her up with Lindows, as long as you configure it to not run as root all the time, and as long as she is paid up on her subscription to Click-N-Run, she won't ever call you to get new software. Click-N-Run is easy to use -- easier than installing software under Windows.
steveha
Debian really needs a slick and easy installer like Red Hat has.
Okay. How about the same one?
Red Hat's "Anaconda" installer has been ported for Debian by Progeny. (Progeny's own installer project, PGI, is now abandoned.)
I really look forward to trying out an Anaconda install of Debian.
steveha
You said the failure of the ACLU to take on this is because it is biased for gun control.
More or less what I said. With examples to back up the assertion.
That's bullshit bashing since there's a better reason for the ACLU not to take the case
Actually, I never said that the ACLU's refusal to support Second Amendment rights was the only reason the ACLU won't touch this. Just that it's sufficient.
It seems reasonable to me that if the ACLU has never, ever supported the Second Amendment, that it is unlikely to start by taking this case.
And I submit that my post wasn't bashing, given that I stated facts and provided some evidence. If I had taken the next step and said "...and because they won't support the Second Amendment, they are worthless slime lords" or some such, that would be bashing.
steveha
It's more of a bill on which several rights are illustrated than an illustrated Bill of Rights.
Interesting. It is the ACLU, not me, that is calling that poster the "Illustrated Bill of Rights" (on the web page; the poster itself appears to say "The American Civil Liberties Union Bill of Rights" in all capital letters). My understanding is that the first ten Amendments are collectively called the "Bill of Rights". It's interesting that the ACLU would use that name while using a different assortment of Amendments.
if you have an explicit anti-2nd Ammendment blocking category
They don't. They have a "weapons" category, and NRA's political arm is filed there too. And the blocking is enabled by default, while anti-gun organizations are not blocked. That ain't right.
steveha
Can we end the mindless ACLU bashing?
Please point out what in my original post is "mindless ACLU bashing".
Symatec's firewall is a private product made by a private company
The ACLU nonetheless has a history of turning a blind eye to Second Amendment issues, which is what the parent post was talking about and what I was talking about.
Yes, they also protect speech, many times speech you may not like.
And for that, at least, I give them some props. Free speech is of little value if it can only be used for non-controversial stuff. We don't need the ACLU to protect our right to comment cheerfully on good weather.
steveha