"I am assuming, or at least it looks like, these tests were run with the default data=journal - so to be fair, they should have been run in data=writeback, or maybe even all three modes. Again, all you have to is specify in/etc/fstab and reboot, no big deal."
And where do you get the idea that this is the default? According to mount(8):
ordered
This is the default mode.
What I really would have liked to see on this benchmark is ext3 on 2.6 with dir_index enabled. (Maybe this would also gain the benefit of the Orlov allocator? I haven't been paying attention to what's been backported.)...In fact, I would have liked to see this whole thing on 2.6.
"fake a bunch of bad transmits and request resends (does the fax protocol even do that?)"
Yes. In places with crappy lines, it's best to override the speed negotiation and set it to a sane value. (Say 9600 bps if the lines can't handle 14400. These are mildly random numbers because my father does this, not me.) Otherwise the fast line speed, which negotiation accepted because the short test burst worked without error, creates too many retransmits and falls behind on average.
"I see now there are binary packages.... Can I install and use binary packages all the time?"
Binaries are solely for speeding up installation; you install Stage 3 + GRP (that is, the binaries) and then it acts like a normal source-installed Gentoo system. You can either leave it for a while doing emerge -e world after you finish the install to put your own USE/CFLAGS settings into effect, or just let things get replaced naturally when you update things.
rac argues for this to continue here; I'm not aware of any other information.
I've been thinking of hacking up some sort of tone generator to a Wacom tablet, and mapping X to pitch, Y to timbre (or maybe gestures for things like wah-wah? That axis is undesigned ATM), and pressure to volume. Of course, this is all vaporware still...
From emerge(5): ENVIRONMENT OPTIONS
ROOT = [path]
Use ROOT to specify the target root filesystem to be used for
merging packages or ebuilds. Defaults to/. This may have been added recently, since I've never looked for it before, but it's definitely there in portage 2.0.50.
I have yet to try gentoo, I tend to like having my distro frozen in time on a set of CD's until I'm ready to jump to the next release (Yes, I build my Slackware ISO's from slackware/current)
There is no rule that says you must put (emerge sync && emerge -uD world) &>/dev/null in/etc/cron.hourly. Just don't sync, and the software won't upgrade (or want to upgrade dependencies) when you install stuff. They're also working on an "emerge security" feature reminiscent of tracking a FreeBSD release, with only security updates done.
Sure, like any distro, an install will blow here and there due to dependencies, but, for the most part, I find it rarely happens with Gentoo....and makes the whole process so easy to do.
You have clearly never worked with mod_php. It likes to install a fresh copy over the old one, install both mod_php and CLI php whether you wanted CLI or not, and then warn you... "Due to previous bloopers with PHP and slotting, you may have multiple instances installed..." They're not too previous if the same thing still happens constantly!
Granted it's a worst case, but it is pretty awful (and repeatable, and persistent). Maybe nobody's fixed it because they all commit suicide after looking at the tangle of PHP ebuilds?
"B....only bundled as part of, and for the sole purpose of running, your Programs,...."
Read what you quoted, hmm? $DEVELOPER is granted the right to redistribute Java as part of THEIR Java program; but the people who receive Java in this manner are NOT free to redistribute it. And distributing Java for other reasons, like allowing people to develop in it, is also NOT allowed by that clause. So section B doesn't grant any rights to Linux distros.
I'm not going to comment on section C because I don't have a copy of the README handy. But I suspect if it was legal to distribute Java as-is, then all distros would do it, perhaps under a separate non-free tree. But they don't.
In modern CPUs, IPC depends on the parallelism of the code, hyperthreading, and so on. Take a look at LAME benchmarks sometime: both Intel and AMD CPUs manage to retire the same amount of IPC for that task, so a P4 kicks Athlon butt in direct relation to the clock difference.
Here's a question for you: a 2GHz chip averages 2.5 IPC and a 1GHz chip averages 3.3. Which is faster?
Fire up 'info bash' sometime and have a look around. They split many things according to whether they're Bourne or Bash shell features. I can see where this information would be useful for someone writing portable sh code, but splitting the contents by it is Just Wrong. It should be done by command. Things seem to be split up among too many indices, as well.
The cvs info page is another horridly designed piece of trash. They may have one index, but you have to use the 'l' command a lot since all the commands (at least) are documented several ways in several places, none of which is ever what you wanted.
Try finding "GCC options" in the GCC info pages. They're stealthily hidden away under "Invoking", like it's some sort of deep black art to run a fscking compiler. It wouldn't be too hard to find with recursive search.
Which brings me to my last point: info's navigation is awful. Why can't there be a sane system for it (like pinfo) with Emacs overriding it for people actually using it in Emacs (who would therefore be expected to know the proper commands)?
I'm pointing this out because a lot of times Linux Oldtimers point people to info, probably without realizing exactly how horrid it is. It is not at all accessible to newbies, and probably should just be chucked in its current form.
glxgears is the crappiest benchmark ever. On my box, I see ~256 FPS from software rendering and ~590 FPS in hardware (32-bit display on a Radeon 7500LE.) Yet, despite glxgears being only 2.5x faster, it makes the difference between 1 fps and 30 fps in actual games.
But did you try them on MSN? "X" on there claims popular categories for X are Final Fantasy X, X-men, and "Malcolm X (civil rights activist...", as well as coming up with a list of non-porn results.
Demand for those backwards-compatible chips is huge. To truly replace the architecture, even with a 32-bit-only version of it, you would have to change the boot protocol, OS, and BIOS as well. In a commoditized market, the odds of all that happening, let alone smoothly, are on the order of a snowball's chance in Hell cubed.
In that scenario, the problems would be common knowledge from all the returns of the printer. It's nothing like this ninja technology that watches unsuspecting printouts until it thinks it needs to act.
Should we quit being concerned with spyware, too, since no law says you have to install it? What if there's nothing but spyware, because Palladium has effectively destroyed anti-spyware companies because Microsoft refuses to sign their products? Just because there's no law granting some company or technology a monopoly does not mean it won't happen.
Something that always bothers me about these articles is that they never seem to specify whether the anti-counterfeiting measures pay attention to the final size of the bill. I imagine Photoshop doesn't, though, because until its printed, it does not know what the final size will be.
To those of us who learned M68K assembly with Assempro on our 1337 C= Amiga machines, where everything is done as "src, dest", it's a real PITA to switch to Intel's "dest, src" mode because we accidentally read instructions backwards from time to time. But then again, there may only be one of me.
lea intname, a1 ; pointer to "intuition.library" in a1
move.l ExecBase, a6 ; base address of core library in a6
jsr OldOpenLibrary(a6) ; open library, unversioned
move.l d0, intbase ; store base address of Intuition (high-level graphics)
"...or sneaky (pretend it didn't hear you, so you don't know if it's there at all)."
If a port is stealthed (i.e. using the DROP target) then it doesn't actually hide your system. If there was nothing at that IP address, normally something upstream would send back something like "no route to host" or "host unreachable". A packet to a stealthed port on a live machine just vanishes, but using traceroute can tell people that the network it's on is actually up and reachable. All the stealthing really does is make a portscan take a bit more resources (time or simultaneous connections) because it has to wait for the full timeout.
Let's just skip over whether grinding out code from a spec is creative. Is programming culturally sensitive? If XUL and XPCOM had been written by Indians instead of Netscape (which, judging by the names in Bugzilla, don't seem to be 100% Pure Olde Anglo-Saxon Heritage), would it matter? That is, assuming they kept I18N issues in mind?
You can get a $250 CRT with a D-sub soldered right into it (this was my 17" Optiquest in 2000), or a $500 Mitsubishi LCD (also 17", 2003) with a separate Dsub-to-Dsub cable. If you want DVI for the LCD, it'll set you back an additional $30 for the cable. (Or $70 if you don't comparison shop and just get it from Best Buy.)
I really would rather have a DVI-only LCD with no A/D conversion circuitry in it at all, but not being a designer of such things, I don't know how much it would actually save. Demand probably isn't all that great either.
It's all dependent on the UI. If the sort action is off under View > Arrange Icons >..., of course fewer people are going to find it. If it's sitting right there on/under the toolbar as Sort: [by date |v], then vastly more people will see and use it.
Generally people are intelligent and creative, despite thinking in so many different ways. They just have no concept of what's possible or impossible with computers, nor any idea of how to do some of the possible things, nor a sense of whether they'll "break" it by doing something. All this greatly hinders their proficiency. A good UI design can alleviate a lot of work of memorizing the mapping between task and command sequence.
Besides GNOME Storage that another poster mentioned, segusoLand looks rather interesting as well. I was actually thinking about inventing the same wheel once; I'm glad I didn't bother.
Open source must embrace patents in the same way that the GNU GPL embraces copyright. With this, we could force closed-source development houses to fund open source, because the license is only free to other open-source projects.
What harm would there be in creating such a license? It would give Red Hat a graceful way to keep their promise that their patents would never be enforced against Open Source, and give the rest of the community a collective defensive patent portfolio.
The harm in not creating such a license is pretty clear; the article is one possibility of the extension of current practice. Another more immediate and likely scenario is that OSS could become a de facto R&D lab for Microsoft. We're already seeing that occur with XUL (or libglade, which is pretty similar) and XAML.
(This post based on the ideas of another; I'll drop the link to him so he can elaborate and/or take credit...)
Slashdot Mirror
"I am assuming, or at least it looks like, these tests were run with the default data=journal - so to be fair, they should have been run in data=writeback, or maybe even all three modes. Again, all you have to is specify in /etc/fstab and reboot, no big deal."
And where do you get the idea that this is the default? According to mount(8):
What I really would have liked to see on this benchmark is ext3 on 2.6 with dir_index enabled. (Maybe this would also gain the benefit of the Orlov allocator? I haven't been paying attention to what's been backported.) ...In fact, I would have liked to see this whole thing on 2.6.
"fake a bunch of bad transmits and request resends (does the fax protocol even do that?)"
Yes. In places with crappy lines, it's best to override the speed negotiation and set it to a sane value. (Say 9600 bps if the lines can't handle 14400. These are mildly random numbers because my father does this, not me.) Otherwise the fast line speed, which negotiation accepted because the short test burst worked without error, creates too many retransmits and falls behind on average.
"I see now there are binary packages. ... Can I install and use binary packages all the time?"
Binaries are solely for speeding up installation; you install Stage 3 + GRP (that is, the binaries) and then it acts like a normal source-installed Gentoo system. You can either leave it for a while doing emerge -e world after you finish the install to put your own USE/CFLAGS settings into effect, or just let things get replaced naturally when you update things.
rac argues for this to continue here; I'm not aware of any other information.
I've been thinking of hacking up some sort of tone generator to a Wacom tablet, and mapping X to pitch, Y to timbre (or maybe gestures for things like wah-wah? That axis is undesigned ATM), and pressure to volume. Of course, this is all vaporware still...
From emerge(5): /.
ENVIRONMENT OPTIONS
ROOT = [path]
Use ROOT to specify the target root filesystem to be used for
merging packages or ebuilds. Defaults to
This may have been added recently, since I've never looked for it before, but it's definitely there in portage 2.0.50.
I have yet to try gentoo, I tend to like having my distro frozen in time on a set of CD's until I'm ready to jump to the next release (Yes, I build my Slackware ISO's from slackware/current)
There is no rule that says you must put (emerge sync && emerge -uD world) &>/dev/null in /etc/cron.hourly. Just don't sync, and the software won't upgrade (or want to upgrade dependencies) when you install stuff. They're also working on an "emerge security" feature reminiscent of tracking a FreeBSD release, with only security updates done.
Sure, like any distro, an install will blow here and there due to dependencies, but, for the most part, I find it rarely happens with Gentoo....and makes the whole process so easy to do.
You have clearly never worked with mod_php. It likes to install a fresh copy over the old one, install both mod_php and CLI php whether you wanted CLI or not, and then warn you... "Due to previous bloopers with PHP and slotting, you may have multiple instances installed..." They're not too previous if the same thing still happens constantly!
Granted it's a worst case, but it is pretty awful (and repeatable, and persistent). Maybe nobody's fixed it because they all commit suicide after looking at the tangle of PHP ebuilds?
"B. ...only bundled as part of, and for the sole purpose of running, your Programs,...."
Read what you quoted, hmm? $DEVELOPER is granted the right to redistribute Java as part of THEIR Java program; but the people who receive Java in this manner are NOT free to redistribute it. And distributing Java for other reasons, like allowing people to develop in it, is also NOT allowed by that clause. So section B doesn't grant any rights to Linux distros.
I'm not going to comment on section C because I don't have a copy of the README handy. But I suspect if it was legal to distribute Java as-is, then all distros would do it, perhaps under a separate non-free tree. But they don't.
"Different formats?"
Yeah, just try and sneakernet something on floppy between an Amiga and a PC...
In modern CPUs, IPC depends on the parallelism of the code, hyperthreading, and so on. Take a look at LAME benchmarks sometime: both Intel and AMD CPUs manage to retire the same amount of IPC for that task, so a P4 kicks Athlon butt in direct relation to the clock difference.
Here's a question for you: a 2GHz chip averages 2.5 IPC and a 1GHz chip averages 3.3. Which is faster?
"but:
:\\\); else echo :\\\(; fi\`] "
:\\\); else echo :\\\(; fi\`] "
export PS1="[\$(date \ +%H%M)][\u@\h:\w][\`if [ \$? = 0 ]; then echo
will not since the $? is examing the output of the last 'command', which in the latter case is 'date +%H%M' and not the last command you entered."
Buried somewhere in info is a list of escapes. You want \D.
export PS1="[\D{%H%M}][\u@\h:\w][\`if [ \$? = 0 ]; then echo
Fire up 'info bash' sometime and have a look around. They split many things according to whether they're Bourne or Bash shell features. I can see where this information would be useful for someone writing portable sh code, but splitting the contents by it is Just Wrong. It should be done by command. Things seem to be split up among too many indices, as well.
The cvs info page is another horridly designed piece of trash. They may have one index, but you have to use the 'l' command a lot since all the commands (at least) are documented several ways in several places, none of which is ever what you wanted.
Try finding "GCC options" in the GCC info pages. They're stealthily hidden away under "Invoking", like it's some sort of deep black art to run a fscking compiler. It wouldn't be too hard to find with recursive search.
Which brings me to my last point: info's navigation is awful. Why can't there be a sane system for it (like pinfo) with Emacs overriding it for people actually using it in Emacs (who would therefore be expected to know the proper commands)?
I'm pointing this out because a lot of times Linux Oldtimers point people to info, probably without realizing exactly how horrid it is. It is not at all accessible to newbies, and probably should just be chucked in its current form.
Done. Tell you what--I'll even include the fix in 2.4.25.
glxgears is the crappiest benchmark ever. On my box, I see ~256 FPS from software rendering and ~590 FPS in hardware (32-bit display on a Radeon 7500LE.) Yet, despite glxgears being only 2.5x faster, it makes the difference between 1 fps and 30 fps in actual games.
But did you try them on MSN? "X" on there claims popular categories for X are Final Fantasy X, X-men, and "Malcolm X (civil rights activist...", as well as coming up with a list of non-porn results.
Demand for those backwards-compatible chips is huge. To truly replace the architecture, even with a 32-bit-only version of it, you would have to change the boot protocol, OS, and BIOS as well. In a commoditized market, the odds of all that happening, let alone smoothly, are on the order of a snowball's chance in Hell cubed.
The Internet interpreted the source code as damage and routed around it.
In that scenario, the problems would be common knowledge from all the returns of the printer. It's nothing like this ninja technology that watches unsuspecting printouts until it thinks it needs to act.
Should we quit being concerned with spyware, too, since no law says you have to install it? What if there's nothing but spyware, because Palladium has effectively destroyed anti-spyware companies because Microsoft refuses to sign their products? Just because there's no law granting some company or technology a monopoly does not mean it won't happen.
Something that always bothers me about these articles is that they never seem to specify whether the anti-counterfeiting measures pay attention to the final size of the bill. I imagine Photoshop doesn't, though, because until its printed, it does not know what the final size will be.
To those of us who learned M68K assembly with Assempro on our 1337 C= Amiga machines, where everything is done as "src, dest", it's a real PITA to switch to Intel's "dest, src" mode because we accidentally read instructions backwards from time to time. But then again, there may only be one of me.
lea intname, a1 ; pointer to "intuition.library" in a1
move.l ExecBase, a6 ; base address of core library in a6
jsr OldOpenLibrary(a6) ; open library, unversioned
move.l d0, intbase ; store base address of Intuition (high-level graphics)
"...or sneaky (pretend it didn't hear you, so you don't know if it's there at all)."
If a port is stealthed (i.e. using the DROP target) then it doesn't actually hide your system. If there was nothing at that IP address, normally something upstream would send back something like "no route to host" or "host unreachable". A packet to a stealthed port on a live machine just vanishes, but using traceroute can tell people that the network it's on is actually up and reachable. All the stealthing really does is make a portscan take a bit more resources (time or simultaneous connections) because it has to wait for the full timeout.
Let's just skip over whether grinding out code from a spec is creative. Is programming culturally sensitive? If XUL and XPCOM had been written by Indians instead of Netscape (which, judging by the names in Bugzilla, don't seem to be 100% Pure Olde Anglo-Saxon Heritage), would it matter? That is, assuming they kept I18N issues in mind?
You can get a $250 CRT with a D-sub soldered right into it (this was my 17" Optiquest in 2000), or a $500 Mitsubishi LCD (also 17", 2003) with a separate Dsub-to-Dsub cable. If you want DVI for the LCD, it'll set you back an additional $30 for the cable. (Or $70 if you don't comparison shop and just get it from Best Buy.)
I really would rather have a DVI-only LCD with no A/D conversion circuitry in it at all, but not being a designer of such things, I don't know how much it would actually save. Demand probably isn't all that great either.
It's all dependent on the UI. If the sort action is off under View > Arrange Icons > ..., of course fewer people are going to find it. If it's sitting right there on/under the toolbar as Sort: [by date |v], then vastly more people will see and use it.
Generally people are intelligent and creative, despite thinking in so many different ways. They just have no concept of what's possible or impossible with computers, nor any idea of how to do some of the possible things, nor a sense of whether they'll "break" it by doing something. All this greatly hinders their proficiency. A good UI design can alleviate a lot of work of memorizing the mapping between task and command sequence.
Besides GNOME Storage that another poster mentioned, segusoLand looks rather interesting as well. I was actually thinking about inventing the same wheel once; I'm glad I didn't bother.
Open source must embrace patents in the same way that the GNU GPL embraces copyright. With this, we could force closed-source development houses to fund open source, because the license is only free to other open-source projects.
What harm would there be in creating such a license? It would give Red Hat a graceful way to keep their promise that their patents would never be enforced against Open Source, and give the rest of the community a collective defensive patent portfolio.
The harm in not creating such a license is pretty clear; the article is one possibility of the extension of current practice. Another more immediate and likely scenario is that OSS could become a de facto R&D lab for Microsoft. We're already seeing that occur with XUL (or libglade, which is pretty similar) and XAML.
(This post based on the ideas of another; I'll drop the link to him so he can elaborate and/or take credit...)