Browsers should allow only session cookies by default, with a way to allow setting a particular site's cookies to permanent.
FireFox+Cookie Monster is almost there -- sadly, the cookie's original timeout is lost so permanency goes in only after the next time that cookie is set.
Chrome is a bad joke -- even 9 allows at most a blanket "all cookies permanent/all cookies session" setting, even Netscape 2.0 at the dawn of time had an "ask" option which, while a bit cumbersome, let you control this.
Many more than just him are fed up with the patent nonsense.
I for one don't do anything but kvetch, but then, it's not like submitting a/. story is much more work. And the article (judging by the blurb, of course) seems to be pretty interesting.
This will break only if IPv6 connection cannot be established and it fails to report that. A vast majority of OSes today are IPv6-capable, and a majority of machines don't have IPv6 connectivity -- they just get to the local machine or the local router in a fraction of a milisecond, then retry on IPv4.
The two sources you point are not broken in any way -- they will just reveal such brokenness. Using 6to4 is perfectly valid, even if not as fast at the moment.
Or, alternatively, put a chip with four of those instead of one. Being faster with one means you'd suddenly be more than four times faster than x86 at the same power.
When cell phones first arrived, receiving calls cost money
Er, receiving? Which phone company dared to pull that? That's the most broken idea related to phones I've heard in a while. Do you mean that people who accepted such shitty plans had to pay whenever some chatty person or a telemarketer called them -- without having any control on how much they get called?
My point is that Wikipedia has a good deal more facts per article, the extra space is not used merely to be more wordy. The chance that fact X is wrong is smaller by far on Wikipedia, thanks to many eyeballs, some of whom are actually competent.
Uhm, you can find "the point of GPL" in its preamble: "The licenses for most software and other practical works are designed to take away your freedom to share and change the works. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change all versions of a program--to make sure it remains free software for all its users.".
The GPL places just one restriction: it restricts you from adding restrictions to the work (and as a pre-requisite, from restricting others to access the source). This fixes a fatal flaw in BSD: in BSD, anyone can place copyright restrictions back.
That's the error rate per article, not for the amount of information. Which, as I recall, the very same study found to be 8x larger per article in Wikipedia, making the error rate tremendously lower.
The whole point of GPL is to make it impossible to put copyright restrictions on that piece of software. And unlike BSD/MIT, it cannot be trivially worked around.
If copyright wasn't there, GPL wouldn't be necessary.
For my comment about battles -- dying in a real war is more important than modding a silly console, but the comparison of these incidents ("battles" as I said) doesn't say anything issues they're about ("wars"). An oppressed peace works only short term.
Imagine two groups of dissidents in China: a thousand-strong group with guns and a few tanks, and a lone guy who manages to send a message to half the population -- which one would their government fear more?
This is a very insignificant battle in the war against culture.
But that war is more important IMO than a war against mere lives. Having peace won't get you free dissemination of ideas, free dissemination of ideas is a doom to oppressive rulers.
Following the "science fault" route: * what does the article's author do for a living? Falsifying of search return. * does the site that published this study have ties to the "winner"? It's among their "sponsors and partners" page.
Somehow, nearly every time you find an "independent" study giving sensational results, it is sponsored by someone with a vested interest in those results.
TCP wouldn't work, but replacing it is damn straightforward[1]. With that done, even regular http would work, although it would suck due to needing several round trips for referenced CSS/images/subframes/whatnot. The latter is also quite obvious -- many of us could whip up a primitive but working proxy in an hour.
If bandwidth isn't a concern, it might be better to do a whole-site wget rip and send it in one go.
[1]. Nearly all complexity in TCP is due to handshaking, retransmissions and adjusting the window. With no handshaking possible and terabyte windows, you can throw most of that away. Adding generous Reed-Solomon codes over the whole message and possibly retransmitting just damaged blocks would be obvious goodies to add, but that's still nothing compared to the mess that current TCP is.
Unless your routing is terminally broken, the IPv6 connection attempt fails immediately. You'd need to have an useless box that not only fails to handle IPv6 but also silently drops the packets.
For that NAT idea -- one of the core purposes of the IPv6 upgrade is to get rid of NAT. If you'd inflict a RFC1918-esque subnet on yourself, your machine cannot be addressed anymore.
I doubt the US is that different -- in Poland you have large swatches of villages where no one works and every single denizen abuses some sort of benefits -- be they unemployment, fake disability or anything you can get. People who try to actually work there are mocked as being resourceless and naive. I'm told that at least in the UK this is rampant as well.
Naturally, this may be a much needed safety net, but kicking off abusers of benefits -- and especially, officials who help their locals obtaining them -- is an important duty towards us taxpayers.
"Nobody contributed code in 10 years"? No, mtr is alive and well (and shipped in default install of all distros), they're either lying or mean that they took a 10 years old release.
Uhm, mtr "abandoned"?? Most distributions seem to be removing old staples like traceroute from default installs, having replaced them with mtr, which certainly doesn't seem to be pining for the fjords.
mtr does everything traceroute did, except instead of doing every probe only after the previous one has timed out, it sends them all concurrently, a packet every a fraction of second, providing a nifty real-time display of statistics.
It's not a fun thing to write on small keyboards, but I find it merely a notch worse than a laptop. Once you get below the comfortable size, the difference seems minor. And n900 can be held in a pocket rather than a car's trunk.
Of course it's better to code at a desktop, it goes without saying.
While n900 beats anything Droid in customizability, the default keymap is worse than abysmal. Please use for example mine, save it as/usr/share/X11/xkb/symbols/nokia_vndr/rx-51; the new assignments use Fn-Shift (or Fn where it's unused).
Having no basic symbols like |, [, ], , {, }, % or ~, or keys like PgUp, PgDn or Esc makes any Unix administration or programming a bad joke. Having to request an on-screen keyboard for those is unacceptable -- it's not an iToy! Fortunately, we can fix it.
You want cp --reflink=auto instead, so it works even if you issue the command across filesystems or on non-btrfs. There's no reason not to plop alias cp='cp --reflink=auto' into your.bashrc.
Slashdot Mirror
Anything called "memory", even in humans, is volatile.
For permanence, you'd want "clay tablets" or newer technology of that kind.
Browsers should allow only session cookies by default, with a way to allow setting a particular site's cookies to permanent.
FireFox+Cookie Monster is almost there -- sadly, the cookie's original timeout is lost so permanency goes in only after the next time that cookie is set.
Chrome is a bad joke -- even 9 allows at most a blanket "all cookies permanent/all cookies session" setting, even Netscape 2.0 at the dawn of time had an "ask" option which, while a bit cumbersome, let you control this.
Rule #13: Do unto others.
Many more than just him are fed up with the patent nonsense.
I for one don't do anything but kvetch, but then, it's not like submitting a /. story is much more work. And the article (judging by the blurb, of course) seems to be pretty interesting.
Or, you realize that e-mail was never designed to lug large binary files around and pass the test programs over http.
This will break only if IPv6 connection cannot be established and it fails to report that. A vast majority of OSes today are IPv6-capable, and a majority of machines don't have IPv6 connectivity -- they just get to the local machine or the local router in a fraction of a milisecond, then retry on IPv4.
The two sources you point are not broken in any way -- they will just reveal such brokenness. Using 6to4 is perfectly valid, even if not as fast at the moment.
Remember that these are not Googlers. While Google picks the cream of the crop, Facebook uses a horde of PHP monkeys.
Or, alternatively, put a chip with four of those instead of one. Being faster with one means you'd suddenly be more than four times faster than x86 at the same power.
When cell phones first arrived, receiving calls cost money
Er, receiving? Which phone company dared to pull that? That's the most broken idea related to phones I've heard in a while.
Do you mean that people who accepted such shitty plans had to pay whenever some chatty person or a telemarketer called them -- without having any control on how much they get called?
My point is that Wikipedia has a good deal more facts per article, the extra space is not used merely to be more wordy. The chance that fact X is wrong is smaller by far on Wikipedia, thanks to many eyeballs, some of whom are actually competent.
Uhm, you can find "the point of GPL" in its preamble: "The licenses for most software and other practical works are designed to take away your freedom to share and change the works. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change all versions of a program--to make sure it remains free software for all its users.".
The GPL places just one restriction: it restricts you from adding restrictions to the work (and as a pre-requisite, from restricting others to access the source). This fixes a fatal flaw in BSD: in BSD, anyone can place copyright restrictions back.
That's the error rate per article, not for the amount of information. Which, as I recall, the very same study found to be 8x larger per article in Wikipedia, making the error rate tremendously lower.
The whole point of GPL is to make it impossible to put copyright restrictions on that piece of software. And unlike BSD/MIT, it cannot be trivially worked around.
If copyright wasn't there, GPL wouldn't be necessary.
For my comment about battles -- dying in a real war is more important than modding a silly console, but the comparison of these incidents ("battles" as I said) doesn't say anything issues they're about ("wars"). An oppressed peace works only short term.
Imagine two groups of dissidents in China: a thousand-strong group with guns and a few tanks, and a lone guy who manages to send a message to half the population -- which one would their government fear more?
This is a very insignificant battle in the war against culture.
But that war is more important IMO than a war against mere lives. Having peace won't get you free dissemination of ideas, free dissemination of ideas is a doom to oppressive rulers.
Following the "science fault" route:
* what does the article's author do for a living? Falsifying of search return.
* does the site that published this study have ties to the "winner"? It's among their "sponsors and partners" page.
Somehow, nearly every time you find an "independent" study giving sensational results, it is sponsored by someone with a vested interest in those results.
TCP wouldn't work, but replacing it is damn straightforward[1]. With that done, even regular http would work, although it would suck due to needing several round trips for referenced CSS/images/subframes/whatnot. The latter is also quite obvious -- many of us could whip up a primitive but working proxy in an hour.
If bandwidth isn't a concern, it might be better to do a whole-site wget rip and send it in one go.
[1]. Nearly all complexity in TCP is due to handshaking, retransmissions and adjusting the window. With no handshaking possible and terabyte windows, you can throw most of that away. Adding generous Reed-Solomon codes over the whole message and possibly retransmitting just damaged blocks would be obvious goodies to add, but that's still nothing compared to the mess that current TCP is.
Unless your routing is terminally broken, the IPv6 connection attempt fails immediately. You'd need to have an useless box that not only fails to handle IPv6 but also silently drops the packets.
For that NAT idea -- one of the core purposes of the IPv6 upgrade is to get rid of NAT. If you'd inflict a RFC1918-esque subnet on yourself, your machine cannot be addressed anymore.
I doubt the US is that different -- in Poland you have large swatches of villages where no one works and every single denizen abuses some sort of benefits -- be they unemployment, fake disability or anything you can get. People who try to actually work there are mocked as being resourceless and naive. I'm told that at least in the UK this is rampant as well.
Naturally, this may be a much needed safety net, but kicking off abusers of benefits -- and especially, officials who help their locals obtaining them -- is an important duty towards us taxpayers.
"Nobody contributed code in 10 years"? No, mtr is alive and well (and shipped in default install of all distros), they're either lying or mean that they took a 10 years old release.
Uhm, mtr "abandoned"?? Most distributions seem to be removing old staples like traceroute from default installs, having replaced them with mtr, which certainly doesn't seem to be pining for the fjords.
mtr does everything traceroute did, except instead of doing every probe only after the previous one has timed out, it sends them all concurrently, a packet every a fraction of second, providing a nifty real-time display of statistics.
Don't forget fbcdn.net, blocking both gives a great boost to your page load speeds. They're as bad as the advertising scum.
It's not a fun thing to write on small keyboards, but I find it merely a notch worse than a laptop. Once you get below the comfortable size, the difference seems minor. And n900 can be held in a pocket rather than a car's trunk.
Of course it's better to code at a desktop, it goes without saying.
While n900 beats anything Droid in customizability, the default keymap is worse than abysmal. Please use for example mine, save it as /usr/share/X11/xkb/symbols/nokia_vndr/rx-51; the new assignments use Fn-Shift (or Fn where it's unused).
Having no basic symbols like |, [, ], , {, }, % or ~, or keys like PgUp, PgDn or Esc makes any Unix administration or programming a bad joke. Having to request an on-screen keyboard for those is unacceptable -- it's not an iToy! Fortunately, we can fix it.
Well, Open/LibreOffice works just fine on ARM today.
You want cp --reflink=auto instead, so it works even if you issue the command across filesystems or on non-btrfs. There's no reason not to plop alias cp='cp --reflink=auto' into your .bashrc.