Communist Soviet Union opposed fascism (Molotov-Riebentropp notwithstanding, they supported republican Spain). Not all authoritarian regime are fascists, just like not all diseases are lupus.
This has nothing to do with fascism. The problem with fascism wasn't censorship. Censorship is bad, fascism included censorship as a matter of course, but it's not what was particularly bad about fascism. Soviet Russia wasn't fascist. It was bad too, just not in the same way. Today the United States are much closer to fascism than Australia, yet they enjoy incomparable freedom of speech. Militarization of the economy, dubious appeals to patriotism, booming prison population, the collusion between corporate interests and government, that's fascist-ish. Censorship, that's what you find in China, which is not nearly as bad as the US in the areas I just listed (but by no means any better overall, don't get me wrong.)
On a random blog, which was rather legit, I ended up getting redirected to this page:
Here's the link: hxxp://gowithscan.com/?uid=13100 (malware! warning!)
It appeared to scan my Windows and find multiple vulnerabilities. Good thing I'm running Linux. Then it proceeded to obnoxiously pop up JS alerts and have me download an install.exe. Major antivirus couldn't find anything wrong with it. I have the file if anyone is interested (submitted it to clamav.org too).
Early Microsoft web frameworks, circa 1998, generated code so ugly it should qualify as crime against humanity. The stench has contaminated many enterprises, which are stuck with those unmaintainable festering sores. Looking at the javscript those beasts produced is fascinating; they could put ";" in places you never expected.
In databases such as PostgreSQL, nothing is guaranteed to be recorded until a transaction has been committed *and* the DB has replied positively to the commit request. The application should not assume that the operation is succesful until then. There is a nice tuning option in PG (and other high end DBs I suppose) where you can tell it to wait a number of milliseconds on every query so that it has a chance to do just one fsync for several transactions. It slows down sequential operations with no load, so you might want to do disable this when doing certain maintenance operations (or you can arrange for those opeartions to be part of one large transaction instead of several small ones, such as with auto commit). In production and under load, however, this improves overall throughput dramatically.
High end disks and modern FS (such as ext{3,4}) support write barriers to enforce ordering in critical sections. I suggest using high end disks on critical systems.
If I understand things correctly, while there is a significant hit when writing lots of small files and fsyncing after each of them, you take a hit except when you're journalling data. But in that case you take a hit when writing big files, since data has to be written twice (first in the journal, then when the journal is flushed).
They use battery-backed cache. The data is written and stored, just not on disk yet. The battery is supposed to last a couple days. If you need to shut a server down for longer than that... well just don't yank the power cord, perform a clean shutdown.
Fedora XGuest is much better than anything you can find on Windows:
Creates xguest user as a locked down user
Installing this package sets up the xguest user to be used as a temporary account to switch to or as a kiosk user account. The account is disabled unless SELinux is in enforcing mode. The user is only allowed to log in via gdm. The home and temporary directories of the user will be polyinstantiated and mounted on tmpfs.
It might not do exactly what you want. No biggie, just create a custom policy to allow what you want, there are GUI tools to help with that.
It's not been implemented that way as far as I know, but you can use SELinux to lock a user down as much as you want. For example Fedora has an "xguest" package where a user type is created that can only do a limited set of things, such as browse the web. Can't even connect to local daemons (except those expressly approved) via the loopback interface. Can't create files but where it is specified. And so on and so forth.
You know what, if one of my lusers manages to write python scripts to play solitaire, I'll tell you what, I'll hire them right away in the sysadmin team.
He proposes using the following format (I believe he uses it in qmtp) instead of ^M terminated lines: first write the number of characters in ASCII digits, followed by a space to indicate the end of the number, followed by that number of character, and a ^M. In any case, you have to take care of quite a few things when serializing data for writing to disk / transmission.
16 bit is too small in some cases? Well then rewrite the library for 32 bits.
Seriously.
Thing is, people rewrote the library that did not exist a thousand times anyway, and more often than not used C strings when they shouldn't have, resulting in countless disastrous buffer overflows.
Oh and btw, there is a simple solution to this non-problem:
for Pascal type strings in C. The fact that null-terminated strings existed wasn't the problem, they make some sense in some respects, such as when you want to pass text of arbitrary length. But the real problem, the real bug was not having a standard way of doing real strings in C. Everybody had to do it himself, poorly. Had there been a standard, no matter how poor, it would have been a starting point to do something better if needed, and would have been better anyway for many uses than C strings. It would have avoided MANY vulnerabilities from common software.
Even if you suscribe to the pro tax cut creed in a normal environment where the economy is working fine, here they are absolutely counter productive. People are rightly afraid of the near future situation. In this situation, most companies / individuals expect their revenues to decline, so they resp. cut wages expenditures and cut spending, which results in people having less money to spend, therefore companies make even less money and cut jobs even further.
It's a vicious circle and you have to break it. If you give people tax cuts, they will just hoard it, because it's the rational thing to do for each economic actor, individually, even if it's negative as a whole.
Slashdot Mirror
Communist Soviet Union opposed fascism (Molotov-Riebentropp notwithstanding, they supported republican Spain).
Not all authoritarian regime are fascists, just like not all diseases are lupus.
This has nothing to do with fascism. The problem with fascism wasn't censorship. Censorship is bad, fascism included censorship as a matter of course, but it's not what was particularly bad about fascism. Soviet Russia wasn't fascist. It was bad too, just not in the same way.
Today the United States are much closer to fascism than Australia, yet they enjoy incomparable freedom of speech.
Militarization of the economy, dubious appeals to patriotism, booming prison population, the collusion between corporate interests and government, that's fascist-ish.
Censorship, that's what you find in China, which is not nearly as bad as the US in the areas I just listed (but by no means any better overall, don't get me wrong.)
On a random blog, which was rather legit, I ended up getting redirected to this page:
Here's the link: hxxp://gowithscan.com/?uid=13100 (malware! warning!)
It appeared to scan my Windows and find multiple vulnerabilities. Good thing I'm running Linux. Then it proceeded to obnoxiously pop up JS alerts and have me download an install.exe. Major antivirus couldn't find anything wrong with it. I have the file if anyone is interested (submitted it to clamav.org too).
Early Microsoft web frameworks, circa 1998, generated code so ugly it should qualify as crime against humanity. The stench has contaminated many enterprises, which are stuck with those unmaintainable festering sores.
Looking at the javscript those beasts produced is fascinating; they could put ";" in places you never expected.
In databases such as PostgreSQL, nothing is guaranteed to be recorded until a transaction has been committed *and* the DB has replied positively to the commit request. The application should not assume that the operation is succesful until then.
There is a nice tuning option in PG (and other high end DBs I suppose) where you can tell it to wait a number of milliseconds on every query so that it has a chance to do just one fsync for several transactions. It slows down sequential operations with no load, so you might want to do disable this when doing certain maintenance operations (or you can arrange for those opeartions to be part of one large transaction instead of several small ones, such as with auto commit). In production and under load, however, this improves overall throughput dramatically.
High end disks and modern FS (such as ext{3,4}) support write barriers to enforce ordering in critical sections.
I suggest using high end disks on critical systems.
If I understand things correctly, while there is a significant hit when writing lots of small files and fsyncing after each of them, you take a hit except when you're journalling data. But in that case you take a hit when writing big files, since data has to be written twice (first in the journal, then when the journal is flushed).
They use battery-backed cache. The data is written and stored, just not on disk yet. The battery is supposed to last a couple days. If you need to shut a server down for longer than that ... well just don't yank the power cord, perform a clean shutdown.
The filesystem doesn't guarantee anything is written until you've called fsync and it has returned.
/ yet another smug, uninfected Linux user.
puppet/modules/nosolitaire/init.pp :
class nosolitaire {
package { "nosolitaire":
name => "solitaire",
ensure => absent,
}
}
Fedora XGuest is much better than anything you can find on Windows:
It might not do exactly what you want. No biggie, just create a custom policy to allow what you want, there are GUI tools to help with that.
It's not been implemented that way as far as I know, but you can use SELinux to lock a user down as much as you want.
For example Fedora has an "xguest" package where a user type is created that can only do a limited set of things, such as browse the web. Can't even connect to local daemons (except those expressly approved) via the loopback interface. Can't create files but where it is specified. And so on and so forth.
What else do you need?
You know what, if one of my lusers manages to write python scripts to play solitaire, I'll tell you what, I'll hire them right away in the sysadmin team.
puppet + custom RPM repository >>> windows
I'm not sure what you're trying to say, though.
is even better, can even get you through some non-free hotspots, and it's hardly ever blocked where most other things are.
offending line.
Step 2: press "DEL".
There is no step 3.
Most people do.
There will always be conservative types whining about how things were so good back in the days.
Get over it.
He proposes using the following format (I believe he uses it in qmtp) instead of ^M terminated lines: first write the number of characters in ASCII digits, followed by a space to indicate the end of the number, followed by that number of character, and a ^M.
In any case, you have to take care of quite a few things when serializing data for writing to disk / transmission.
the anonymous idiot looks at the finger.
16 bit is too small in some cases? Well then rewrite the library for 32 bits.
Seriously.
Thing is, people rewrote the library that did not exist a thousand times anyway, and more often than not used C strings when they shouldn't have, resulting in countless disastrous buffer overflows.
Oh and btw, there is a simple solution to this non-problem:
#define STRING(typ) typ##string
#define DEFINESTRING(typ) typedef struct { \
typ len; \
char c[0]; \
} STRING(typ);
DEFINESTRING(int);
DEFINESTRING(long);
DEFINESTRING(uint128);
int println(STRING(int) a) {
for Pascal type strings in C. The fact that null-terminated strings existed wasn't the problem, they make some sense in some respects, such as when you want to pass text of arbitrary length. But the real problem, the real bug was not having a standard way of doing real strings in C. Everybody had to do it himself, poorly. Had there been a standard, no matter how poor, it would have been a starting point to do something better if needed, and would have been better anyway for many uses than C strings. It would have avoided MANY vulnerabilities from common software.
Even if you suscribe to the pro tax cut creed in a normal environment where the economy is working fine, here they are absolutely counter productive. People are rightly afraid of the near future situation. In this situation, most companies / individuals expect their revenues to decline, so they resp. cut wages expenditures and cut spending, which results in people having less money to spend, therefore companies make even less money and cut jobs even further.
It's a vicious circle and you have to break it. If you give people tax cuts, they will just hoard it, because it's the rational thing to do for each economic actor, individually, even if it's negative as a whole.
j/k.
I see you're still too modest to plug your own awesome code, so I'm doing it for you.