-ffast-math on gcc only has it do what both Intel and MS are already doing - using the bloody fp unit instead of software emulation!!!
This is Obviously Wrong. The default -fno-fast-math behavior is to use the FP unit, but strategically use fstp/fld (store-and-pop/load) pairs to cut the accuracy down to the actual number of needed bits (from the 80-bit internal representation of the 80x87). This makes the final value reflect what a naïve program expects, if it breaks on slight differences in rounding and comes from an architecture where e.g. 32-bit float precision means 32-bit float precision.
Software emulation of the x87 (in case the app is running on, say, a naked 386 or a 486SX) is entirely an OS and library issue. The resulting app-level assembly code is identical.
The U.S. was settled in 1607. By 1776, much of the population of the colonies had already been isolated from British dialect changes for quite some time.
Actually, no, a no-execute bit doesn't prevent buffer overflow attacks. It just stops an attacker from directly executing code that he/she left lying around on the stack or heap. However, return-into-libc exploits are still possible, and although they're a bit trickier to write, they aren't really mind-blowingly difficult, either.
You can just allocate a data page, decrypt to that, then change the protections on the data page to turn it to code (mprotect(3) in *nix, VirtualProtectEx in Windows). Y'know, just like the hoops a JIT has to jump through. Essentially, you're using the same argument as people who say, "Man, this NT stuff is harder than 95. I'll just grant Full Access to Everyone in the installer."
Actually, no. I was suggesting that IE's cache would appear to either (1) bloat all running programs embedding IE, or (2) not show up. Some experimental testing under Win2K says that my guess was wrong -- I dusted IE off and tried some web surfing, and EXPLORER.EXE didn't grow in Task Manager while IEXPLORE.EXE did.
Besides VMWare (which is a solid, if slow, piece of software), you also have the options of Knoppix (a Debian based distro that runs entirely off a bootable CD) and Cooperative Linux (a beta-but-working project to run Linux as a device driver inside a Windows 2000/XP kernel). I haven't tried out Co-Linux (yet), but it sounds like something that takes some previous Linux experience to set up, so I'd recommend Knoppix or VMware.
It's not a secret API. It's MSHTML.DLL, which EXPLORER.EXE (since Win95+IE4 or Win98) and IEXPLORE.EXE both use. It probably uses (documented, non-secret) APIs to create shared r/w data pages for an interprocess in-memory cache. (And, to be fair, if you were writing an embeddable shared-object web browser control meant to be part of 20 apps at once, all owned by the same user, why wouldn't you?)
Okay, I'm going to give you the benefit of a doubt and assume that you're as free of anti-gay bias as you claim. However, you're ignoring the cardinal rule of language -- words have the meaning they do because of mutual agreement between speaker and listener. Go out and ask ten people, "Do you think an average gay person would be insulted by the word 'fag'?" I'm willing to bet that a solid majority would say yes -- definitely in a random sample, but probably even if you just poll your friends and family. So, the people that you are speaking to hear you say "fag", and they assume that you don't care if you insult gay people. And why don't you care? They'll assume you dislike gay people. And, if you dislike gay people, why shouldn't they? Peer pressure is the mechanism of social change at the individual level -- and every time you choose to use that word, the rate of change slows down, just a bit.
And that's not even touching on how it affects gay people who overhear you say it. Despite years of trying to train myself out of it, "fag" is still one of the few triggers words I have that makes me punch through walls in lieu of someone's face. To be honest, I'm surprised I've never broken my hand.
Here in the US, "queer" is worse when used offensively than "gay", but nowhere near as strong as "fag"/"faggot". It's been semi-adopted as an umbrella term to replace the ever-growing and unwieldly GLBTQ acronym (Gay, Lesbian, Bisexual, Transgender (of which transsexuals are only one part), Questioning). The Straight Person's Guide To Gay Etiquette has a chapter discussing language usage that touches on the subject.
There's a very big flaw in your example. The two meanings of "straight" are both... "positive" is the wrong word, how about "non-insulting". However, the new bad!gay meaning was created *specifically* to reinforce the idea that homosexual!gay was also wrong. Every time someone uses it that way, it strengthens the idea that the two are related -- it's basically the opposite of the "reclaiming" process that has (finally) made "queer" and "dyke" acceptable under certain circumstances, which was accomplished by linking the words with new, positive meanings.
I'd have to stick up for them against calling them "homophobic", because homophobia really implies an intent to harm. Most of them are just mimicking their peers, and while the practice was certainly started by actual homophobes, they themselves aren't guilty of it.
HOWEVER
Just because they're wallowing in ignorance doesn't mean that it's not (a) insulting, and (b) fostering an anti-gay atmosphere. If someone points it out to them, they should make an attempt to stop (with allowances for habit). If they refuse, then they're being outright assholes. I mean, really, it's depressing enough reading the news without being reminded that a statistically significant portion of the population thinks us gay people are sub-human. Every time I hear someone use "gay" as an insult, and they don't know that I am, I have to mentally stop what I'm doing and figure out if that person was merely "joking" or if they fall into the category of people that would ostracise me, foam at the mouth and rant, (attempt to) beat me up, or worse if they found out. It gets fucking tiring after a while.
The concept behind Habeas is that the poem is copyrighted, and anyone who attaches the poem to their mail without permission (e.g. when spamming) is violating copyright law and open to a lawsuit. However, since the legal system is so glacial, Habeas is getting abused so heavily these days that it's pretty much useless.
To make SpamAssassin ignore Habeas entirely, edit/etc/mail/spamassassin/local.cf (or your system's equivalent) and add the line score HABEAS_SWE 0 to the end.
Re:What do...
on
Real's Reality
·
· Score: 4, Informative
My Windows software uses the "deprecated" INI files in the executable directory instead of the super-bloated registry...
Eep. Stop right there, you're standing on a landmine. INI files on a modern system should be put in the per-user "Application Data" folder. You can retrieve its path using SHGetSpecialFolderPath (95/IE 4.0 or 98+) or SHGetFolderPath (98/IE 5.0 or 98SE+) using the CSIDL_APPDATA constant. If you've fail to do this, your app is currently giving ulcers to some innocent admin of WinNT-family boxes who now has to manually add extra NTFS permissions for the Everyone group to your app's install folder. Your app also doesn't work correctly with multiple users (all users share the same settings), under roaming profiles (settings are per-machine, not per-user), or running off a network drive (Ha! Like *your* app deserves chmod a+w on the Samba server!). As an added bonus, your app may stop working under XP SP2 (or after some Critical Update in the unspecified future) and almost certainly will be b0rked by the time Longhorn comes out (if MS isn't a completely lost cause, they'll have stopped making users Admin by default by that timeframe).
Actually, both. Tolkien based Quenya on Finnish, and Sindarin on Welsh. Which should be very confusing from a linguistic evolutionary standpoint, but works surprisingly well since he didn't borrow any Welsh vocabulary.
In my experience, repeatable crashes in Windows from programs doing non-crash-inducing things are generally caused by (1) flakey video drivers, (2) other flakey drivers, (3) bad RAM, (4) bad video hardware. If Firefox runs on thousands upon thousands of computers, yet crashes only on yours...
The other filters are still very useful because they allow training an empty Bayesian filter -- very spammy and very hammy messages are automatically used as training material. Besides, the DNS blacklist checks catch nearly as much spam as the Bayesian filter, at least for me.
It's not navigating, it's just downloading the file (to RAM). It might sound a bit pedantic at first, but it doesn't try to render the contents, so there's no risk of e.g. exploits against the HTML render engine.
As others mentioned, each serial line is independently clocked, so it doesn't matter if the timing gets out of sync. Theoretically, they could even run at different speeds (but practically speaking there'd be no reason to).
Actually, that bit of code just downloads the malicious.EXE. It's a bit dodgy that it's allowed to do it automatically (after all, it could be asking for http://spy.malware.com/cgi-bin/report?firstname=Jo hn&lastname=Doe&underwear_type=boxers...), but it's not an instant security breach itself. The actual bug is...
s.SaveToFile("C:\\Program Files\\Windows Media Player\\wmplayer.exe",2);
...which overwrites Media Player with the downloaded malware using ADODB.Stream (which probably never should have been enabled as a trusted ActiveX control in the first place, and certainly shouldn't be automatically overwriting files without user intervention).
Re:Similiaries to Netscape vs MS not unfounded
on
Google v. Microsoft
·
· Score: 1
Yeah, I used Netscape 4.77 up until Mozilla got to 1.0, actually. And the crashes were annoying.
You waited that long? I switched when Mozilla hit Milestone 18, way back in late '00. Netscape 4 was just awful, despite being faster on my old 300MHz system at the time.
Slashdot Mirror
-ffast-math on gcc only has it do what both Intel and MS are already doing - using the bloody fp unit instead of software emulation!!!
This is Obviously Wrong. The default -fno-fast-math behavior is to use the FP unit, but strategically use fstp/fld (store-and-pop/load) pairs to cut the accuracy down to the actual number of needed bits (from the 80-bit internal representation of the 80x87). This makes the final value reflect what a naïve program expects, if it breaks on slight differences in rounding and comes from an architecture where e.g. 32-bit float precision means 32-bit float precision.
Software emulation of the x87 (in case the app is running on, say, a naked 386 or a 486SX) is entirely an OS and library issue. The resulting app-level assembly code is identical.
Or quadratically, if the distribution is from a line instead of a point....
Touché. :-)
The reason wireless electricity is a problem is that as distance increases linearly, power drops off exponentially
Actually, cubically.
The U.S. was settled in 1607. By 1776, much of the population of the colonies had already been isolated from British dialect changes for quite some time.
Actually, no, a no-execute bit doesn't prevent buffer overflow attacks. It just stops an attacker from directly executing code that he/she left lying around on the stack or heap. However, return-into-libc exploits are still possible, and although they're a bit trickier to write, they aren't really mind-blowingly difficult, either.
You can just allocate a data page, decrypt to that, then change the protections on the data page to turn it to code (mprotect(3) in *nix, VirtualProtectEx in Windows). Y'know, just like the hoops a JIT has to jump through. Essentially, you're using the same argument as people who say, "Man, this NT stuff is harder than 95. I'll just grant Full Access to Everyone in the installer."
Actually, no. I was suggesting that IE's cache would appear to either (1) bloat all running programs embedding IE, or (2) not show up. Some experimental testing under Win2K says that my guess was wrong -- I dusted IE off and tried some web surfing, and EXPLORER.EXE didn't grow in Task Manager while IEXPLORE.EXE did.
Besides VMWare (which is a solid, if slow, piece of software), you also have the options of Knoppix (a Debian based distro that runs entirely off a bootable CD) and Cooperative Linux (a beta-but-working project to run Linux as a device driver inside a Windows 2000/XP kernel). I haven't tried out Co-Linux (yet), but it sounds like something that takes some previous Linux experience to set up, so I'd recommend Knoppix or VMware.
It's not a secret API. It's MSHTML.DLL, which EXPLORER.EXE (since Win95+IE4 or Win98) and IEXPLORE.EXE both use. It probably uses (documented, non-secret) APIs to create shared r/w data pages for an interprocess in-memory cache. (And, to be fair, if you were writing an embeddable shared-object web browser control meant to be part of 20 apps at once, all owned by the same user, why wouldn't you?)
Okay, I'm going to give you the benefit of a doubt and assume that you're as free of anti-gay bias as you claim. However, you're ignoring the cardinal rule of language -- words have the meaning they do because of mutual agreement between speaker and listener. Go out and ask ten people, "Do you think an average gay person would be insulted by the word 'fag'?" I'm willing to bet that a solid majority would say yes -- definitely in a random sample, but probably even if you just poll your friends and family. So, the people that you are speaking to hear you say "fag", and they assume that you don't care if you insult gay people. And why don't you care? They'll assume you dislike gay people. And, if you dislike gay people, why shouldn't they? Peer pressure is the mechanism of social change at the individual level -- and every time you choose to use that word, the rate of change slows down, just a bit.
And that's not even touching on how it affects gay people who overhear you say it. Despite years of trying to train myself out of it, "fag" is still one of the few triggers words I have that makes me punch through walls in lieu of someone's face. To be honest, I'm surprised I've never broken my hand.
Here in the US, "queer" is worse when used offensively than "gay", but nowhere near as strong as "fag"/"faggot". It's been semi-adopted as an umbrella term to replace the ever-growing and unwieldly GLBTQ acronym (Gay, Lesbian, Bisexual, Transgender (of which transsexuals are only one part), Questioning). The Straight Person's Guide To Gay Etiquette has a chapter discussing language usage that touches on the subject.
There's a very big flaw in your example. The two meanings of "straight" are both... "positive" is the wrong word, how about "non-insulting". However, the new bad!gay meaning was created *specifically* to reinforce the idea that homosexual!gay was also wrong. Every time someone uses it that way, it strengthens the idea that the two are related -- it's basically the opposite of the "reclaiming" process that has (finally) made "queer" and "dyke" acceptable under certain circumstances, which was accomplished by linking the words with new, positive meanings.
WRT use of "gay" as an insult...
I'd have to stick up for them against calling them "homophobic", because homophobia really implies an intent to harm. Most of them are just mimicking their peers, and while the practice was certainly started by actual homophobes, they themselves aren't guilty of it.
HOWEVER
Just because they're wallowing in ignorance doesn't mean that it's not (a) insulting, and (b) fostering an anti-gay atmosphere. If someone points it out to them, they should make an attempt to stop (with allowances for habit). If they refuse, then they're being outright assholes. I mean, really, it's depressing enough reading the news without being reminded that a statistically significant portion of the population thinks us gay people are sub-human. Every time I hear someone use "gay" as an insult, and they don't know that I am, I have to mentally stop what I'm doing and figure out if that person was merely "joking" or if they fall into the category of people that would ostracise me, foam at the mouth and rant, (attempt to) beat me up, or worse if they found out. It gets fucking tiring after a while.
I don't know ANYONE that is a gamer and gay...
*waves hand frantically*
*waves hand unenthusiastically*
*rolls eyes and walks off to play some Metroid Prime*
The concept behind Habeas is that the poem is copyrighted, and anyone who attaches the poem to their mail without permission (e.g. when spamming) is violating copyright law and open to a lawsuit. However, since the legal system is so glacial, Habeas is getting abused so heavily these days that it's pretty much useless.
To make SpamAssassin ignore Habeas entirely, edit /etc/mail/spamassassin/local.cf (or your system's equivalent) and add the line score HABEAS_SWE 0 to the end.
Actually, we should be much more worried about crystal skulls that teleport people to offworld caverns populated by giant white aliens. Oh, and some old geezer named Nick who sees dead people.
My Windows software uses the "deprecated" INI files in the executable directory instead of the super-bloated registry ...
Eep. Stop right there, you're standing on a landmine. INI files on a modern system should be put in the per-user "Application Data" folder. You can retrieve its path using SHGetSpecialFolderPath (95/IE 4.0 or 98+) or SHGetFolderPath (98/IE 5.0 or 98SE+) using the CSIDL_APPDATA constant. If you've fail to do this, your app is currently giving ulcers to some innocent admin of WinNT-family boxes who now has to manually add extra NTFS permissions for the Everyone group to your app's install folder. Your app also doesn't work correctly with multiple users (all users share the same settings), under roaming profiles (settings are per-machine, not per-user), or running off a network drive (Ha! Like *your* app deserves chmod a+w on the Samba server!). As an added bonus, your app may stop working under XP SP2 (or after some Critical Update in the unspecified future) and almost certainly will be b0rked by the time Longhorn comes out (if MS isn't a completely lost cause, they'll have stopped making users Admin by default by that timeframe).
Actually, both. Tolkien based Quenya on Finnish, and Sindarin on Welsh. Which should be very confusing from a linguistic evolutionary standpoint, but works surprisingly well since he didn't borrow any Welsh vocabulary.
Like Janet Jackson is just Michael dressed up as a girl?
Whoa there! Just whoa! Janet Jackson is actually Michael in disguise? That's just sick and wrong.
I mean, everyone knows it's *LaToya* who's actually Michael.
In my experience, repeatable crashes in Windows from programs doing non-crash-inducing things are generally caused by (1) flakey video drivers, (2) other flakey drivers, (3) bad RAM, (4) bad video hardware. If Firefox runs on thousands upon thousands of computers, yet crashes only on yours...
...well, whatever's causing it, it ain't Firefox.
The other filters are still very useful because they allow training an empty Bayesian filter -- very spammy and very hammy messages are automatically used as training material. Besides, the DNS blacklist checks catch nearly as much spam as the Bayesian filter, at least for me.
It's not navigating, it's just downloading the file (to RAM). It might sound a bit pedantic at first, but it doesn't try to render the contents, so there's no risk of e.g. exploits against the HTML render engine.
As others mentioned, each serial line is independently clocked, so it doesn't matter if the timing gets out of sync. Theoretically, they could even run at different speeds (but practically speaking there'd be no reason to).
Actually, that bit of code just downloads the malicious .EXE. It's a bit dodgy that it's allowed to do it automatically (after all, it could be asking for http://spy.malware.com/cgi-bin/report?firstname=Jo hn&lastname=Doe&underwear_type=boxers...), but it's not an instant security breach itself. The actual bug is...
...which overwrites Media Player with the downloaded malware using ADODB.Stream (which probably never should have been enabled as a trusted ActiveX control in the first place, and certainly shouldn't be automatically overwriting files without user intervention).
Yeah, I used Netscape 4.77 up until Mozilla got to 1.0, actually. And the crashes were annoying.
You waited that long? I switched when Mozilla hit Milestone 18, way back in late '00. Netscape 4 was just awful, despite being faster on my old 300MHz system at the time.