The C standard, n1570 5.1.2.3pp2,4, disagrees, as I read it. I would claim that the zeroing of memory is a "needed side effect", and that the compiler has no reason to assert that it is not needed. The burdon of proof of non-neededness should be on the one doing the optimisation. It might be that the pointer was locally allocated, and never communicated to an external function, and therefore the object being pointed to was not accessible elsewhere (such as from a signal handler), that would probably be a satisfactory proof, but I'm fairly sure things weren't that cut and dry.
I don't disagree with your second paragraph at all. Many optimisations are semantics preserving. Use of register variables (this is the big win, once you can do this, you can do most of the others, as there are no visible side effects regarding what's stored in registers), redundant store elimination, common subexpression elimination, loop unrolling,...
I agree about review, though. And that review should include runs through static code analysis tools such as coverity, purify, and this new one.
I get a completely contradictory opinion from the Taiwanese that I know and work with. The Taiwanese were the Chinese government in exile - if anything, the "real" Chinese government. And therefore, "Chinese".
We're spinning round the sun at about 30 km/s, so the average speed of something we encounter will be around that midpoint. Some things we'll be catching up (so subtract speeds[*]), some things will be catching us up (also subtract speeds[*]), but sometimes they'll be approaching head on (so add speeds[*]). A direct head on colision cuts through only 10km of atmosphere, so indeed there's less than a second of that. However, at the very limits, a grazing collision (or near miss) can cut through over 500km (1000km) of atmosphere (from pythagoras' theorem: hypotenuse = earth's radius + atmosphere, adjecent edge = earth's radius). That's a fair few seconds. Typically, you'll be between the two, more likely to be at the shorter end, grazing is a corner case, so to speak.
[* of course, this is a velocity calculation, directions matter, but I'm simplifying onto a 1-dimensional case.]
> I find it very curious that you'd be prohibited from using the value of the pointer proper. Do you have a citation for it being undefined?
N1570 L.3p2 "The value of a pointer that refers to space deallocated by a call to the free or realloc function is used (7.22.3)."
realloc() *always* deallocates, unless it returns NULL. Simply growing an area counts as a deallocation of the old area followed by a new allocation at the same spot.
> In C? Assuming low and high are unsigned > (low >> 1) + (high >> 1) + (low & high & 1). Ick.
(low & high) + ((low ^ high) >> 1);
Every bit in the & expression would be included twice in the sum, and then halved. So leave them alone. Every bit in the ^ expression would be only included once in the sum, on one side or the other, we care not, and then halved. So just half those.
The example given (memset before free) was not dead code. The compiler is entirely at fault in removing that code. It did not see writing to non-automatic storage as being a side effect, when in truth writing to non-automatic storage is *nothing but* a side effect.
TFS doesn't say "packed together tightly" it says "packed more tightly".
It's a comparitive. You would appear to have a problem with the statement that Peter Dinklage is taller than Warwick Davis.
And only a nob would use metres as the unit for describing interplanetary distances. What's worse - only a complete nob would give those distances to 10 significant digits.
There needs to be viral downmodding on/., so that it's not just the idiot who posts crap who gets punished, but everyone who upmodded it gets punished too for their ignorance.
> It is neither a bullet [...] It doesn't use gunpowder, it uses compressed air.
Bollocks - it is a projectile flung at force towards a target. You're not confusing the word "bullet" and "cartridge", are you? There's precisely no need for gunpowder or any other explosive to be involved in the flinging of bullets.
Personally, I think the alu ones are *very* well built, but that's the only one I've ever owned (dual G5, and likewise I've done HD and RAM upgrades). I thought the bubblegum-coloured plastic all-in-ones just looked as if they were cheap plastic, but am preparet to believe that the engineering inside them is just as good, even if the design is awful.
Being an amateur photographer, I wanted to design my own business cards for one of my businesses. Being exclusively linux/FOSS, I tried GIMP. On screen, I was quite proud of what I'd designed. Until I saw it on card.
Alas, my bold ambers came out a kind of bilberry blue in the test run of the cards. It's my belief that until I've got end-to-end RAW/CMYK, all I will be able to do is tweak curves and pay for another test run (less than 5e for 36 cards, and the kinds of people I'm giving these to don't care about the visuals, so it's an annoyance rather than a disaster). No idea how many iterations will be necessary.
Well, the clip in my own follow-up contains one example of it, but it wasn't the one I was thinking of. I'm 99.99% sure it was Spitting Image, up from 99% sure before finding the above clip, but it was 25 years ago.
Accepted. Even if they were made in China, thinkpads still fell into the "branded" category. Indeed they were good - pretty much best of the bunch. I've had Toshiba and HP laptops fail repeatedly, my (proper IBM-branded) thinkpad is still running well. I've never owned an apple laptop, I have heard good things about their build quality, but of course that would be from people who've invested their money in them. However, I've never seen anyone with an 8 year old Mac laptop (the age of my thinkpad, IIRC), so there's no way of judging what their longevity really is.
I would think that such funding would make more sense if it was per party rather than per candidate. Fortunately all of the 4 main parties, and a representative from the ad hoc election coalition, were already all represented on the pre-election debates, so there's no apparently no need here for any more encouragement for the little guy. One of the benefits of only having democracy for a couple of decades is that the futility of Duverger's law and the structures that support that hasn't kicked in yet.
Nobody with any experience in HCI thinks that the puck mouse was anything apart from half-baked. And the iPod shuffle with no buttons at all is pretty brain-dead, as validated by them putting a button back on the next version.
Well, the last time I was in my city of birth, it appears that the youth's primary source of amusement was burning furniture shops and looting. It was quite a spooky view at about 5am, as there was little left but pillars of smoke with a few red glints from the soon-to-be sunrise, as we rode the taxi to the airport, crossing most of north London.
I can't understand why the UK police have this logic: - Electrician going to work: shoot 7 times in the head - Looters setting fire to listed buildings: do bugger all Personally, I'd reverse both of those policies.
I did say "PC's", look again. I deliberately chose the "pee-cee" expansion as a dumbification of the term in order to reinforce the payload in that part of the sentence.
If choice of words was purely out of "need", there would be nothing that was interesting to read, no personal expression, no literature. I don't think that's a good thing.
In 2006, the consortium rebranded the architecture, 32-bit instances and all, to just being the "Power architecture". I know this, as I worked for one of the companies that made the 32-bit members of the family. We were specifically told to stop using the terms "PowerPC" and "PPC", even when refering to legacy chips (we had a customer-facing office, this was important). Only people who didn't get the memo call it by the retired name.
Slashdot Mirror
The C standard, n1570 5.1.2.3pp2,4, disagrees, as I read it. I would claim that the zeroing of memory is a "needed side effect", and that the compiler has no reason to assert that it is not needed. The burdon of proof of non-neededness should be on the one doing the optimisation. It might be that the pointer was locally allocated, and never communicated to an external function, and therefore the object being pointed to was not accessible elsewhere (such as from a signal handler), that would probably be a satisfactory proof, but I'm fairly sure things weren't that cut and dry.
...
I don't disagree with your second paragraph at all. Many optimisations are semantics preserving. Use of register variables (this is the big win, once you can do this, you can do most of the others, as there are no visible side effects regarding what's stored in registers), redundant store elimination, common subexpression elimination, loop unrolling,
I agree about review, though. And that review should include runs through static code analysis tools such as coverity, purify, and this new one.
I get a completely contradictory opinion from the Taiwanese that I know and work with. The Taiwanese were the Chinese government in exile - if anything, the "real" Chinese government. And therefore, "Chinese".
400 ton rocks are small potatoes. We still don't know the whereabouts of every 10000000 ton rock that's got a potential earth orbit intersection.
We're spinning round the sun at about 30 km/s, so the average speed of something we encounter will be around that midpoint. Some things we'll be catching up (so subtract speeds[*]), some things will be catching us up (also subtract speeds[*]), but sometimes they'll be approaching head on (so add speeds[*]). A direct head on colision cuts through only 10km of atmosphere, so indeed there's less than a second of that. However, at the very limits, a grazing collision (or near miss) can cut through over 500km (1000km) of atmosphere (from pythagoras' theorem: hypotenuse = earth's radius + atmosphere, adjecent edge = earth's radius). That's a fair few seconds. Typically, you'll be between the two, more likely to be at the shorter end, grazing is a corner case, so to speak.
[* of course, this is a velocity calculation, directions matter, but I'm simplifying onto a 1-dimensional case.]
> I find it very curious that you'd be prohibited from using the value of the pointer proper. Do you have a citation for it being undefined?
N1570 L.3p2
"The value of a pointer that refers to space deallocated by a call to the free or realloc function is used (7.22.3)."
realloc() *always* deallocates, unless it returns NULL. Simply growing an area counts as a deallocation of the old area followed by a new allocation at the same spot.
> Switching rings was too damn slow. So slow, that only academics ever used them, and not very well. It is no wonder it died.
I can't think of any current system that doesn't have something running at ring 0 and something running at ring 3.
>> Quick summary: (low + high) / 2
> In C? Assuming low and high are unsigned
> (low >> 1) + (high >> 1) + (low & high & 1). Ick.
(low & high) + ((low ^ high) >> 1);
Every bit in the & expression would be included twice in the sum, and then halved. So leave them alone.
Every bit in the ^ expression would be only included once in the sum, on one side or the other, we care not, and then halved. So just half those.
The example given (memset before free) was not dead code.
The compiler is entirely at fault in removing that code. It did not see writing to non-automatic storage as being a side effect, when in truth writing to non-automatic storage is *nothing but* a side effect.
TFS doesn't say "packed together tightly" it says "packed more tightly".
/., so that it's not just the idiot who posts crap who gets punished, but everyone who upmodded it gets punished too for their ignorance.
It's a comparitive. You would appear to have a problem with the statement that Peter Dinklage is taller than Warwick Davis.
And only a nob would use metres as the unit for describing interplanetary distances. What's worse - only a complete nob would give those distances to 10 significant digits.
There needs to be viral downmodding on
The problem is that jammers have, by design, a rather large RF footprint.
> It is neither a bullet [...] It doesn't use gunpowder, it uses compressed air.
Bollocks - it is a projectile flung at force towards a target. You're not confusing the word "bullet" and "cartridge", are you? There's precisely no need for gunpowder or any other explosive to be involved in the flinging of bullets.
Personally, I think the alu ones are *very* well built, but that's the only one I've ever owned (dual G5, and likewise I've done HD and RAM upgrades). I thought the bubblegum-coloured plastic all-in-ones just looked as if they were cheap plastic, but am preparet to believe that the engineering inside them is just as good, even if the design is awful.
Do you not understand the difference between "not needed at early boot" and "not going to be run, ever"?
Good. Now compare what I was talking about, and what you are talking about.
Now say sorry, and promise to post non-anonymously in the future, so that you have some incentive not to post crap.
Being an amateur photographer, I wanted to design my own business cards for one of my businesses. Being exclusively linux/FOSS, I tried GIMP. On screen, I was quite proud of what I'd designed. Until I saw it on card.
Alas, my bold ambers came out a kind of bilberry blue in the test run of the cards. It's my belief that until I've got end-to-end RAW/CMYK, all I will be able to do is tweak curves and pay for another test run (less than 5e for 36 cards, and the kinds of people I'm giving these to don't care about the visuals, so it's an annoyance rather than a disaster). No idea how many iterations will be necessary.
If GIMP wanted CMYK, then it could have done it a decade ago when it was first asked for. When they were laughed at for not having it. Repeatedly.
Well, the clip in my own follow-up contains one example of it, but it wasn't the one I was thinking of. I'm 99.99% sure it was Spitting Image, up from 99% sure before finding the above clip, but it was 25 years ago.
Example here: http://foaas.com/this/slashdot%20readership
Accepted. Even if they were made in China, thinkpads still fell into the "branded" category. Indeed they were good - pretty much best of the bunch. I've had Toshiba and HP laptops fail repeatedly, my (proper IBM-branded) thinkpad is still running well. I've never owned an apple laptop, I have heard good things about their build quality, but of course that would be from people who've invested their money in them. However, I've never seen anyone with an 8 year old Mac laptop (the age of my thinkpad, IIRC), so there's no way of judging what their longevity really is.
I would think that such funding would make more sense if it was per party rather than per candidate. Fortunately all of the 4 main parties, and a representative from the ad hoc election coalition, were already all represented on the pre-election debates, so there's no apparently no need here for any more encouragement for the little guy. One of the benefits of only having democracy for a couple of decades is that the futility of Duverger's law and the structures that support that hasn't kicked in yet.
Don't drunken Oxbridge types do pranks with cows? So it really could be either.
What makes you think I'm a US citizen?
> Apple doesn't release half-baked products.
Nobody with any experience in HCI thinks that the puck mouse was anything apart from half-baked. And the iPod shuffle with no buttons at all is pretty brain-dead, as validated by them putting a button back on the next version.
Well, the last time I was in my city of birth, it appears that the youth's primary source of amusement was burning furniture shops and looting.
It was quite a spooky view at about 5am, as there was little left but pillars of smoke with a few red glints from the soon-to-be sunrise, as we rode the taxi to the airport, crossing most of north London.
I can't understand why the UK police have this logic:
- Electrician going to work: shoot 7 times in the head
- Looters setting fire to listed buildings: do bugger all
Personally, I'd reverse both of those policies.
I did say "PC's", look again. I deliberately chose the "pee-cee" expansion as a dumbification of the term in order to reinforce the payload in that part of the sentence.
If choice of words was purely out of "need", there would be nothing that was interesting to read, no personal expression, no literature. I don't think that's a good thing.
In 2006, the consortium rebranded the architecture, 32-bit instances and all, to just being the "Power architecture". I know this, as I worked for one of the companies that made the 32-bit members of the family. We were specifically told to stop using the terms "PowerPC" and "PPC", even when refering to legacy chips (we had a customer-facing office, this was important). Only people who didn't get the memo call it by the retired name.