Can't even shoot the messenger
-
@PJH, we need a legwork badge
-
@PJH, we need a legwork badge
i'll vote for that, and nominate @nightware as first recipient!
I suggest "Fancy Footwork" as the badge name
-
https://sociorocketnewsen.files.wordpress.com/2012/10/itasha.jpg
You can tell it's Japanese by the shortness of the skirt, and the apparent total lack of underwear. Unless she's wearing a thong or something akin to.Seriously, what is it with Japan and tiny skirts on teenagers?
-
Yeah ok, but to be clear, I didn't write all that stuff, or even gather it up and format it. I just pasted a link, and all the contents puked out into the post. Also, preview didn't look like it was going to do that, but it did show the top photo from the destination page.
Link was this
Feel more like I should get a "Your first thread derailment" badge.
-
Feel more like I should get a "Your first thread derailment" badge.
... we'd have to hand out do many of those it wouldn't even be funny!
-
-
... we'd have to hand out do many of those it wouldn't even be funny!
No, just one per user... Their first one. Its icon would be a cherry.
-
-
Feels like around 10, most of the time.
-
Around 893 would qualify for that badge.
-
-
Almost half our users don't post?
-
That can't be right... Discounting.
-
Is Discourse that frustrating, that so many people register to comment, but don't even make it that far?
-
There were probably at least a few imported from CS who never posted.
-
Ahhh, that makes sense.
-
Feels like around 10, most of the time.
well yes... the noisy ones....
http://what.thedailywtf.com/badges/162/5-poster
http://what.thedailywtf.com/badges/165/too-much-time-on-my-hands
-
we'd have to hand out do many of those it wouldn't even be funny!
they could just be set to trigger off the 2^2 badge or something, on the assumption that if you've made that many posts, you've probably done it by then.
-
I guess that's a negative zero, judging by the direction of the triangle?
-
I guess that's a negative zero, judging by the direction of the triangle?
i read that as "zero, which is down from the last reporting cycle"
-
i read that as "zero, which is down from the last reporting cycle"
So, we're agreeing? ;)
-
So, we're agreeing?
i don't think so. i think you were making a joke about -0 <> 0 and i was commenting in common reporting symbology.
-
@OffByOne's comment reminded me of taking limits as something goes to 0, where you have to specify if you're coming from positive or negative. Also, those weird architectures that had -0, but less so than the limit thing.
-
i think you were making a joke about -0 <> 0
I was making a joke that 0 is represented as a negative number (orange color), not as signless (what I would expect to be black or something).
Now that I reread your previous reply and finally understand what you mean, yeah, we were not agreeing, but what you say makes sense.
-
@OffByOne's comment reminded me of taking limits as something goes to 0, where you have to specify if you're coming from positive or negative.
That's not the same thing. The sign denotes the direction you're approaching from and is written in superscript after the value you're approaching, not before.
How would you write "right limit approaching -8" or "left limit approaching 5"?I used http://mathworld.wolfram.com/Limit.html to refresh my memory on the exact notation.
Also, those weird architectures that had -0, but less so than the limit thing.
You mean one's complement representation? A separate representation of -0 and +0 has its niche uses, I guess. It also simplifies arithmetic calculation implementations a little bit.
Having two separate representations of what is basically the same thing does smell like a leaked implementation detail though.
-
That's not the same thing.
Yes. Like I said, it's just what it reminded me of.
You mean one's complement representation?
Probably. I've never used one, just remember reading about it.
-
@OffByOne said:
That's not the same thing.
Yes. Like I said, it's just what it reminded me of.Indeed you did. I didn't want to quibble with you.
@OffByOne said:
You mean one's complement representation?
Probably. I've never used one, just remember reading about it.
I haven't used any of those either. I did have to learn the various number representations and their characteristics in my CS curriculum.
-
-
No one wants to quibble any more.
It's your own fault:
@ijij said:I fear to quibble with you... your quibbling has been honed to great heights by Blakey.
You've leveled up your quibbling skill too much ;)
-
Also, those weird architectures that had -0..
Oh?
[HA:pjh@sofa tmp]$ cat ./zero.c #include <stdio.h> int main(void){ float x=0.0, y=-0.0; printf("%g, %g\n", x, y); return 0; } [HA:pjh@sofa tmp]$ make zero cc zero.c -o zero [HA:pjh@sofa tmp]$ ./zero 0, -0 [HA:pjh@sofa tmp]$
-
floating point doesn't count. it was designed with -0. also infinity and -infinity and NaN and -NaN (two different types of NaN actually and both could be positive or negative)
-
-
...you forgot to mention denormalized numbers...
-
those too.
-
NaN and -NaN
I can't imagine what use there would be for maintaining a bit of information to separate those, except that it would be too much work in the hardware to do anything else with it…
-
it's a consequence of how NAN values are stored. they have a particular mantissa that says they are NAN an a particular value in the base that says what NaN they are. that leaves the sign bit free so all NaN values have a positive and negative value.
whether there's meaning associated with the sign of a NaN determines on which NaN it is and how it was created and whether the app cares.
-
The way I've had it told to me is that a NaN is what you use when you would otherwise have to throw a floating point fault of some kind. It's for stuff that is Just Plain Wrong (e.g., 0.0/0.0, which has no sane meaning at all). Given that you're doing the equivalent of throwing an exception, only having one bit to convey debugging information is nuts…
-
well the sign bit really isn't intended to be debugging information. as far as i can tell it's to simplify the design of hardware FPUs and that implementation detail leaked into the standard.
-
well the sign bit really isn't intended to be debugging information. as far as i can tell it's to simplify the design of hardware FPUs and that implementation detail leaked into the standard.
Well yes. That's what I meant by this:
too much work in the hardware to do anything else with it
-
The way I've had it told to me is that a NaN is what you use when you would otherwise have to throw a floating point fault of some kind. It's for stuff that is Just Plain Wrong (e.g., 0.0/0.0, which has no sane meaning at all). Given that you're doing the equivalent of throwing an exception, only having one bit to convey debugging information is nuts…
it's a consequence of how NAN values are stored. they have a particular mantissa that says they are NAN an a particular value in the base that says what NaN they are. that leaves the sign bit free so all NaN values have a positive and negative value.
It's a particular exponent, @accalia. The significand of a NaN is free to be tagged in system-specific ways...
-
so i used a fancy word wrong. i thought mantissa was exponent. it's the base.... whoopsie.
-
so i used a fancy word wrong. i thought mantissa was exponent. it's the base.... whoopsie.
Your typo disease seems to have spread to aphasia. You might want to talk to a doctor.
-
you really think i should see a shaman?
if you inside.....
<of course you realize i'm joking here>
-
Shaman, trepanist, whatever "works".
-
just so long as whatever solution used doesn't impair my mental faculties i'm good with that
<honk if you read these hidden messages>
-
ust so long as whatever solution used doesn't impair my mental faculties i'm good with that
According to trepanning enthusiasts it opens your mind or something, so they would claim so. At least, they would before the procedure. :) Technically they'd be correct.
I watched a video of some woman doing it to herself in the 70s. All you need is a local anaesthetic.
-
I watched a video of some woman doing it to herself in the 70s. All you need is a local anaesthetic.
And coconut water to wash it clean afterwards (it's sterile).
-
Given that you're doing the equivalent of throwing an exception, only having one bit to convey debugging information is nuts…
Theoretically, the standard does specify more types of NaN than the commonly used one, including signaling nans, which do throw an exception (generate an interrupt, whatever) when used, but in practice the one type of nan seems to be enough for anybody.
@OffByOne's comment reminded me of taking limits as something goes to 0, where you have to specify if you're coming from positive or negative. Also, those weird architectures that had -0, but less so than the limit thing.
Actually, positive and negative 0.0 are
exactlyapproximately the same thing as negative and positive limits to zero, in that they mean “a value arbitrarily close to zero, but not necessarily zero”. Try:```
[HA:pjh@sofa tmp]$ cat ./zero.c
#include <stdio.h>
int main(void){
float x=0.0, y=-0.0;
x == y &&
1/x != 1/y &&
printf("Positive and negative infinities.");
return 0;
}
[HA:pjh@sofa tmp]$ make zero
cc zero.c -o zero
[HA:pjh@sofa tmp]$ ./zero
Positive and negative infinities.
[HA:pjh@sofa tmp]$
-
Given that you're doing the equivalent of throwing an exception
And then you have sNaNs, which actually do generate traps (basically exceptions).
Some architectures use them, while others don't. When you have programs that depend on a sNaN and qNaN doing different things, and you're trying to emulate them, fun things result.
-
weird architectures that had -0
Oh, like IEEE 754 floating point as used just about everywhere?Edit: where are the fucking discotoasters when you need them