First off, the techniques involved are such that you'd be emulating cell on cell, not x86 on cell - what it's doing is managing to run most of the instructions natively, without any emulation.
Porting this wouldn't make a whole lot of sense other than collecting existing work into one project; virtualization is already available on most other architectures, because it's not as hard there. If the instruction set was designed for it and the chip gives a good way to hook the instructions that need to be emulated, it can be fairly straightforward. See http://maconlinux.org/ (which is opensource, and does PPC-on-PPC), HP's VPAR, IBM's LPAR, etc.
x86 on x86 is a particularly difficult case though; plex86, Xen, etc all do it to some extent, but forbid certain operations that are too hard to emulate. If you have control of your guest kernel (which is the only place you have to worry about privileged instructions anyway), this works well. If you want to run unmodified copies of windows, it doesn't. VMware pioneered some techniques to trap these instructions, and thus provide a full x86 instruction set, Microsoft followed with Virtual PC.
Now qemu apparently can do this too, which is quite cool, but mostly because it's now possible in an architecture where the rest of the emulation system is GPL.
No, if the hash is perfect the complexity of a collision attack is the square root of the complexity of a pre-image attack. This is known as the birthday paradox.
I dunno - windows has a leg up on shiny editors (though I'm with you on vim - it may be a terminal program at heart, but it's a programmer's best friend). I'm watching yzis with some interest - building a ttuly GUI-oriented implementation of vim's core design would allow some nice improvements.
But, I would argue that windows trails far behind in other development tool areas - tools like valgrind and oprofile. On the win32 projects I work with, I would trade visual studio for mingw any day if that meant I could have valgrind.
If I'm wrong, and equivalents do exist (don't have to be free, but thousands like Purify is pushing it), please tell me:-)
linux is capable of forcing assignments, but it's also aware of cache effects as a factor when making its decision about scheduling. So you generally don't need to override it, it tries to avoid thrashing a process from CPU-to-CPU anyway.
Horrifyingly bad is a bit of an overstatement. I built more or less the machine I described (in a scythe e-Otonashi case) for my mother. I left the case off the breakdown, since the case I used for hers wouldn't qualify for the mac mini's size - it's 280x190x90mm.
It can run neverball fine, and even quake3 if I want to really badly (in linux even). But it's not up to radeon 9200 specs by any stretch of the imagination.
You can probably come close, but beating it is going to be pretty hard. To match the size is going to take a mini-itx motherboard ($175 or so for an M10000), a 2.5" harddrive ($75 or so for 40G), optical drive ($100 or so for a dvd/cdrw), ram ($40 for 256M). I've now spent $390, still needing a case/PSU (and still larger, albiet only slightly). This is going to be *well* short of the Mac mini in performance, (especially at graphics - unichrome is nowhere near a radeon 9200 mobility), and you still don't get OSX.
If your wife wants a cute little Mac, let her have it:-)
The same isn't necessarily true for DSL (the wire is dedicated, so with the right equipment on each end the split could probably be anything desired). But I suspect it *is* still something of a zero-sum game, where increasing upstream decreases downstream. So for home use a split favoring downstream still does make a reasonable amount of sense, espescially when they have to compete with the very high (albiet shared) downstream numbers that the cable modems can advertise.
no, see, they don't care that this works. That's fine, and it doesn't undermine the effectiveness of the block as an anti-spam measure. Smarthosting like this means your mailserver puts the message together, but sends it through theirs for forwarding rather than doing delivery directly. If you send a reasonable amount of mail, this means that everything just works. If you start spewing spam out, they can realize the insane amount of traffic coming from you and block it. And if you're machine is compromised and spewing spam via some trojan (which is what the blocks are really chasing after) then it will just work because the trojan won't be configured for the right smarthost.
If they didn't do this, their IP block would be on everbody's receive blacklist as a source of spam, so it really doesn't change much one way or the other;-)
It's form factor was pretty much set by the a fairly immovable constraint - the 5.25" frisbee we've settled upon:-) Ok, it's 3/8" inch bigger on each size, but that's still doesn't leave much room to play with dimensions:-)
virgin has nice small phones now. So does tracfone, though they aren't as fancy as the phones virgin has. My brother has virgin (since it's a nationwide plan w/o an extra roaming charge) and I have tracfone (since they have fewer gaps in coverage, and I don't travel enough for the extra the 10/min when roaming to be a deal-breaker). Both are good deals if you're not a really heavy user.
What I was trying to point out is *how* timing jitter (in recording or playback) affects signals near the limiting frequency for the system. "why a CD can't go to 22050 Hz". Jitter is all-to-often mentioned out as the bogeyman of digital systems relative to the the number of people who have any clue what the effect even is.
The frequency spectrum given was for the hypothetical pure sine wave at n Hz (pre-sampling), or the distorted one (post-reconstruction). The resulting pulse train would of course have a very different characteristic if you looked at it directly, since the ideal impulse function has frequency content at all frequencies, most of which is filtered back off as part of reconstruction. Not sure if you were referring to this or to the fact that the jitter "noise" is also quite high frequency. Both are definitely true, and the playback system's filtering will get most of both.
Also, I'll freely admit that the high frequency noise introduced by the jitter (the inaccurate waveform of the signal) is going to be at or beyond human hearing (for a CD anyway). It's perceptible *at most* as intonation, and not even that much unless the gear in question is really impressively bad. But it's harmonic, so I suspect that the true golden-ear types may be able to notice it, or notice interference patterns between it and the true tones.
Amusingly enough, the comments tape-deck aficionados seemed to really like this effect as long as it come from motor speed variations.
Yep. And sampling @96kHz gives you response up to 48kHz (in theory, though in principal it the last parts suffer from intonation problems if there's any jitter at all to your sampling or reproduction clock.
Frequency distribution is a nice sharp spike at n Hz... | | | | |_______|__ n
So we'll sample at 2n Hz
_ _ / \ / \ - forgive the ascii art and \_/ \_/ pretend that was a sine wave
_|__ __|__ - sampled at 2n Hz | |
We'll even pretend that the studio gear sampled it perfectly (no clock jitter) since that gear is likely pretty damned good. So our digital signal is +1,-1,+1,-1 just like it should be
But now we play it back on a cheapo walkman that doesn't have a perfect clock, so what it synthesizes is
_|_ ___|_ _ | |
and after filtering, the analog signal it produces now looks like this _ _ / | _/ | - again, forgive the ascii art, |_/ |_/ but clearly it has steeper sections and shallower ones so it's no longer a pure tone
So the frequency distribution now looks like
| | | |_____|_|_| n
it has some frequency content to both sides of the 'real' signal (how much and how far depends on the amount of jitter present). Obviously, the signals very close to the nyquist limit suffer most from this - the lower pitches get to average the wave-shape out over multiple samples, so they will not spread out as much in the freuency domain if a point is a little off in time. But this is why the nyquist limit is not the whole story. Along with the fact that no filter is a completely sharp dropoff, this is why CD's lowpass filter to <20kHz, not at the 22050Hz Nyquist limit).
Sampling beyond 96kHz is not (yet, anyway) mainstream gear. So I think the grandparen't claim that digital equipment works to 48kHz, but has a hard time as that limit is approached is pretty fair - that's the theoretical limit (for prosumer-grade stuff), and in practice it will have trouble near the edge.
but the die is tested *before* packaging. So yes, a P4 die (full-cache) which fails testing on one bank of cache could end up packaged on the lower pinout package with the bad bank not wired up, and sold as a celeron - if (and only if) the rest of the circuitry passed.
From how the sensors work, it ought to help, but it *has* to move. A stationary magnet won't help at all, but a powerful one, close to the ground, and moving, should have some effect. Assuming the system is edge-triggered, it will probably at least register you as you pull onto the sensor. For $15 it might be worth a shot...
I've never tried it, since on my bicycle just leaning it sideways (to increase the cross-section area) is enough to reliably trigger them. But tipping a 20lb bike is an entirely easier prospect than tipping a 400lb bike:-)
pull along one edge of the sensor (you should be able to see the saw cut), and lay the bike down on its side across the pickup. Usually you don't have to go all the way down unless the sensitivity adjustment is really far out of whack - just tilting it some should be enough.
What the magnetic sensor is really measuring is flux through a wire loop that it's driving current around. This flux is proportional to the current it's using (which you don't control) and the surface area of the loop. As the signal is quite low frequency (60Hz - duh), any surface area enclosed at ground level by a continuous loop of conductive material would be essentially blocked (this is the standard "Faraday cage" effect. If the loops is higher up, the effect is to lessen the apparent area (since the field lines are curving).
A bike standing straight up is a line, with essentially no area, and the one part that might offer some area (the triangle of rear-axle and arms) is relatively far from the ground. Hence, it's nearly invisible. A bike on on it's side blocks a very large area (the wheels, the main triangle of the frame etc), and is so close to the ground that the signal is probably stronger than most cars. Thus, somewhere in between (ie, just tilting over) is usually enough to get it.
Still sucks if you're on a motorcycle, since they weigh too much to just hold at an angle (much less lay it down). I suppose one could carry a loop of wire and toss it down. But for ordinary bikes I've never found one that this won't trip (unless it's not magnetic at all, and those are very rare).
It doesn't move furniture, but it does drive around it quite tightly, pushing the loose fabric drape that's at the bottom of mine in (and reaching just a bit beyond that with its side brush). It does a very nice job of working up against things. This probably doesn't replace moving them, but I really doubt you move furniture every time either, and the side brush does a lot better job of getting tight against things than a regular head can.
Anything with open legs it seems to drive right under and edge around the legs.
And of course it doesn't do stairs, but that was just trolling anyway;-)
I have one, though I haven't had it very long (just a few days, once through the house so far). It's the newer discovery model (the white 4210). Based my experience so far, it's much more than a conversation piece. We may be 5 years behind schedule, but I finally have my robot housecleaner! Woo!
It doesn't hold a *lot*, but it didn't totally fill up when I just let it have at the downstairs (which hadn't been done for several weeks, so it was pretty bad) all at once, so it's got enough capacity to be useful; it's a bagless vac anyway, so you just empty it after it finishes up.
I watched it wander between the living room/dining room/kitchen until the novelty wore of (and/or until I got done with dishes), and it seemed to be doing fine so I went to bed. I found it on on it's charger the next morning, and my floor looked lots better... it had even founds its way into all the closets I'd left open (I never seem to bother to do them, so it was pretty obvious that it had been in there from the lack of dust-bunny civilization). Not bad at all as far as coverage went, I couldn't didn't find anything it had missed.
So, I next sent it through a room that had just been cleaned with my upright (an oreck XL, should you care), to judge it's suction. I figured this would give me a better feel for how it actually worked as a vacuum, though I didn't expect it to come back with much. While It didn't have near as much as from the living room, but it still pulled up quite a lot of hair and general crud.
It does lack for raw power (it's on batteries after all!), but on the plus side it's pretty quiet (I'd estimate somewhere around 70dB?). The actual air inlet has two flexible rubber 'lips' that constrain the airflow to a very small cross-section, which helps it have a decent pressure differential even without a lot of fan power. It also has what seems to me like a much better than normal brush design. There are two brushes spinning opposite directions, one that's a pretty typical bristle brush, and one that's slightly sticky rubber in a paddle-wheel sort of design, probably to help debris things flying back off the bristle brush and swing it into the suction.
So, in short, I'm reasonably impressed with it, though it will take more time to tell if it's really vacuuming well or just sweeping on the surface. Either way it's damned good at sweeping, far better than my normal vacuum at getting hair worked free. If it succeeds in keeping the surface clean (I'm sure it will clean more often than I would have!) there won't be near as much to get ground down in the first place. So, at this point I'm definitely giving it a "+1 Finally!" rating...
Aye. I've definitely had this argument with plenty of people back in college. If you wanted to ride with me, you had to wear a helmet. No excuses. Besides, "I'll be wearing my *spandex*. Who's going to bother to laugh at *you*?"
My roomate took a spill about two weeks after he brought his bike to school and we had this argument. But he'd bought one and *was* wearing it (thankfully!). He pretty much shattered the whole back half of the helmet, but was basically unhurt beyond some minor 'road rash'. Funny thing is, there wasn't a repeat discussion about whether or not he was going to buy a replacement... and the remains of the helmet are always helpful when another conversion needs to be made.
Slashdot Mirror
First off, the techniques involved are such that you'd be emulating cell on cell, not x86 on cell - what it's doing is managing to run most of the instructions natively, without any emulation.
Porting this wouldn't make a whole lot of sense other than collecting existing work into one project; virtualization is already available on most other architectures, because it's not as hard there. If the instruction set was designed for it and the chip gives a good way to hook the instructions that need to be emulated, it can be fairly straightforward. See http://maconlinux.org/ (which is opensource, and does PPC-on-PPC), HP's VPAR, IBM's LPAR, etc.
x86 on x86 is a particularly difficult case though; plex86, Xen, etc all do it to some extent, but forbid certain operations that are too hard to emulate. If you have control of your guest kernel (which is the only place you have to worry about privileged instructions anyway), this works well. If you want to run unmodified copies of windows, it doesn't. VMware pioneered some techniques to trap these instructions, and thus provide a full x86 instruction set, Microsoft followed with Virtual PC.
Now qemu apparently can do this too, which is quite cool, but mostly because it's now possible in an architecture where the rest of the emulation system is GPL.
No, if the hash is perfect the complexity of a collision attack is the square root of the complexity of a pre-image attack. This is known as the birthday paradox.
I dunno - windows has a leg up on shiny editors (though I'm with you on vim - it may be a terminal program at heart, but it's a programmer's best friend). I'm watching yzis with some interest - building a ttuly GUI-oriented implementation of vim's core design would allow some nice improvements.
:-)
But, I would argue that windows trails far behind in other development tool areas - tools like valgrind and oprofile. On the win32 projects I work with, I would trade visual studio for mingw any day if that meant I could have valgrind.
If I'm wrong, and equivalents do exist (don't have to be free, but thousands like Purify is pushing it), please tell me
bust the screen off and then it's probably close. So yes, the mini is basically an iBook stacked up vertically. But it's still cool :-)
linux is capable of forcing assignments, but it's also aware of cache effects as a factor when making its decision about scheduling. So you generally don't need to override it, it tries to avoid thrashing a process from CPU-to-CPU anyway.
Horrifyingly bad is a bit of an overstatement. I built more or less the machine I described (in a scythe e-Otonashi case) for my mother. I left the case off the breakdown, since the case I used for hers wouldn't qualify for the mac mini's size - it's 280x190x90mm.
It can run neverball fine, and even quake3 if I want to really badly (in linux even). But it's not up to radeon 9200 specs by any stretch of the imagination.
You can probably come close, but beating it is going to be pretty hard. To match the size is going to take a mini-itx motherboard ($175 or so for an M10000), a 2.5" harddrive ($75 or so for 40G), optical drive ($100 or so for a dvd/cdrw), ram ($40 for 256M). I've now spent $390, still needing a case/PSU (and still larger, albiet only slightly). This is going to be *well* short of the Mac mini in performance, (especially at graphics - unichrome is nowhere near a radeon 9200 mobility), and you still don't get OSX.
:-)
If your wife wants a cute little Mac, let her have it
The same isn't necessarily true for DSL (the wire is dedicated, so with the right equipment on each end the split could probably be anything desired). But I suspect it *is* still something of a zero-sum game, where increasing upstream decreases downstream. So for home use a split favoring downstream still does make a reasonable amount of sense, espescially when they have to compete with the very high (albiet shared) downstream numbers that the cable modems can advertise.
no, see, they don't care that this works. That's fine, and it doesn't undermine the effectiveness of the block as an anti-spam measure. Smarthosting like this means your mailserver puts the message together, but sends it through theirs for forwarding rather than doing delivery directly. If you send a reasonable amount of mail, this means that everything just works. If you start spewing spam out, they can realize the insane amount of traffic coming from you and block it. And if you're machine is compromised and spewing spam via some trojan (which is what the blocks are really chasing after) then it will just work because the trojan won't be configured for the right smarthost.
;-)
If they didn't do this, their IP block would be on everbody's receive blacklist as a source of spam, so it really doesn't change much one way or the other
It's form factor was pretty much set by the a fairly immovable constraint - the 5.25" frisbee we've settled upon :-) Ok, it's 3/8" inch bigger on each size, but that's still doesn't leave much room to play with dimensions :-)
virgin has nice small phones now. So does tracfone, though they aren't as fancy as the phones virgin has. My brother has virgin (since it's a nationwide plan w/o an extra roaming charge) and I have tracfone (since they have fewer gaps in coverage, and I don't travel enough for the extra the 10/min when roaming to be a deal-breaker). Both are good deals if you're not a really heavy user.
What I was trying to point out is *how* timing jitter (in recording or playback) affects signals near the limiting frequency for the system. "why a CD can't go to 22050 Hz". Jitter is all-to-often mentioned out as the bogeyman of digital systems relative to the the number of people who have any clue what the effect even is.
The frequency spectrum given was for the hypothetical pure sine wave at n Hz (pre-sampling), or the distorted one (post-reconstruction). The resulting pulse train would of course have a very different characteristic if you looked at it directly, since the ideal impulse function has frequency content at all frequencies, most of which is filtered back off as part of reconstruction. Not sure if you were referring to this or to the fact that the jitter "noise" is also quite high frequency. Both are definitely true, and the playback system's filtering will get most of both.
Also, I'll freely admit that the high frequency noise introduced by the jitter (the inaccurate waveform of the signal) is going to be at or beyond human hearing (for a CD anyway). It's perceptible *at most* as intonation, and not even that much unless the gear in question is really impressively bad. But it's harmonic, so I suspect that the true golden-ear types may be able to notice it, or notice interference patterns between it and the true tones.
Amusingly enough, the comments tape-deck aficionados seemed to really like this effect as long as it come from motor speed variations.
Yep. And sampling @96kHz gives you response up to 48kHz (in theory, though in principal it the last parts suffer from intonation problems if there's any jitter at all to your sampling or reproduction clock.
Frequency distribution is a nice sharp spike at n Hz...
| |
| |
|_______|__
n
So we'll sample at 2n Hz
_ _
/ \ / \ - forgive the ascii art and
\_/ \_/ pretend that was a sine wave
_|__ __|__ - sampled at 2n Hz
| |
We'll even pretend that the studio gear sampled it perfectly (no clock jitter) since that gear
is likely pretty damned good. So our digital signal is +1,-1,+1,-1 just like it should be
But now we play it back on a cheapo walkman that doesn't have a perfect clock, so what it synthesizes is
_|_ ___|_ _
| |
and after filtering, the analog signal it produces now looks like this
_ _
/ | _/ | - again, forgive the ascii art,
|_/ |_/ but clearly it has steeper
sections and shallower ones
so it's no longer a pure tone
So the frequency distribution now looks like
|
| |
|_____|_|_|
n
it has some frequency content to both sides of the 'real' signal (how much and how far depends on the amount of jitter present). Obviously, the signals very close to the nyquist limit suffer most from this - the lower pitches get to average the wave-shape out over multiple samples, so they will not spread out as much in the freuency domain if a point is a little off in time. But this is why the nyquist limit is not the whole story. Along with the fact that no filter is a completely sharp dropoff, this is why CD's lowpass filter to <20kHz, not at the 22050Hz Nyquist limit).
Sampling beyond 96kHz is not (yet, anyway) mainstream gear. So I think the grandparen't claim that digital equipment works to 48kHz, but has a hard time as that limit is approached is pretty fair - that's the theoretical limit (for prosumer-grade stuff), and in practice it will have trouble near the edge.
but the die is tested *before* packaging. So yes, a P4 die (full-cache) which fails testing on one bank of cache could end up packaged on the lower pinout package with the bad bank not wired up, and sold as a celeron - if (and only if) the rest of the circuitry passed.
It would be quite difficult to make such sockets work at the sort of speeds that VRAM usually gets run at though...
I think that both a&b are true. PCI-E really is better at pulling this off than AGP, resulting in a card that runs HL2/Doom3 decently for $100
Doesn't exactly sound like a bad thing to me... no, it's not as fast as the $200 version. Surprise, surprise.
As you pointed out, a print server provides printing. A file server provides files. etc.
:-)
I usually like to phrase the Xserver case as follows:
An X server provides software with a human who can "sign on the X_______________, please"
Or, there's always "In Soviet Russia, X serves you!"
this of course assumes you have infinite money. I suppose if you have infinite money and like gambling, you're all set :-)
:-)
Otherwise the math works just the same, eventually you lose
From how the sensors work, it ought to help, but it *has* to move. A stationary magnet won't help at all, but a powerful one, close to the ground, and moving, should have some effect. Assuming the system is edge-triggered, it will probably at least register you as you pull onto the sensor. For $15 it might be worth a shot...
:-)
I've never tried it, since on my bicycle just leaning it sideways (to increase the cross-section area) is enough to reliably trigger them. But tipping a 20lb bike is an entirely easier prospect than tipping a 400lb bike
it's not distance, it's cross-section area. But yes, this is the #1 thing you can do to make a bike more visible to the sensor details
pull along one edge of the sensor (you should be able to see the saw cut), and lay the bike down on its side across the pickup. Usually you don't have to go all the way down unless the sensitivity adjustment is really far out of whack - just tilting it some should be enough.
What the magnetic sensor is really measuring is flux through a wire loop that it's driving current around. This flux is proportional to the current it's using (which you don't control) and the surface area of the loop. As the signal is quite low frequency (60Hz - duh), any surface area enclosed at ground level by a continuous loop of conductive material would be essentially blocked (this is the standard "Faraday cage" effect. If the loops is higher up, the effect is to lessen the apparent area (since the field lines are curving).
A bike standing straight up is a line, with essentially no area, and the one part that might offer some area (the triangle of rear-axle and arms) is relatively far from the ground. Hence, it's nearly invisible. A bike on on it's side blocks a very large area (the wheels, the main triangle of the frame etc), and is so close to the ground that the signal is probably stronger than most cars. Thus, somewhere in between (ie, just tilting over) is usually enough to get it.
Still sucks if you're on a motorcycle, since they weigh too much to just hold at an angle (much less lay it down). I suppose one could carry a loop of wire and toss it down. But for ordinary bikes I've never found one that this won't trip (unless it's not magnetic at all, and those are very rare).
It doesn't move furniture, but it does drive around it quite tightly, pushing the loose fabric drape that's at the bottom of mine in (and reaching just a bit beyond that with its side brush). It does a very nice job of working up against things. This probably doesn't replace moving them, but I really doubt you move furniture every time either, and the side brush does a lot better job of getting tight against things than a regular head can.
Anything with open legs it seems to drive right under and edge around the legs.
And of course it doesn't do stairs, but that was just trolling anyway;-)
I have one, though I haven't had it very long (just a few days, once through the house so far). It's the newer discovery model (the white 4210). Based my experience so far, it's much more than a conversation piece. We may be 5 years behind schedule, but I finally have my robot housecleaner! Woo!
It doesn't hold a *lot*, but it didn't totally fill up when I just let it have at the downstairs (which hadn't been done for several weeks, so it was pretty bad) all at once, so it's got enough capacity to be useful; it's a bagless vac anyway, so you just empty it after it finishes up.
I watched it wander between the living room/dining room/kitchen until the novelty wore of (and/or until I got done with dishes), and it seemed to be doing fine so I went to bed. I found it on on it's charger the next morning, and my floor looked lots better... it had even founds its way into all the closets I'd left open (I never seem to bother to do them, so it was pretty obvious that it had been in there from the lack of dust-bunny civilization). Not bad at all as far as coverage went, I couldn't didn't find anything it had missed.
So, I next sent it through a room that had just been cleaned with my upright (an oreck XL, should you care), to judge it's suction. I figured this would give me a better feel for how it actually worked as a vacuum, though I didn't expect it to come back with much. While It didn't have near as much as from the living room, but it still pulled up quite a lot of hair and general crud.
It does lack for raw power (it's on batteries after all!), but on the plus side it's pretty quiet (I'd estimate somewhere around 70dB?). The actual air inlet has two flexible rubber 'lips' that constrain the airflow to a very small cross-section, which helps it have a decent pressure differential even without a lot of fan power. It also has what seems to me like a much better than normal brush design. There are two brushes spinning opposite directions, one that's a pretty typical bristle brush, and one that's slightly sticky rubber in a paddle-wheel sort of design, probably to help debris things flying back off the bristle brush and swing it into the suction.
So, in short, I'm reasonably impressed with it, though it will take more time to tell if it's really vacuuming well or just sweeping on the surface. Either way it's damned good at sweeping, far better than my normal vacuum at getting hair worked free. If it succeeds in keeping the surface clean (I'm sure it will clean more often than I would have!) there won't be near as much to get ground down in the first place. So, at this point I'm definitely giving it a "+1 Finally!" rating...
Since I carved two fairly rad pumpkins this year, I guess I'll throw out a link :-)
d =f8b25c8cc851ba2e69b93ba3c3470767
Both are based on prints from http://itswalky.com/.
http://zoo.nightstar.net/viewtopic.php?t=10011&si
Aye. I've definitely had this argument with plenty of people back in college. If you wanted to ride with me, you had to wear a helmet. No excuses. Besides, "I'll be wearing my *spandex*. Who's going to bother to laugh at *you*?"
My roomate took a spill about two weeks after he brought his bike to school and we had this argument. But he'd bought one and *was* wearing it (thankfully!). He pretty much shattered the whole back half of the helmet, but was basically unhurt beyond some minor 'road rash'. Funny thing is, there wasn't a repeat discussion about whether or not he was going to buy a replacement... and the remains of the helmet are always helpful when another conversion needs to be made.