My proposal does not address the problem of RMS occasionally being apparently petty, or the argument over "GNU/Linux" but it does address the problem of giving proper credit to the work done by the GNU project.
My proposal is this: If you use GNU software on a regular basis, particularly if you use the software as part of your work, accredit them. Find a place to say "This work would not have been possible without the many fine utilities produced by the GNU project." or something similar.
Y'know, I dig GNU and all, and I surely appreciate that all those GNU applications give me something to run on my nice, shiny Linux box (GCC and libc would be especially hard to live without) - but sometimes RMS just comes across as being so petty! Is the important issue here the -name- of an operating system, or a club, or the -respect- which the GNU project deserves?
Consider: If that SIGLinux had changed its name to SIGFree, its name wouldn't have meant much to anybody. SIGFree is a pretty stupid name. "SIGLinux", even if it doesn't represent everything a "Linux club" is interested in ("Linux Clubs" in my experience also appeal to BSD users and people with a general interest in hacking, Free Software, etc.) the meaning is recognizable: someone who said one day, "You know, I'd like to set up a Linux system" could search, find the club, and then the club could educate them on the importance of GNU.
Consider: The name of the club, or OS, isn't nearly as important with regard to this issue as the treatment of GNU. If "Linux"-oriented organizations made it abundantly clear that they, and Linux itself, owe a great deal to the GNU project, then proper accreditation is served.
IMO the GNU project owes about as much to free kernels like Linux as the free kernels owe to GNU. GNU took way too long to develop their own kernel: and GNU's system software without a kernel is as useless as a kernel without system software. RMS can nitpick over stupid little issues in the Linux kernel all he wants: but it works, it supports all my hardware, and it has journalling filesystems. Nothing GNU has produced can compete with that.
Defining "OS" strictly in terms of the kernel is a very limited perspective, and generally not in-line with current OS design. An Operating System includes a kernel: it also includes system software, that is, anything a user might need (or, more generally, want) in order to perform basic, common operations on the system. This is the basis behind GNU's contention that Linux systems are actually GNU/Linux systems: take the GNU software away and you don't have much to work with.
KDE and Gnome -are- operating systems. They define their own rules for how interactions take place between applications, and the user, and the "low level" services provided by the underlying system.
10 years ago the term "system software" would include file manipulation commands and a command shell, because those were the basic services users and programmers needed. These days applications are routinely expected to carry on various kinds of networking transactions, work with common data types hosted by other applications, and hardly any user wants to do without simple services like a media player or a web browser. That means that MS must at least include these things with the OS, or the system isn't complete.
As for whether the features should be decoupled: if a vendor like Microsoft provides a set of programs in a single bundle, it's not unreasonable for some of those programs to rely upon each other. The OS kernel wouldn't be expected to rely on something like the web browser in such a case, but important system software -might-. And in the effort to provide a fault-free system, it's not at all hard to imagine why such a vendor might want to prevent the use of a third-party substitute module, whose quality the vendor can't vouch for.
HAL 9000 went insane because it was given contradictory commands. It didn't have the option of saying "hey, that doesn't make much sense now that I think about it, I ought to just do what I think is right." HAL was ordered to place the mission first - and it appeared to him that the crew endangered the mission. It resolved the contradiction with the simplest means available: eliminating the crew.
That wouldn't have been my first guess why naming it HAL was a bad idea anyway - I would have thought maybe "Lawsuit" or "Tedious, unoriginal name idea".
What if -everyone- who received unsolicited spam replied to it? Then the spammers would be spending too much time weeding out all the "Sorry, I'm not interested, and by the way I forged my reply address" posts to find those poor slobs who -are- interested in the Herbal-V and toner cartridges.
I think the biggest flaw in the given solution (and one that wasn't pointed out too well) is that, regardless of what filesystem is used, how efficiently the filenames are stored, etc., the compressed files cannot be stored in less space than the original data.
Think about this. For every one byte he takes out, he adds another file to his collection. Even if you store all the files in a linear block, like a TAR file or something, and store their filenames by a mechanism that requires no storage, you still need to encode the *length* of each of those blocks - requiring at least one or two bytes, depending on the block size. The challenge didn't say anything about the cost of storing the files in a filesystem (inodes or the like) but it did say that the compressed files, with the decompressor, needed to take less space than the original file - and the solution can't be stored in a way that meets those requirements, even under very generous assumptions about the conditions of the storage.
The challenge should have been stated better, but the challenger still lost IMO.
Re:Why it's better... not great
on
PDAs, PDAs
·
· Score: 1
I think the Motorolas are still the right choice, for a little longer at least, in terms of power efficiency... Besides, this whole Palm thing is helping a whole new generation of programmers the importance of optimization (myself included)
You know, for some odd reason, the name "HandEra 330" makes me think of Tuna Fish with mayonnaise. Especially when I think about going into grayscale mode on that nice QVGA screen. Anybody else get this feeling?
I just like the standardization of the form factor, myself. Until the m500/505 Palm hasn't been able to stick to one form factor - VII, IIIc with different length than III, m100 with no siblings of the same size/shape, V series now replaced by the m5 series... Now I can use my new Rand GPS without the CF card getting in the way... and I have a lot more pixels to see my map with, too.
As for going smaller, I think going smaller is more convenient than going thinner - that is, more like the Sony form factor rather than the m500 or Visor Edge form factor. (Palm and Handspring have gone for super-thin without reducing the profile. Sony's kept the dimensions proportionate and IMO given it a better feel at a smaller size - if there were more than one manufacturer of Memory Stick equipment in the world, I might be sorely tempted to buy the new Sony instead of the new TRG..)
Re:I'd rather have 2 CF slots
on
PDAs, PDAs
·
· Score: 1
Uhm, if you want to have storage and CF I/O at the same time, why not buy an SD memory card?? Besides, having two CF slots would screw up the form factor - and CF is slower anyway.
Plus, keep in mind that Palm is going to drive a lot of new PalmOS products to the SD format, including SD I/O devices, eventually. It's certainly not unreasonable to expect that someone would release, say, an SD I/O GPS with a PalmOS mapping application and a shitload of maps installed on it, within the next year. New PalmOS devices aiming to have mass storage (other than those setting their own standards) include SD. It's that simple. TRG isn't going for a lot of new standards with this device - CF inherited from TRGPro, SD to keep up with Palm, Palm III form factor so geeks like me can use all that year-old accessory hardware, the only place they're coming up with new specs is the screen - the extra resolution adds a lot, I can't wait to get mine.:)
Re:Things to do to new PDAs
on
PDAs, PDAs
·
· Score: 1
You don't need to "hack" ethernet. You can buy a CF ethernet adaptor off-the-rack and plug it in.
Heh, your Visor just cracked? That's funny, my TRGPro recently suffered a mysterious accident. Only part of the digitizer that still works is the menu key.
Why do people ask questions like "Why would anybody want this?"? Duh, if you don't want a microdrive, don't buy one. For that matter, it's not as though microdrives were designed for TRG's line of handhelds. More likely Wince machines.
However, one question that was asked about mass storage at last December's Palmsource went like this: "When Palm OS devices support mass storage, will I be able to load up my 800 MB database for such-and-such.." (I can't remember what the guy wanted it for... inventory database, medical, I dunno. I just vividly remember thinking that either the guy was a TRG plant or the TRG guys were jumping for joy at getting to say "Yes, you can do that already with a TRGPro"
Short answer: Under PalmOS, 1GB is probably gonna be most useful if you have something you want to put on it before you go out. Yeah, you're not going to be accessing, much less writing, all that data over the course of a single charge - but having it all with you means you can access it without going back to your desk if you've gone, say, a few hundred miles away.
As far as 1: The plan is they're going to make the new connector common across most or all their future models: Meaning model 500 accessories (wireless modems, cradles, GPS units, etc.) will also work with model 700 or 800 or whatever.
2: SD isn't nearly as proprietary as Memory Stick - there are, for instance, companies besides Palm selling SD and MMC cards, and companies besides Palm selling devices that use SD/MMC cards. For instance, the 64MB Panasonic SD card and USB adaptor I ordered last week. Yeah, copy control sucks, but it's a fact of life that companies developing media these days are gonna do it.
Why support another new technology? How's this: better transfer rates, smaller size, promise of a hell of a lot of nice I/O toys in the future... sounds good to me.
OK, yeah, normally it's kind of stupid to set up random device X as a Linux box. But in some cases it's not such a bad idea. Consider:
PS2 has USB. Hence, you can hook up ethernet, keyboard, etc.
PS2 has TV output, meaning you don't have to lug a monitor around.
PS2 has FMV and 3-d video acceleration, meaning you can use it to display spiffy graphics.
I can understand why you might not see it as particularly compelling (esp. since PS2 does not read CDRs AFAIK) but if accessible and programmable in Linux, it could be a neat little system to use. Of course, you would need a hard drive - Sony added a large bay for such a device, so it may be a possibility in the future.
"...or have you been able to compile a 2.4 kernel as zImage?"
Yes, I have. I needed to build a kernel for an IBM 730T (tablet computer with PCMCIA hard drive) which I could loadlin and fit on a ramdisk. Disable the extra features and you get rid of the bloat. It's really that simple.
Maybe the new icon should be a Sony dual-shock controller. It certainly looks as though that's what the new Nintendo controller is - a poorly redesigned dual-shock.
Looks pretty cool... but I'm a PSX-type, so I probably won't go for it.
(Side note - with one of those USB adaptors, PSX controllers are excellent for Linux gaming)
I saw that too in the article - he was taking both ESR's and RMS's positions as the heart and soul of all free software developers. The biggest problems with that are
1: They're not. I find some of RMS's ramblings insightful, informative, even inspiring at times. Others I find misguided. I think any intelligent programmer who pays attention to the central figures of Free Software/Open Source will agree that it's essential to make one's own decisions about which of all these assertions is and is not worthwhile. 2: They're incompatible. RMS mentioned this emphatically in the recent Slashdot interview. So if ESR's behavior conflicts with RMS's ideals, that is not a basis for considering the movement flawed.
I also think he was off on a bad tangeant with the whole ESR gun control thing. From what I've seen ESR does try to plug his personal cause in the same venues where he supports Free/Open Source software. But frankly, that's his business. I think anyone motivated enough to have a well-developed opinion would want to promote it, if they found themselves in the position to do so. The ethical question is this - if ESR really believes the right to own and carry guns is an important one to preserve, isn't he in fact morally obligated to promote that view - much in the same way as those of us who believe he happens to be wrong on that point are obligated to oppose him? Again, his word isn't law any more than Stallman's is. But hopefully we'll find some useful ideas.
I think the article sort of contradicts itself a bit with the ending - as though its body were written to downplay RMS's ideals, but its conclusion was written to tell everybody to have a good day. The article puts down GNU's accomplishments, saying isn't it ironic (doncha think) that GNU's goal was to create a free system, but wound up being a central piece of commercial unixes... A little too ironic. yeah, I really do think. It's like rain. He also puts down GCC because the Windows port isn't bulletproof (IIRC GNU doesn't even maintain the Windows port)... ho-hum.
I just think it's exciting that one of GNU's original goals, the free system, is finally being realized. Reading RMS material it sounds more like a return to free computing, rather than a "revolution".
//The poster wrote this referring to another message - saying the first poster wasn't making a good argument PocketPC is about 10x faster than Palm, but you have a feeling that this is irrelevant. Anyway, 20MHz is enough for anyone.
I remember someone famous once said "640K is enough memory for anyone". I'm guessing that someone has learned a thing or three since then.
Different users, different needs, different devices. Currently I need a Palm.
I think one of the few things anti-Palm types do sort of have right is their comments on battery life figures. We Palm users do tend to exaggerate the numbers for "actual use". I haven't used alkalines for a while but I believe my usual recorded powered-on uptime was only about 34 hours, which for me was one month of use (note taking, scheduling, Scheme programming in LispMe, Palm development, etc.) Certainly nowhere near the 50 hours on/3 months active (or more) figures you hear from some, but a hell of a lot more than 8 hours "on".
There are some situations where limited (rechargable) battery life is acceptable. I often find myself in situations where it's not, and I often encounter new ones - for starters, in my job I develop palmtop mapping and GPS products - if someone wants to use one of those products to navigate on a hiking trip or something I don't see how an 8 hour pack and a cradle recharger is gonna cut it. Luggable battery pack, maybe (You'd need to carry extras for the Palm and GPS anyway...) hrmmmm....
Both the Microsoft Wince/Palm comparison and the Palm "Stay calm, we still rock your world" E-mail were pretty much pure marketing, fairly unattached to reality and each repulsive in its own special way - but the one bit I enjoyed was when the Palm message said "Besides, there's no question that Windows CE is less unpleasant than it used to be". That was a kick, one of the high points of a Palm propaganda counter-strike that just screamed out "We own the market, so we can be super-cocky sons of bitches!"
Personally, I say we let the marketing monkeys have their war - mean time the rest of us can do something useful. I want to reach a new level of hacker zen. Reading the 1987 edition of the RMS Emacs manual is good for the soul, I highly recommend it.
Slashdot Mirror
My proposal does not address the problem of RMS occasionally being apparently petty, or the argument over "GNU/Linux" but it does address the problem of giving proper credit to the work done by the GNU project.
My proposal is this: If you use GNU software on a regular basis, particularly if you use the software as part of your work, accredit them. Find a place to say "This work would not have been possible without the many fine utilities produced by the GNU project." or something similar.
Y'know, I dig GNU and all, and I surely appreciate that all those GNU applications give me something to run on my nice, shiny Linux box (GCC and libc would be especially hard to live without) - but sometimes RMS just comes across as being so petty! Is the important issue here the -name- of an operating system, or a club, or the -respect- which the GNU project deserves?
Consider: If that SIGLinux had changed its name to SIGFree, its name wouldn't have meant much to anybody. SIGFree is a pretty stupid name. "SIGLinux", even if it doesn't represent everything a "Linux club" is interested in ("Linux Clubs" in my experience also appeal to BSD users and people with a general interest in hacking, Free Software, etc.) the meaning is recognizable: someone who said one day, "You know, I'd like to set up a Linux system" could search, find the club, and then the club could educate them on the importance of GNU.
Consider: The name of the club, or OS, isn't nearly as important with regard to this issue as the treatment of GNU. If "Linux"-oriented organizations made it abundantly clear that they, and Linux itself, owe a great deal to the GNU project, then proper accreditation is served.
IMO the GNU project owes about as much to free kernels like Linux as the free kernels owe to GNU. GNU took way too long to develop their own kernel: and GNU's system software without a kernel is as useless as a kernel without system software. RMS can nitpick over stupid little issues in the Linux kernel all he wants: but it works, it supports all my hardware, and it has journalling filesystems. Nothing GNU has produced can compete with that.
Defining "OS" strictly in terms of the kernel is a very limited perspective, and generally not in-line with current OS design. An Operating System includes a kernel: it also includes system software, that is, anything a user might need (or, more generally, want) in order to perform basic, common operations on the system. This is the basis behind GNU's contention that Linux systems are actually GNU/Linux systems: take the GNU software away and you don't have much to work with.
KDE and Gnome -are- operating systems. They define their own rules for how interactions take place between applications, and the user, and the "low level" services provided by the underlying system.
10 years ago the term "system software" would include file manipulation commands and a command shell, because those were the basic services users and programmers needed. These days applications are routinely expected to carry on various kinds of networking transactions, work with common data types hosted by other applications, and hardly any user wants to do without simple services like a media player or a web browser. That means that MS must at least include these things with the OS, or the system isn't complete.
As for whether the features should be decoupled: if a vendor like Microsoft provides a set of programs in a single bundle, it's not unreasonable for some of those programs to rely upon each other. The OS kernel wouldn't be expected to rely on something like the web browser in such a case, but important system software -might-. And in the effort to provide a fault-free system, it's not at all hard to imagine why such a vendor might want to prevent the use of a third-party substitute module, whose quality the vendor can't vouch for.
HAL 9000 went insane because it was given contradictory commands. It didn't have the option of saying "hey, that doesn't make much sense now that I think about it, I ought to just do what I think is right." HAL was ordered to place the mission first - and it appeared to him that the crew endangered the mission. It resolved the contradiction with the simplest means available: eliminating the crew.
That wouldn't have been my first guess why naming it HAL was a bad idea anyway - I would have thought maybe "Lawsuit" or "Tedious, unoriginal name idea".
What if -everyone- who received unsolicited spam replied to it? Then the spammers would be spending too much time weeding out all the "Sorry, I'm not interested, and by the way I forged my reply address" posts to find those poor slobs who -are- interested in the Herbal-V and toner cartridges.
"GOD-DAMMIT!!"
It wouldn't fit.
Seriously, have you seen overhead bins these days? They aren't big enough to hold much of anything - and certainly not one carry-on per passenger.
---GEC
(M-x depeche-mode)
"Mercy is bad for the vision, ruthless will clear it away"
I think the biggest flaw in the given solution (and one that wasn't pointed out too well) is that, regardless of what filesystem is used, how efficiently the filenames are stored, etc., the compressed files cannot be stored in less space than the original data.
Think about this. For every one byte he takes out, he adds another file to his collection. Even if you store all the files in a linear block, like a TAR file or something, and store their filenames by a mechanism that requires no storage, you still need to encode the *length* of each of those blocks - requiring at least one or two bytes, depending on the block size. The challenge didn't say anything about the cost of storing the files in a filesystem (inodes or the like) but it did say that the compressed files, with the decompressor, needed to take less space than the original file - and the solution can't be stored in a way that meets those requirements, even under very generous assumptions about the conditions of the storage.
The challenge should have been stated better, but the challenger still lost IMO.
And the winner is... CmdrTaco!
I think the Motorolas are still the right choice, for a little longer at least, in terms of power efficiency... Besides, this whole Palm thing is helping a whole new generation of programmers the importance of optimization (myself included)
You know, for some odd reason, the name "HandEra 330" makes me think of Tuna Fish with mayonnaise. Especially when I think about going into grayscale mode on that nice QVGA screen. Anybody else get this feeling?
I just like the standardization of the form factor, myself. Until the m500/505 Palm hasn't been able to stick to one form factor - VII, IIIc with different length than III, m100 with no siblings of the same size/shape, V series now replaced by the m5 series... Now I can use my new Rand GPS without the CF card getting in the way... and I have a lot more pixels to see my map with, too.
As for going smaller, I think going smaller is more convenient than going thinner - that is, more like the Sony form factor rather than the m500 or Visor Edge form factor. (Palm and Handspring have gone for super-thin without reducing the profile. Sony's kept the dimensions proportionate and IMO given it a better feel at a smaller size - if there were more than one manufacturer of Memory Stick equipment in the world, I might be sorely tempted to buy the new Sony instead of the new TRG..)
Uhm, if you want to have storage and CF I/O at the same time, why not buy an SD memory card?? Besides, having two CF slots would screw up the form factor - and CF is slower anyway.
:)
Plus, keep in mind that Palm is going to drive a lot of new PalmOS products to the SD format, including SD I/O devices, eventually. It's certainly not unreasonable to expect that someone would release, say, an SD I/O GPS with a PalmOS mapping application and a shitload of maps installed on it, within the next year. New PalmOS devices aiming to have mass storage (other than those setting their own standards) include SD. It's that simple. TRG isn't going for a lot of new standards with this device - CF inherited from TRGPro, SD to keep up with Palm, Palm III form factor so geeks like me can use all that year-old accessory hardware, the only place they're coming up with new specs is the screen - the extra resolution adds a lot, I can't wait to get mine.
You don't need to "hack" ethernet. You can buy a CF ethernet adaptor off-the-rack and plug it in.
Heh, your Visor just cracked? That's funny, my TRGPro recently suffered a mysterious accident. Only part of the digitizer that still works is the menu key.
Why do people ask questions like "Why would anybody want this?"? Duh, if you don't want a microdrive, don't buy one. For that matter, it's not as though microdrives were designed for TRG's line of handhelds. More likely Wince machines. However, one question that was asked about mass storage at last December's Palmsource went like this: "When Palm OS devices support mass storage, will I be able to load up my 800 MB database for such-and-such.." (I can't remember what the guy wanted it for... inventory database, medical, I dunno. I just vividly remember thinking that either the guy was a TRG plant or the TRG guys were jumping for joy at getting to say "Yes, you can do that already with a TRGPro" Short answer: Under PalmOS, 1GB is probably gonna be most useful if you have something you want to put on it before you go out. Yeah, you're not going to be accessing, much less writing, all that data over the course of a single charge - but having it all with you means you can access it without going back to your desk if you've gone, say, a few hundred miles away.
As far as 1: The plan is they're going to make the new connector common across most or all their future models: Meaning model 500 accessories (wireless modems, cradles, GPS units, etc.) will also work with model 700 or 800 or whatever.
2: SD isn't nearly as proprietary as Memory Stick - there are, for instance, companies besides Palm selling SD and MMC cards, and companies besides Palm selling devices that use SD/MMC cards. For instance, the 64MB Panasonic SD card and USB adaptor I ordered last week. Yeah, copy control sucks, but it's a fact of life that companies developing media these days are gonna do it.
Why support another new technology? How's this: better transfer rates, smaller size, promise of a hell of a lot of nice I/O toys in the future... sounds good to me.
I have played the Starfighter demo - I agree, don't go for it. Compared to other games the gameplay is just way too bland.
:)
By "Other Games" I particularly mean Armored Core II - build anime-style giant robots and fight your way through missions... what could be better?
OK, yeah, normally it's kind of stupid to set up random device X as a Linux box. But in some cases it's not such a bad idea. Consider: PS2 has USB. Hence, you can hook up ethernet, keyboard, etc. PS2 has TV output, meaning you don't have to lug a monitor around. PS2 has FMV and 3-d video acceleration, meaning you can use it to display spiffy graphics. I can understand why you might not see it as particularly compelling (esp. since PS2 does not read CDRs AFAIK) but if accessible and programmable in Linux, it could be a neat little system to use. Of course, you would need a hard drive - Sony added a large bay for such a device, so it may be a possibility in the future.
"...or have you been able to compile a 2.4 kernel as zImage?"
Yes, I have. I needed to build a kernel for an IBM 730T (tablet computer with PCMCIA hard drive) which I could loadlin and fit on a ramdisk. Disable the extra features and you get rid of the bloat. It's really that simple.
Maybe the new icon should be a Sony dual-shock controller. It certainly looks as though that's what the new Nintendo controller is - a poorly redesigned dual-shock.
Looks pretty cool... but I'm a PSX-type, so I probably won't go for it.
(Side note - with one of those USB adaptors, PSX controllers are excellent for Linux gaming)
Um, Megacheese isn't new. It's been out for at least a year or two now.
Still, in light of the "50 least influential movies list" (or whatever) it might be worth a look.
I saw that too in the article - he was taking both ESR's and RMS's positions as the heart and soul of all free software developers. The biggest problems with that are
1: They're not. I find some of RMS's ramblings insightful, informative, even inspiring at times. Others I find misguided. I think any intelligent programmer who pays attention to the central figures of Free Software/Open Source will agree that it's essential to make one's own decisions about which of all these assertions is and is not worthwhile.
2: They're incompatible. RMS mentioned this emphatically in the recent Slashdot interview. So if ESR's behavior conflicts with RMS's ideals, that is not a basis for considering the movement flawed.
I also think he was off on a bad tangeant with the whole ESR gun control thing. From what I've seen ESR does try to plug his personal cause in the same venues where he supports Free/Open Source software. But frankly, that's his business. I think anyone motivated enough to have a well-developed opinion would want to promote it, if they found themselves in the position to do so. The ethical question is this - if ESR really believes the right to own and carry guns is an important one to preserve, isn't he in fact morally obligated to promote that view - much in the same way as those of us who believe he happens to be wrong on that point are obligated to oppose him? Again, his word isn't law any more than Stallman's is. But hopefully we'll find some useful ideas.
I think the article sort of contradicts itself a bit with the ending - as though its body were written to downplay RMS's ideals, but its conclusion was written to tell everybody to have a good day. The article puts down GNU's accomplishments, saying isn't it ironic (doncha think) that GNU's goal was to create a free system, but wound up being a central piece of commercial unixes... A little too ironic. yeah, I really do think. It's like rain. He also puts down GCC because the Windows port isn't bulletproof (IIRC GNU doesn't even maintain the Windows port)... ho-hum.
I just think it's exciting that one of GNU's original goals, the free system, is finally being realized. Reading RMS material it sounds more like a return to free computing, rather than a "revolution".
//The poster wrote this referring to another message - saying the first poster wasn't making a good argument
PocketPC is about 10x faster than Palm, but you have a feeling that this is irrelevant. Anyway, 20MHz is enough for anyone.
I remember someone famous once said "640K is enough memory for anyone". I'm guessing that someone has learned a thing or three since then.
Different users, different needs, different devices. Currently I need a Palm.
I think one of the few things anti-Palm types do sort of have right is their comments on battery life figures. We Palm users do tend to exaggerate the numbers for "actual use". I haven't used alkalines for a while but I believe my usual recorded powered-on uptime was only about 34 hours, which for me was one month of use (note taking, scheduling, Scheme programming in LispMe, Palm development, etc.) Certainly nowhere near the 50 hours on/3 months active (or more) figures you hear from some, but a hell of a lot more than 8 hours "on".
There are some situations where limited (rechargable) battery life is acceptable. I often find myself in situations where it's not, and I often encounter new ones - for starters, in my job I develop palmtop mapping and GPS products - if someone wants to use one of those products to navigate on a hiking trip or something I don't see how an 8 hour pack and a cradle recharger is gonna cut it. Luggable battery pack, maybe (You'd need to carry extras for the Palm and GPS anyway...) hrmmmm....
Both the Microsoft Wince/Palm comparison and the Palm "Stay calm, we still rock your world" E-mail were pretty much pure marketing, fairly unattached to reality and each repulsive in its own special way - but the one bit I enjoyed was when the Palm message said "Besides, there's no question that Windows CE is less unpleasant than it used to be". That was a kick, one of the high points of a Palm propaganda counter-strike that just screamed out "We own the market, so we can be super-cocky sons of bitches!"
Personally, I say we let the marketing monkeys have their war - mean time the rest of us can do something useful. I want to reach a new level of hacker zen. Reading the 1987 edition of the RMS Emacs manual is good for the soul, I highly recommend it.