User Sarenne on Wikipedia was running a (mostly) one-man campaign to replace all power-of-two MB/KB/etc with MiB/KiB/etc, with some support from others on the Manual of Style: Units and Numbers page. A while back someone had said they should use MiB/etc, the (small) number of people who frequent that style page agreed (after lots of argument), but not much changed, until Sarenne started a mass-change campaign, backed up by the consensus on the style page. Each time someone complained, they'd come there, complain, a few others would agree, and they'd be told "it's already been decided, it's been discussed to death, you lose". After a while, they'd go away, guaranteeing that there never would be enough of them around to change the consensus. Eventually, enough users were annoyed and stuck around to convince the MOS:UaN people to hold a poll, and Sarenne started going too far in forcing edits into pages and got himself banned. It looks like it finally got changed to:
There is no consensus to use the newer IEC-recommended prefixes in Wikipedia articles to represent binary units. There is consensus that editors should not change prefixes from one style to the other, especially if there is uncertainty as to which term is appropriate within the context--one must be certain whether "100 GB" means binary not decimal units in the material at hand before equating it with 100 GiB. When this is certain, the use of parentheses for IEC binary prefixes, for example, 256 KB (KiB) is acceptable, as is the use of footnotes to disambiguate prefixes. When in doubt, stay with established usage in the article, and follow the lead of the first major contributor.
Anyone really interested in this should read about this bill (the "Holt bill") at both http://blackboxvoting.org/ and also http://blackboxvoting.com/. Note that these two sites have very different takes on this bill!!! IMHO, they both have valid points. BBV (.org) has done a great job pushing forward the problems with the current systems and raising the visibility of the issue (and ferreting out some real doosies by Diebold, etc). However, since we're not likely to get anything better for this cycle, people should consider if the Holt bill would improve things. The danger would be that people might assume that "ok, now that we have paper trails all is OK and we can stop worrying about it."
Here in Chester County, PA we had a local house race that went to a full hand recount last year. (Chester County bucked the trend in Pennsylvania and installed optical-scan machines, where a recount is possible.) The original count had the Republican winning the PA house seat by 19 votes, and after a recount Barbara McIlvaine Smith (D-PA 156th) unseated the Republican by 23 votes, which switched control the PA house (by 1 vote) for the first time in around a dozen years to the Democrats.
Right - Gonzales gets it a bit of added press (though I agree that "Gonzales" attached to it probably hurts it right now). The question is how many dems have been tempted with RIAA money... Or is it just Lamar making sure his buddies keep throwing money towards him.
The administration being behind it will help, and it will get more notice. The real question is whether the RIAA has bought off enough democrats to get this on the docket for a vote.
If this really is an artifact of the old 'core' of AOL, then it's probably due to the original password functions we put into PlayNET back in 1984-1985. (For those that don't know, AOL was originally a port of QuantumLink to the PC, and QuantumLink was licensed from PlayNET. See http://en.wikipedia.com/wiki/PlayNET.)
The original core was all done in PL/1 on Stratus fault-tolerant minicomputers. They continued to run the core up until a few years ago, but much of the design was so ingrained that it continues to exist in places until this day. That was why it took so long to have more than 10-character usernames, and why last I checked (a couple of years ago), the login protocol still 'uses' my old error-correcting protocol from PlayNET, which was designed to allow error-free communication across non-corrected 300 baud modems (sliding windows, piggybacked acks, CRC-16, special tricks to avoid 0x0D/etc (because Tymnet/Telenet/X.25 pads took that as a "forward the string" code).
Precisely - total release is lower with CFLs. It can be a point-source release, which is an issue. NEMA tells people who break one to sweep up the fragments (no vacuum) and use paper towels.
Note that a broken CFL has 5mg; a broken old-style thermostat has 500-2000mg.
Or discharge (largely) the ultra-cap when parked in a garage or at a charger at work, and make sure the charger can give a complete charge in a few minutes (maybe combined with a timer). Read AMPS - probably not possible unless there's a local storage device that ISN'T an ultracap.... Could install vacuum flywheels in every home.
So don't expect them to replace batteries in cars, but they could greatly improve electric-only cars in combination with Lithium-ion batteries. Don't use the ultracap for long-term storage, but use it for performance and possibly for a just-before-trip "top-off".
I said innovation has slowed, not what's commonly used or used by a particular programmer. A lot of what's happened recently has been implementation of older ideas and incremental change. There are a lot less "left-field" languages being dreamed up, less entirely new paradigms for languages. Java is not exactly new. C# is probably the "newest" of those, dating to ~2000. Java and Javascript are circa 1995 (11 years old), and Python is 1991 (16 years old).
See http://www.levenez.com/lang/history.html. Note that most of the recent stuff is just things like "Python 2.3.2". Look how many "new" languages have appeared since 2000 on that chart (close to 0). Hell, even Ruby dates to 1993.
Back in the 1980's (and into the 90's) there was an entire magazine devoted to this area called "Computer Language".
Re:marketing works
on
Who won?
·
· Score: 2, Interesting
For fairness, you should note that the "colleages" you mention often either didn't know him personally or were misquoted or duped. Others (including some who did know him) had a deep, vitriolic, long-standing grudge against him over his testimony about the war.
My father is good friends with the soldier who was rescued by Kerry, and who spoke at the convention for him. Sure, his perspective is colored by Kerry having saved his life - but he can say with certainty that Kerry did save his life, which the Swift Boaters tried to poo-poo. And he was a life-long Republican as well (he switched for one reason - to vote for Kerry in the primary). Kerry didn't seek him out for publicity for the election, either - he came to Kerry and offered. Kerry hadn't seen him since Vietnam.
The questioner says they haven't really advanced, but are worried because their salary requirements have increased.
Either a) you're wrong about advancing (perhaps you're underestimating yourself), b) you've advanced some (experience and maturity at least), but not enough to justify your current salary, or c) you're appropriately paid for what you do but you've shown yourself not to be interested in advancing and learning, and so may fall out of sync with current practices (in which case, crack the books and start learning!)
New languages every year isn't important in the way it once was; language innovation seems to have slowed compared to say the 1980's. Substitute new design patterns, application frameworks, libraries, etc instead.
Hold the speculation - there have been cases before of kidnap victims (including adults) who were either brainwashed or threatened or both into not reporting it and not running away from the kidnapper when they could. ("I'll kill your parents...") In the right setting, they can end up believing it.
Without the new kidnap case, the odds might tilt to runaway. With the new case and the circumstances (not similar to a runaway situation), the odds may tilt towards kidnapping/etc. Or maybe the guy 'befriended' these kids somewhere (physically or online) and offered them a "safe" way to get away from home.... (how "safe" it was might be a question...).
So, can the speculation for now - because that's what it is, speculation.
FSF requires copyright assignment for contibuting to GNU projects. This assists copyright enforcement (but is not necessary), and may have unintended consequences in the event of an FSF bankruptcy and auction of assets.
Ho boy, that's a wrinkle I hadn't thought of. Though sale of the copyrights can't change existing licenses people are using the code under (and so they can continue to use it), it does also mean that the code a) could be separately sold to other companies to be used in closed-source projects, and b) the under-new-management/ownership-FSF could issue a new GPL that would apply to all existing V2+ projects and code. See my other post in this topic for other ways V2+ could produce unexpected results if the FSF "published" a radically different version of the license, for example by mistake, or even a minor typo.
Recent email exchanges have convinced me that RMS is rather more radical than I had assumed. (absent code better than closed, derivative wedge). Closer to what his critics claim. A zealot. There is a surprisingly large gulf between "Free Software" and "Open Source".
Yeah, I realized that exchanging email with him back in 1986.... and he has gotten if anything more radical since then. He made the same argument back then (pre-GPLv1 I think), even in response to the case of "closed-source life-saving device that saves millions of people" vs "no life-saving device made". I was at BoF/session at Boscone around that time ('85-87ish) where he and some closed-source developers were having shouting matches.
I trust him to try. And they may. But even "binding" trusts backed by billions of dollars in art and hundreds of millions in cash can end up being "broken" by the courts if a crisis is created, subverting the original intent. (See Barnes Museum in Philadelphia, now slated to move from his mansion in the near suburbs to the middle of Philadelphia to "solve" an (avoidable) crisis.) Now that's a different sort of thing, but it illustrates the point that it's hard to bind successors to even hard-written details ("the artwork is not to be moved from current location"), let alone philosophies.
Even assuming good faith, successors may disagree what actions compromise following the intentions of the founder. And (to use another not-direct example), the "party" Leiberman created in Connecticut so he could get a better spot on the ballot ("Connecticut for Lieberman Party") has been taken over by one of his critics, who has set the rules so that only "critics, bloggers and anyone named Lieberman can join the party".
The "or any later" clause is a blank check. It doesn't stop someone from using V2, but by contributing to that you're allowing for your contributions (once someone else submits a V(>2)-only patch, effectively making the package V(>2)-only from then on) to be licensed under a license that didn't even exist when you effectively agreed to it. For example, if you agreed with V2, but now think the (likely) V3 restrictions on DRM are too restrictive - too bad, you effectively agreed to your patches being used as a tool in the war against DRM. (And perhaps you're happy with that instead of upset - great! But my point is that your contribution can be used to promote goals that didn't even exist when you agreed to it.)
Hell, what if there were to be a schism in the FSF over some issue? What if there were two competing V3's? It says
If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation."
For that matter, what does "Publish" mean there? If FSF "publishes" on their website something meant as a draft of V10, it seems like that counts as "publish" and could be used. What if there's a typo in a version they "publish"? Once it's "published", it can't be taken back per the above wording. Note that "publish" doesn't mean "gave a number to", "publish" means
1 a : to make generally known b : to make public announcement of
2 a : to disseminate to the public b : to produce or release for distribution; specifically PRINT c : to issue the work of (an author)
So which of those applies? Now, perhaps the legal argument is that a draft isn't a "version" of "this License", but typoed 'versions' would be, wildly different versions would be, and inadvertently-released versions would be (if someone put it on their website or any other FSF "publication" forgetting to put "draft" on it). You might even have a case where someone who works for the FSF "publishes" (without permission from Stallman) a "GPLv10" that says "copy in any way you want" for 5 minutes on their website or on FSF letterhead. A court might find that to be an FSF "publication" of a version of the license, allowing it to be used with all V2+ projects and code submitted under V2+.
Basically, you can drive a truck through "V2+", which is exactly how Stallman wanted it - he has his own personal truck, which you may (or may not) agree with where he's driving it, or where he may decide to drive in the future.
Back in the original A3000/16 (16MHz) days, we timed bootup (including loading kickstart from HD) as 11 seconds cold, 7 seconds warmboot - including starting TCP/IP. I used to regularly boot an A3000+ (lab rat) into WB, TCPIP, mount 4 or so NFS volumes, and have a telnet session up in about 15, maybe 20 seconds - and I think most of that delay was NFS.
My comment was in regards partly to your implication that there's no such thing as "late/end-stage diabetes". My father is very lucky and very unusual - but even he is walking an ever-narrower tightrope. Most people, even those who do a good job controlling blood sugar, don't get anywhere hear his age (or relatively low amount of side-effects). And he's had several close calls, especially over the last 10 years.
As for the possible treatment: I'm not advocating "rushing" into human trials (though early trials with related suppressives have begun). The Wikipedia page on the suppressive mentions that it's banned in human use due to toxic side-effects. The trial in Israel (see the SciAm article) uses a known-safe-in-humans suppressant that doesn't suppress as much (but also didn't seem to work well enough; I think the SciAm article said they were going to try higher doses). Also note that the suppressive only suppresses part of the immune system; my understanding is that it's less than what is used in transplants, which are the other main hope for Type 1 diabetics, and as pointed out by someone else, the protocol doesn't involve permanent suppression like a transplant - it's temporary while the immune system is "retrained" not to attack them.
The interesting thing about the new study is that it verifies that the new islets come from spleen cells.
For "brittle"/end-stage diabetics, the loss of the ability to recognize blood-sugar changes and the fact that it starts swinging without warning leads to increasing severe side-effects and often death. Those are some of the patients who've taken part in transplant trials in Canada over the last few years, which have seemed to work, but with some problems with slow rejection/die-off of the transplanted islets. They were chosen because for them, the risks of lifetime use of immuno-suppressives is lower than the risk of not doing something. The same reasoning may end up applying here, but with hopefully lower risks (partial suppression, temporary not lifetime, lower chance of rejection/re-attack of islets).
There is no such thing as severe/end-stage type I diabetes. Usually by the time you are diagnosed, you are at the "severe/end stage" - Your pancreatic beta cells are gone or nearly so. Insulin can prolong your life for decades, and if your bloodsugars are carefully controlled (via aggressive and careful diet, insulin dosing, and glucose monitoring), you will live just as long a life as a normal person.
If you're talking about severe diabetes complications (Kidney damage, retina damage, etc.)- By the time those present themselves, the cumulative damage of years of abnormal bloodsugars is done and curing the underlying diabetes isn't going to help.
As the son of a type-1 diabetic, I have to disagree in part. My father is 72 and has had type 1 since he was 9 - way beyond expected survival for a diabetic. He has lived around as long as a normal person, with a tightly-controlled diet, weight, exercise, etc. Due to a kidney tumor, he has one kidney, and diabetes had reduced them to 40%; post-removal the remaining one has stabilized at 30%, which makes for a tricky diet, but doesn't require dialysis. Minor retinal damage until last winter, though still not too bad.
However, an effective cure would help him a lot, even at this late date. It becomes increasingly hard to control blood sugar levels, and he doesn't notice when levels wander into dangerous levels anymore. While it wouldn't reverse the damage to the kidneys, vision, etc, it would remove a major cause of further damage and life-threatening blood sugar swings. Even a partial reversal would help a lot.
Also, some very early experiments in humans have been done in Israel, using a less-toxic immune-suppressive (which doesn't suppress as much). No success, but there may be some data from it that it was heading in the right direction (see SciAm article).
This would be great if it works; my father is a 72-year-old juvenile diabetic (since age 9 - WAY outliving the probabilities), and my cousin once-removed on my father's side is also a juvenile diabetic (age ~23, diabetic since ~19 or 20). Many type-1 diabetics die before they're 40, often with severe complications.
Lemmings (original or Lemmings 2) had simply FANTASTIC multiplayer on an Amiga with two mice. The fun of digging a hole that starts draining all your opponent's lemmings into a pit while they're busy elsewhere was amazing. It was so good, at an Amiga DevCon a bunch of Commodore employees and a few developers all dressed up as Lemmings as a surprise to the attendees and "acted out" a Lemmings skit.
MULE was groundbreaking on the C64. Simple as sin on the surface, but plain fun for 4.
Gauntlet - fast and fun on the Amiga for two or 4.
One last one that launched a genre of god-games, and was a ancestor of games like Black and White - Populous. In some ways similar to a much fancier, 3-d Lemmings.
The EPA said commodore replaced a tank in the ground with an illegal unlined tank in 1975!
Commodore didn't buy MOS until after that (a year or so later, can't find the exact date). MOS was in serious financial difficulty around that time due to the bottom falling out of the calculator market.
The stock 1541 was roughly the equivalent to 2400 baud or a bit faster; around 1 sector/second. Note that most programs/people used "fastloaders" that programmed the 6502 in the 1541 and upped load speeds by 3-10x.
Tripos... Some aspects of Tripos were good. The big problem was mostly
BCPL, and how it interacted with the rest of the system. Two generations before C, it had some problematic issues, like the
global vector. "global vector" was a kindof-replacement for libraries, or
if you prefer a equivalent to the standard C library, but as a shared lib. Pointers not being native pointers (pointer to an integer was a machine pointer shifted-right two) was a real thorn too.
BCPL on a 68000 had real issues, in that they set the stack up
"upside-down" by normal 68000 standards. This caused all sorts of problems
integrating it with system libraries, and for user code trying to call it
(dos.library calls all had to go through a translation layer). It also had
real issues with strings, due to not liking pointers to bytes much. If
Tripos hadn't been in BCPL, it would have not been a huge issue.
More documentation wouldn't have solved these problems really (and even Commodore had essentially NO documentation other than the mostly uncommented code). I had
fought these as a developer trying to write a replacement shell before
joining Commodore. And in reality it was still Tripos, just ported to C
and with a big "translation" layer in ASM so that existing BCPL code could
run (i.e. BCPL now went through translation functions, and C/ASM/etc went
direct). Plus a bunch of additional interfaces for things like replacing the shell, etc. Look at the docs on the dos.library interface differences between 1.3 and 2.0 for an example.
The ramdisk, for example, was dramatically easier to work on and extend
(and speed up) after I rewrote it. The ramdisk got quite fast. FFS already existed in ASM (and also
handled OFS), and in reality didn't lose any real-world safety (disks
already had per-block checksums, adding another layer of them to the FS
just slows things down (hugely)). Many calls it didn't matter for, but for some the speedup was dramatic. But most of the speedups were from redesign of bigger, complex pieces. The biggest BCPL->C/ASM speedups were in really simple calls that didn't do a lot, but in BCPL invoked some heavy translation layers.
MEMF_PUBLIC - of little use unless something was checking that from the
start. And it would only have allowed protection, not per-process address
spaces. VM would have been useful; people did work on trying to implement
VM without per-process address spaces or inter-process protection. It
kinda-sorta worked, but not really. If you wanted VM and process protection in an exec-like message-passing microkernel, QNX was a lot closer to what you'd have designed (it was later, too).
As for the need for VM - I did all my development on Amigas; when I finally had a machine with 16MB of ram (which wasn't until a while after I had an A3000T), I can't remember getting below 4MB free, and rarely below 8MB free (until I started running an early version of NCSA Mosaic to browse the web in '93-94). That's not to say people couldn't have used it, but it wasn't as critical as it is today, when my WinXP machine is using 600MB (out of 1GB), and 275MB of that is a browser (which a huge number of tabs open, and which has been running for a month), and my Linux box is using around 1GB of 1.5GB (again, a big browser running for weeks and X are the hogs). Note, however, that neither machine is paging, and things slow down a lot when you do. But it is useful at times, and the #1 use of paging:
Stopping programmers from having to bother about checking memory allocations or error codes from lots of leaf functions.
When an Amiga ran out of memory, it remained almost totally stable, because we (and many of our developers) were pedantic about checking returns and handling error paths, and we had LOTS of tools to make it easy to stress-test programs and the OS for this. An awfully high percentage of programs now simply call exit() if they run out of memory... assuming they notice at all.
George (designer of A500 and A1200, admin of cbmvax) owned a former railroad station around 30 miles away, and mostly lived in his cube 24/7. Once in a while he'd go home to pick up mail, see if the water heater had burst, etc. Most of the time I was there he didn't drive, just rode a bike.
He paid off the mortgage and didn't even notice, just kept sending checks. They started having to send him checks with notes asking him to please stop. I think it was after he finally stopped that he started just throwing his checks in a drawer (too much hassle to go deposit them).
His primary leisure activities were admining cbmvax, USENET, inline rollerskating in the huge warehouse/production floor, reading SF, and railroads.
Slashdot Mirror
Anyone really interested in this should read about this bill (the "Holt bill") at both http://blackboxvoting.org/ and also http://blackboxvoting.com/. Note that these two sites have very different takes on this bill!!! IMHO, they both have valid points. BBV (.org) has done a great job pushing forward the problems with the current systems and raising the visibility of the issue (and ferreting out some real doosies by Diebold, etc). However, since we're not likely to get anything better for this cycle, people should consider if the Holt bill would improve things. The danger would be that people might assume that "ok, now that we have paper trails all is OK and we can stop worrying about it."
Here in Chester County, PA we had a local house race that went to a full hand recount last year. (Chester County bucked the trend in Pennsylvania and installed optical-scan machines, where a recount is possible.) The original count had the Republican winning the PA house seat by 19 votes, and after a recount Barbara McIlvaine Smith (D-PA 156th) unseated the Republican by 23 votes, which switched control the PA house (by 1 vote) for the first time in around a dozen years to the Democrats.
Right back at 'cha, Buddy! ;-)
Right - Gonzales gets it a bit of added press (though I agree that "Gonzales" attached to it probably hurts it right now). The question is how many dems have been tempted with RIAA money... Or is it just Lamar making sure his buddies keep throwing money towards him.
The administration being behind it will help, and it will get more notice. The real question is whether the RIAA has bought off enough democrats to get this on the docket for a vote.
If this really is an artifact of the old 'core' of AOL, then it's probably due to the original password functions we put into PlayNET back in 1984-1985. (For those that don't know, AOL was originally a port of QuantumLink to the PC, and QuantumLink was licensed from PlayNET. See http://en.wikipedia.com/wiki/PlayNET.)
The original core was all done in PL/1 on Stratus fault-tolerant minicomputers. They continued to run the core up until a few years ago, but much of the design was so ingrained that it continues to exist in places until this day. That was why it took so long to have more than 10-character usernames, and why last I checked (a couple of years ago), the login protocol still 'uses' my old error-correcting protocol from PlayNET, which was designed to allow error-free communication across non-corrected 300 baud modems (sliding windows, piggybacked acks, CRC-16, special tricks to avoid 0x0D/etc (because Tymnet/Telenet/X.25 pads took that as a "forward the string" code).
Precisely - total release is lower with CFLs. It can be a point-source release, which is an issue. NEMA tells people who break one to sweep up the fragments (no vacuum) and use paper towels.
Note that a broken CFL has 5mg; a broken old-style thermostat has 500-2000mg.
Not unless it was connected to a PSTN bridge, which from your description it wasn't.
Or discharge (largely) the ultra-cap when parked in a garage or at a charger at work, and make sure the charger can give a complete charge in a few minutes (maybe combined with a timer). Read AMPS - probably not possible unless there's a local storage device that ISN'T an ultracap.... Could install vacuum flywheels in every home.
So don't expect them to replace batteries in cars, but they could greatly improve electric-only cars in combination with Lithium-ion batteries. Don't use the ultracap for long-term storage, but use it for performance and possibly for a just-before-trip "top-off".
I said innovation has slowed, not what's commonly used or used by a particular programmer. A lot of what's happened recently has been implementation of older ideas and incremental change. There are a lot less "left-field" languages being dreamed up, less entirely new paradigms for languages. Java is not exactly new. C# is probably the "newest" of those, dating to ~2000. Java and Javascript are circa 1995 (11 years old), and Python is 1991 (16 years old).
See http://www.levenez.com/lang/history.html. Note that most of the recent stuff is just things like "Python 2.3.2". Look how many "new" languages have appeared since 2000 on that chart (close to 0). Hell, even Ruby dates to 1993.
Back in the 1980's (and into the 90's) there was an entire magazine devoted to this area called "Computer Language".
For fairness, you should note that the "colleages" you mention often either didn't know him personally or were misquoted or duped. Others (including some who did know him) had a deep, vitriolic, long-standing grudge against him over his testimony about the war.
My father is good friends with the soldier who was rescued by Kerry, and who spoke at the convention for him. Sure, his perspective is colored by Kerry having saved his life - but he can say with certainty that Kerry did save his life, which the Swift Boaters tried to poo-poo. And he was a life-long Republican as well (he switched for one reason - to vote for Kerry in the primary). Kerry didn't seek him out for publicity for the election, either - he came to Kerry and offered. Kerry hadn't seen him since Vietnam.
The questioner says they haven't really advanced, but are worried because their salary requirements have increased.
Either a) you're wrong about advancing (perhaps you're underestimating yourself), b) you've advanced some (experience and maturity at least), but not enough to justify your current salary, or c) you're appropriately paid for what you do but you've shown yourself not to be interested in advancing and learning, and so may fall out of sync with current practices (in which case, crack the books and start learning!)
New languages every year isn't important in the way it once was; language innovation seems to have slowed compared to say the 1980's. Substitute new design patterns, application frameworks, libraries, etc instead.
Hold the speculation - there have been cases before of kidnap victims (including adults) who were either brainwashed or threatened or both into not reporting it and not running away from the kidnapper when they could. ("I'll kill your parents...") In the right setting, they can end up believing it.
Without the new kidnap case, the odds might tilt to runaway. With the new case and the circumstances (not similar to a runaway situation), the odds may tilt towards kidnapping/etc. Or maybe the guy 'befriended' these kids somewhere (physically or online) and offered them a "safe" way to get away from home.... (how "safe" it was might be a question...).
So, can the speculation for now - because that's what it is, speculation.
Even assuming good faith, successors may disagree what actions compromise following the intentions of the founder. And (to use another not-direct example), the "party" Leiberman created in Connecticut so he could get a better spot on the ballot ("Connecticut for Lieberman Party") has been taken over by one of his critics, who has set the rules so that only "critics, bloggers and anyone named Lieberman can join the party".
The "or any later" clause is a blank check. It doesn't stop someone from using V2, but by contributing to that you're allowing for your contributions (once someone else submits a V(>2)-only patch, effectively making the package V(>2)-only from then on) to be licensed under a license that didn't even exist when you effectively agreed to it. For example, if you agreed with V2, but now think the (likely) V3 restrictions on DRM are too restrictive - too bad, you effectively agreed to your patches being used as a tool in the war against DRM. (And perhaps you're happy with that instead of upset - great! But my point is that your contribution can be used to promote goals that didn't even exist when you agreed to it.)
Hell, what if there were to be a schism in the FSF over some issue? What if there were two competing V3's? It says
For that matter, what does "Publish" mean there? If FSF "publishes" on their website something meant as a draft of V10, it seems like that counts as "publish" and could be used. What if there's a typo in a version they "publish"? Once it's "published", it can't be taken back per the above wording. Note that "publish" doesn't mean "gave a number to", "publish" means So which of those applies? Now, perhaps the legal argument is that a draft isn't a "version" of "this License", but typoed 'versions' would be, wildly different versions would be, and inadvertently-released versions would be (if someone put it on their website or any other FSF "publication" forgetting to put "draft" on it). You might even have a case where someone who works for the FSF "publishes" (without permission from Stallman) a "GPLv10" that says "copy in any way you want" for 5 minutes on their website or on FSF letterhead. A court might find that to be an FSF "publication" of a version of the license, allowing it to be used with all V2+ projects and code submitted under V2+.Basically, you can drive a truck through "V2+", which is exactly how Stallman wanted it - he has his own personal truck, which you may (or may not) agree with where he's driving it, or where he may decide to drive in the future.
Back in the original A3000/16 (16MHz) days, we timed bootup (including loading kickstart from HD) as 11 seconds cold, 7 seconds warmboot - including starting TCP/IP. I used to regularly boot an A3000+ (lab rat) into WB, TCPIP, mount 4 or so NFS volumes, and have a telnet session up in about 15, maybe 20 seconds - and I think most of that delay was NFS.
They aren't using this drug in a trial; see the Scientific American article I linked to in another comment.
My comment was in regards partly to your implication that there's no such thing as "late/end-stage diabetes". My father is very lucky and very unusual - but even he is walking an ever-narrower tightrope. Most people, even those who do a good job controlling blood sugar, don't get anywhere hear his age (or relatively low amount of side-effects). And he's had several close calls, especially over the last 10 years.
As for the possible treatment: I'm not advocating "rushing" into human trials (though early trials with related suppressives have begun). The Wikipedia page on the suppressive mentions that it's banned in human use due to toxic side-effects. The trial in Israel (see the SciAm article) uses a known-safe-in-humans suppressant that doesn't suppress as much (but also didn't seem to work well enough; I think the SciAm article said they were going to try higher doses). Also note that the suppressive only suppresses part of the immune system; my understanding is that it's less than what is used in transplants, which are the other main hope for Type 1 diabetics, and as pointed out by someone else, the protocol doesn't involve permanent suppression like a transplant - it's temporary while the immune system is "retrained" not to attack them.
The interesting thing about the new study is that it verifies that the new islets come from spleen cells.
For "brittle"/end-stage diabetics, the loss of the ability to recognize blood-sugar changes and the fact that it starts swinging without warning leads to increasing severe side-effects and often death. Those are some of the patients who've taken part in transplant trials in Canada over the last few years, which have seemed to work, but with some problems with slow rejection/die-off of the transplanted islets. They were chosen because for them, the risks of lifetime use of immuno-suppressives is lower than the risk of not doing something. The same reasoning may end up applying here, but with hopefully lower risks (partial suppression, temporary not lifetime, lower chance of rejection/re-attack of islets).
However, an effective cure would help him a lot, even at this late date. It becomes increasingly hard to control blood sugar levels, and he doesn't notice when levels wander into dangerous levels anymore. While it wouldn't reverse the damage to the kidneys, vision, etc, it would remove a major cause of further damage and life-threatening blood sugar swings. Even a partial reversal would help a lot.
Read the Nov 12th Scientific American article on this release (http://sciam.com/print_version.cfm?articleID=CE7B B73A-E7F2-99DF-3069CE90D77629FB). According to Wikipedia, Freund's adjuvant is highly toxic (http://en.wikipedia.org/wiki/Freund's_adjuvant).
Also, some very early experiments in humans have been done in Israel, using a less-toxic immune-suppressive (which doesn't suppress as much). No success, but there may be some data from it that it was heading in the right direction (see SciAm article).
This would be great if it works; my father is a 72-year-old juvenile diabetic (since age 9 - WAY outliving the probabilities), and my cousin once-removed on my father's side is also a juvenile diabetic (age ~23, diabetic since ~19 or 20). Many type-1 diabetics die before they're 40, often with severe complications.
Lemmings (original or Lemmings 2) had simply FANTASTIC multiplayer on an Amiga with two mice. The fun of digging a hole that starts draining all your opponent's lemmings into a pit while they're busy elsewhere was amazing. It was so good, at an Amiga DevCon a bunch of Commodore employees and a few developers all dressed up as Lemmings as a surprise to the attendees and "acted out" a Lemmings skit.
MULE was groundbreaking on the C64. Simple as sin on the surface, but plain fun for 4.
Gauntlet - fast and fun on the Amiga for two or 4.
One last one that launched a genre of god-games, and was a ancestor of games like Black and White - Populous. In some ways similar to a much fancier, 3-d Lemmings.
Commodore didn't buy MOS until after that (a year or so later, can't find the exact date). MOS was in serious financial difficulty around that time due to the bottom falling out of the calculator market.
The stock 1541 was roughly the equivalent to 2400 baud or a bit faster; around 1 sector/second. Note that most programs/people used "fastloaders" that programmed the 6502 in the 1541 and upped load speeds by 3-10x.
A2200 was correct also; if I remember there were A2200 and A2400 monikers for slightly different variants before it became the A4000.
Tripos... Some aspects of Tripos were good. The big problem was mostly BCPL, and how it interacted with the rest of the system. Two generations before C, it had some problematic issues, like the global vector. "global vector" was a kindof-replacement for libraries, or if you prefer a equivalent to the standard C library, but as a shared lib. Pointers not being native pointers (pointer to an integer was a machine pointer shifted-right two) was a real thorn too.
BCPL on a 68000 had real issues, in that they set the stack up "upside-down" by normal 68000 standards. This caused all sorts of problems integrating it with system libraries, and for user code trying to call it (dos.library calls all had to go through a translation layer). It also had real issues with strings, due to not liking pointers to bytes much. If Tripos hadn't been in BCPL, it would have not been a huge issue.
More documentation wouldn't have solved these problems really (and even Commodore had essentially NO documentation other than the mostly uncommented code). I had fought these as a developer trying to write a replacement shell before joining Commodore. And in reality it was still Tripos, just ported to C and with a big "translation" layer in ASM so that existing BCPL code could run (i.e. BCPL now went through translation functions, and C/ASM/etc went direct). Plus a bunch of additional interfaces for things like replacing the shell, etc. Look at the docs on the dos.library interface differences between 1.3 and 2.0 for an example.
The ramdisk, for example, was dramatically easier to work on and extend (and speed up) after I rewrote it. The ramdisk got quite fast. FFS already existed in ASM (and also handled OFS), and in reality didn't lose any real-world safety (disks already had per-block checksums, adding another layer of them to the FS just slows things down (hugely)). Many calls it didn't matter for, but for some the speedup was dramatic. But most of the speedups were from redesign of bigger, complex pieces. The biggest BCPL->C/ASM speedups were in really simple calls that didn't do a lot, but in BCPL invoked some heavy translation layers.
MEMF_PUBLIC - of little use unless something was checking that from the start. And it would only have allowed protection, not per-process address spaces. VM would have been useful; people did work on trying to implement VM without per-process address spaces or inter-process protection. It kinda-sorta worked, but not really. If you wanted VM and process protection in an exec-like message-passing microkernel, QNX was a lot closer to what you'd have designed (it was later, too).
As for the need for VM - I did all my development on Amigas; when I finally had a machine with 16MB of ram (which wasn't until a while after I had an A3000T), I can't remember getting below 4MB free, and rarely below 8MB free (until I started running an early version of NCSA Mosaic to browse the web in '93-94). That's not to say people couldn't have used it, but it wasn't as critical as it is today, when my WinXP machine is using 600MB (out of 1GB), and 275MB of that is a browser (which a huge number of tabs open, and which has been running for a month), and my Linux box is using around 1GB of 1.5GB (again, a big browser running for weeks and X are the hogs). Note, however, that neither machine is paging, and things slow down a lot when you do. But it is useful at times, and the #1 use of paging:
When an Amiga ran out of memory, it remained almost totally stable, because we (and many of our developers) were pedantic about checking returns and handling error paths, and we had LOTS of tools to make it easy to stress-test programs and the OS for this. An awfully high percentage of programs now simply call exit() if they run out of memory... assuming they notice at all.George (designer of A500 and A1200, admin of cbmvax) owned a former railroad station around 30 miles away, and mostly lived in his cube 24/7. Once in a while he'd go home to pick up mail, see if the water heater had burst, etc. Most of the time I was there he didn't drive, just rode a bike.
He paid off the mortgage and didn't even notice, just kept sending checks. They started having to send him checks with notes asking him to please stop. I think it was after he finally stopped that he started just throwing his checks in a drawer (too much hassle to go deposit them).
His primary leisure activities were admining cbmvax, USENET, inline rollerskating in the huge warehouse/production floor, reading SF, and railroads.