I'd have to go look up exactly when copyright was conceptually founded (I believe someone posted in a article a couple of months ago that it has existed since the days of the Romans conceptually that puts it back into at least 1000AD or so), but it is explicitly mentioned in the U.S. Constitution. So it's been around since the late 1700's. I believe John Locke wrote about it prior to the 1780's or so. Johnson and Johnson is currently suing the American Red Cross over a trade mark registered in the 1890's. The U.S. Patent Office has been around since around the time of the founding of the United States. For instance, Abraham Lincoln was proud of the fact that he was a patent holder.
So at least two such concepts pre-date things like Women's Sufferage, or the concept that African American's shouldn't be held as Slave's in the South in the United States. Given those dates, I'm reasonable confident there is no one alive who remembers before the three concepts of Intellectual Property existed (alright, there might be a handful alive from the trademark date I quote, but I think trademarks pre-date the early 1890's, I'm just too lazy to go find out when).
So while you refer to them as "new"... You can only mean new in comparison to concepts like "bipedal humans that walk upright" or "humans forming civiliations and moving from hunter gather to agricultural modes of survival", and still be intellectually honest (or grossly uninformed on the concepts).
We have the works of Shakespeare and Newton, because they eventually fell into the public domain. Now, if you want to argue that current U.S. copyright law is just stupid, I'll back you wholeheartedly. Unfortunately, as a citizen of the U.S. and the U.S. being a signer of the Berne Convention, means that Copyright Law can't be made to be sane. It could however be lowered to limits of the Berne convention, and then at least copyright would expire in 50 years after the work was published.
Assuming that Disney isn't successful forever, Mickey Mouse and Donald Duck will fall into the public domain. The entire music catalog of the Beetle's will be in the public domain in Britain within the next 20 years (they refused to increase copyright past 50 years recently). The original works of Einstien, Dirac, Godel, Turing, Plank, Hemmingway, Authur Miller, Tennesse Williams and others should eventually fall into the public domain (contemporary notable scientists and and playwrights). Hopefully within my life time (the next 40-60 years). The works will be preserved as long as it takes to get them into the public domain. The sad part is that 99% won't be. Only the things that were recognized as great works at the time will be. Who knows, maybe Shakespeare had a truely great pupil lost to the sands of time. It'd be far easier for libraries and other archivest to preserve if they didn't have to worry about copyright being an issue. It'd be easier to stand on the sholders of giants if I could use giants who were alive during my lifetime...
Any chance you have a cite for that usage of "fold"? I've always thought that it was a straight multiplication factor. Looking in online dictionaries I don't see any explicit version of that usage. However, several point out that the "ply" in "multiply" is derived from "fold", so it'd seem reasonable to thing the grandparents usage is correct. I've always thought "a 2 fold" increase and "a 10 fold" increase meant to multiply by 2 and 10 respectively. However, I've been wrong before, and be interested to see the background of "X fold".
I believe your answer to number 3 is wrong according to MySQL AB (the corporation that owns the copyright of MySQL). They changed from the LGPL in version 2.x to GPL in version 3.x. See http://mmmysql.sourceforge.net/. I can't find the page right now, but at the time all this changed it was fairly clear that MySQL was tired of losing the licensing fees. Because the Connector was LGPL, you could embed MySQL in virtually any application and never need to purchase a license. The LGPL had little burden on a commercial use (essentially, ship the source, or offer a copy of the source to any one you shipped the connector to). The change to using the GPL should mean that your application should be bound by the GPL'ed. While your interpretation might be correct, the copyright holder disagrees with you. Even if your right, that could be very expensive.
As for 1, I'm not sure you can relicense it, I'd have to read the licenses closely, but you could release your stuff under GPLv3 and include any requirements the Apache license has (including the notice files, not removing the copyright notice, or the waivers of any warrantee). Effectively you can do what you want.
If they what you describe, that is problematic. If however, a search for "Pepsi" brings up a link that is for "Coke", but the link isn't labelled "Pepsi" I've got no problem with it.
I'd suggest two things, first: it's not well defined what is and isn't copyright infringement with respect to source code. I suspect that's double true when the GPL was originally written (circa 1984 for the precursor licenses of the GPL, the Emacs and GNU C Compiler license, I think the GPL v1 was in 1987 or 89, while GPL v2 was 1991). Copyright infringement for books, pictures and music aren't completely well defined. If I took 5 lines from a book, is that infringement? If I took the basic plot, but re-wrote everything. How many seconds of sampled music is a copyright violation? How many long a sequence of notes can be re-used before it's copyright infringement. I'm reasonable sure each of those has to be decided on a case by case basis, and that the court is very inconsistent.
Second, it's very clear that they'll take you to court if you violate what they deem a derived work. My hunch is that their definition of infringement is vastly looser then any court in the U.S. would ever interpret it to be. If you even came close to any usage allowed by the GPL on code while using a "All Rights Reserved" license source, you'd be infringing on the copyright. I guarantee every interpretation of a derived work by a U.S. Court that is disallowed for GPL'ed code would also be disallowed for "All Rights Reserved" (which are essentially the only rights you have if you don't comply with the GPL). They spell out for the court the terms, so it's clear, if they don't follow the following terms, they have Fair Use rights. Just like I could license my code to say only guys born on days of the months that are prime can use it, they can dictate the terms of their license, and what they consider a derived work. As their definition is vastly looser then any standard definition used by a court, it's not like the court will use a stronger interpretation. It is also very difficult for someone to argue that they felt in compliance of the license, while using a different definition of "derived work".
If you feel that analysis is wrong, please point out something that would be allowed by a courts interpretation of "derived work" on code which was licensed as "All Rights Reserved", while being disallowed by the FSF definition of "derived work" in the GPL. Anything under fair use doesn't count, as that is allowed on any piece of source regardless of the license.
'Unfortunately, the FSF has seen fit to encourage such nonsense, with the "anything that links to this is derived" claims they put in the GPL.'
It's a fairly reasonable assertion, they define the terms under which you are allowed to use their programs. It's not meant to line up with the legal term "derived work" that is fairly well established in most contexts (Software is one context where I'm not sure there is a well defined precedent for what is and isn't a derived work, see the bullshit look and feel lawsuits), but it's their working definition of "derived work". If you do what they consider a "derived work", you have none of the rights granted to you via the GPL (specifically the rights to modify or distribute the software, I believe technically you can still run it). It's not like you could ship their software or modify their software and distribute it without the consent of the Copyright Holder. The GPL merely explains explicitly what the copyright holder thinks is acceptable terms.
If they had spelled out that "here are the things we conditions that we consider not to be copyright violations", and spelled them out, and dropped the words "derived work", it'd all have the same legal weight. It's flat out copyright infringement to ship the Linux Kernel in binary or source form if you construct anything the GPL defines to be a "derived work" and distribute it. It's not like a court is going to find in favor of the VMware because of fair use rights. The only way VMware will win this is if they prove that it's a well separated piece of functionality that works outside of the standard Linux kernel, or only uses a well defined subset of the kernel API's that are common to numerous UNIX operating systems (what Linus would approve of essentially, according to several widely quoted public statements).
First, RHEL did cause a fairly big splash. A number of people were fairly vocal, and called for the downfall of RedHat. However, there is one critical difference: RedHat freely gave away everything needed to build RHEL to anybody who wanted to show up and use it. It sure sounds like MySQL is not giving away the source or binaries to anyone. They are following the letter of the GPL (which is very good!). But they are not playing in an open space. To the best of my knowledge, RHEL didn't get particularly upset if you gave away the binaries (as long as you took the time to strip off their trademarked materials), and were clear that it wasn't RHEL. I don't really understand why MySQL doesn't just put the source out and let folks build their own binaries (which I think would cause most of this to go away, as there'd be a sourceforge project that distributed the binaries). In the end, their client license will ensure they get paid by folks who use their software in any non-Opensource project (They GPL their client, which is a problem for anybody who uses it in a proprietary application). That is however, their right, and I support them to do it. I just don't use their database at work for anything because of it. (I have a historical preference for PostGres over MySQL, and the change from LGPL to GPL on the client has made it a moot point to re-evaluate the technical merits).
I'd suggest that the article didn't say this is "the only way to go past the industrial revolution", here merely pointed out that this might have been a driving factor behind the one in England. I'd say that Africa's situation is strong evidence that they haven't yet had to get over the hump of industrialization. Clearly there's some reason why England has, and a Third World nation hasn't. This line of reasoning seems at least plausible.
Just like historians who say that the Nazi's committed genocide against the Jews, aren't saying that every political party world wide has an interest in genocide. Which is the logical leap you seem to want to make.
You can read whatever you want into it, but it sure looks like your generalizing and projecting something that you wished it said. I see no implication that starts with what the author has to say, and ends where you suggest (everything after the word therefore in your post is you looking to pick a fight, not you being rational or logical).
After more thought, on average you get everything out of Life Insurance you pay into it (minus the profit of the company). It's not uncommon for Life Insurance to be a net loss over the long haul, relative to conservative investments. The only reason to have life insurance is in case your dependents need the money to survive in case of an untimely demise. Life insurances is really nothing more then a group of people pooling their money and having an agreement to take care of the survivors. It's just that the folks who manage that money also get to keep a small percentage as their profit. (It's slighly more complex in the case of term life, but in general that's the concept behind life insurance). Health Insurance is likely to have no financial windfall to anyone, the only benefit to you is well, your health and extending your life. I've got different scales of what I'm willing to accept on those two things. You and I might just disagree on that, which is fine. But that's one mans perspective on it.
Having life insurance or not in the general case, won't make a 5-10 year difference in your life span (over a large population I'd guess that is a reasonable estimate, I wouldn't be shocked to see that number be as high as 20-30 for specific at risk groups). Besides, health insurance I see as good thing for society and the quality of life. Life insurance is a good thing for your family in case of a serious accident. My life insurance pays off when I'm dead it doesn't really affect my quality of life in a measurable way. I've got lots of control over my choice of job, and quite a bit of my lifestyle. However, there are many things which I have no control over that are easy to measure, and statistically important. I'm fairly sure if you examine my DNA, you'll find I have several predispositions (no idea what they are, but I'm sure I have them). Smoking might be something you can say is a legitimate choice. Weight, I'm not so sure there are a number of legitimate medical conditions that make maintaining a healthy weight very difficult. I'm also really sure, it's none of my employers business how much I weight, or what various other specifics of chemical balance is. Shit it'd be a HIPPA violation if a nurse gave that information to anyone outside my immediate family. There are a lot of reasons why those values can be out of whack. Some of which are legitimate choices, some of which are the entire reason we have insurance. Until they are capable of differentiating between those fairly it sure seems like hurting the precise people that insurance is meant to help. I pay a disproportionate amount of a lot of things in taxes and other fees that I tolerate because they are good for society as a whole, and thus indirectly good for me. Life Insurance I don't feel that way about. It's directly good for a specific small group of people. If folks fall on hard times because of a death without insurance, society has other more general ways of helping such folks out.
Besides that, my other gripe about health insurance, is that insurance and litigation are the two things that drive the cost of health care up. The other is that medical providers charge crazy amounts for common things to cover up the costs of really expensive things insurance won't cover. I'm positive that there's no good reason for Aspirin to cost $50, but it does if you read a hospital bill closely. I'm highly unlikely to abuse my insurance or to sue anybody, yet I get to foot the bill for everybody who does. The pure amount of paperwork and headache insurance causes for my General Practitioner is why it costs ~$150-200 a visit. Where it's not uncommon to see a doctor for $30-40 if I have no insurance. I enjoyed an article about a move by a number of doctors to never accept insurance. It's a flat fee of $40-50 dollars, no insurance, they'll treat you just like they treat everybody else. Cash up front, if it takes 60 minutes and 2 tests great. If it takes 2 minutes and they send you out with a placebo, fine. But the fact that I'm billed $100 for the 5 minutes it took a nurse to flush my ears is assine. It's not a complex procedure, I'm fairly sure I could buy or fabricate the equipment used to do it for less then that. I'm fairly sure just about anybody could do the procedure. The only reason to charge so much is to cover the costs of other more expensive procedures, or the risk of a multi-million dollar lawsuit if something happens to my eardrum. Because it costs so much, nobody pays for anything out of their pocket to keep expenses insurance costs down. I like Health Savings Accounts, where you can put Tax Deferred money into an account up to the deductible amount, and have a high initial deductible of ~$2K-4K per year, folks who rarely need health care can carry that money over year to year, and saved a substantial amount on my premiums. It's essentially a relatively easy way to motivate consumers to choose efficient ways to deal with their medical costs, while still covering major costs in cases of disaster.
Have you considered a future in marketing or politics? Seriously, there's absolutely no difference between charging people extra who have more risk. If you frame it as a discount for good behavior or as a fee for bad behavior, it's doing absolutely the same thing. The only difference is how you present it. That is marketing, and that sort of thing is good for convincing people of an idea.
There's absolutely no way to differentiate between those two cases. I've got no problem with them basing things like this on a statistical basis, but at some point as a generally healthy person I'd just opt out. BMI is a bullshit measure, when I'm at the top weight that is considered "healthy" by BMI and you'd think I was a recovering from an eating disorder. I can't imagine what it'd take to get to the middle of that range (6'6" and 185 is pretty darn thin, I'd nearly have to give up eating to maintain that weight even while working out regularly). I weight about 60lbs more then anybody every guess I weight. Most of the folks in my family do the same thing. Nobody ever guesses close to what any of the men in my family weight.
Insurance is a system where we all pay in to help spread risk across everyone. This is the first step towards just having you pay in what you cost after they skim off their profits. If that's what they want to do, that's fine. Give me some catastrophic coverage for a major emergency, and let me pocket the ~$150/month difference in cost (and more like ~$600 once I negotiate with my employer to get the 75% of my insurance they pay for me). I'll happily invest that money, and pay for my own $4 antibiotic prescription I need once a year, along with the $40 for the doc visit. I'm generally very healthy. I don't have a wife or kids. I lose a tremendous amount of money to the insurance industry every year. I'm very low risk. Anything under $10K would be easy for me to pay cash, anything under about $50K that allowed me to continue working, I could easily finance. Out to about $100K I wouldn't be particularly worried about my ability to pay for it assuming I could continue to work or some form of short/long term disability covered a reasonable percentage my salary. Especially if I got another $600 a month to save for just such a situation. But then I'm freakishly good about saving for a rainy day.
Hmmm, I live in Nebraska and I'm fairly sure that with the current tax and tuition situation would have left them with a number of other less then desirable options. I graduated from the University of Nebraska at Omaha. I believe both institutions have similar tuition rates and increases.
Currently tuition, is about $120/hr. When I started there in 1995 it was around $60/hr. All of those are the in-state prices. In all probability, this was seen as a way to fairly have the folks who are costing the most to absorb the costs more efficiently (Maybe not the business dept).
Currently the University of Nebraska has rules about who has to be admitted if you graduated from a state High School. Essentially, if you score a 20 on the ACT and graduated from an in-state HS, they have to admit you.
The costs associated with the University are going way up. I wouldn't be shocked if some of this has to do with the University's problem of hiring and retaining top Engineering, CS and MBA PhD's. They can easily command a great deal more money in the regular job market then say an English Professors salaries.
The alternative is to make it so that nobody but the rich can afford to send their kids to the state school they help subsidize.
Yeah, maybe if I got several of folks together I could negociate. However, I've done that sort of thing before. One guy doesn't want to pony up the money that month. Or something else has come up. The rest of you get to cover for him, or risk having your service cut off or paying late fees. I'm much happier to pay a little extra and disentangle myself from my friends finances or have to deal with cutting him out of the LLC.
Next, what I do find very irritating about the phone companies lately is that they way overprice the phones. Then if you don't sign the contract, you get to pay them the profit up front on the phone. Then you get to pay extra every month, because they won't lock you in with a good rate. In my experience, they also build the phones to last about the length of a 2 year contract. I still have my circa 1997 cell phone. Damn thing still works, and runs like a champ. It's just the battery is long dead, and I can't find a replacement. It had a few other quirks (like you couldn't stop it from ringing without answering it). But the phone took an absolute beating for 5 years. I've had 5 phones since then, and I'm generally much easier on them. They die within a year or two.
Finally, I'd really like the phone makers to get together on a system of making phones fairly uniform in UI. I've had 6 phones. Each of them had 2-3 really nice features... But no phone has had the list of all of the features I wish it had. My current list:
100 speed dials, I press and hold a digit for 2-3 seconds, the phone dials a one digit speed dial. I press one digit and release, followed by hold the next one for 2-3 seconds, it dials a two digit speed dial.
A clamshell that no matter what button on the outside is pressed the phone does nothing at all. Have it be modal based on when I close the phone, or how long the phone has been closed (60 seconds with no button presses means the phone is in my pocket).
People have multiple home, work, and office numbers. Do not have a stupid menu system that only allows one phone number per person of each type (even if I can put in as many "Others" as I'd like, it's nicer to have them qualified).
Allow me to put a 3 speed dials for the same person, do not make me pick their home vs. cell phone for a speed dial (far more useful once you have 100 phone numbers).
When I store a number, allow me to override an existing one without having to delete the damn thing first.
Sell me a phone without a damn camera, they are a serious problem if you work for a DoD contractor or near a classified space.
Add sync software that'll work on Linux, or at least with open formats and standards. I wish I could sync my phone with my iCal settings from my Thunderbird + Lightening.
Make it simple to cancel another incoming caller without hanging up on a current phone call.
When I'm on the phone, and I ignore a call, I really don't need the obnoxious voice mail alert right in my ear, honest I understood that someone called.
Make a speaker phone that I can choose the latency of the pickup and drop off... I've had one too many phones that people couldn't hear the first or last syllable I said, rendering the speaker phone useless.
Integrate more and better contact information (hopefully after you have sync, because believe or not, some of us can't stand using a phone to type in information, I'm 6'7" and have large hands and fingers, typing on a phone sucks).
Make the cables to connect the phone be storable with or inside of the phone. It sucks to lose a $40 cable because you it's useful and carry it with you all the time, or to never have the cable when you need it, for fear of losing it. Or make it a 4 inch cable, or a cable I can buy at any damn best buy, or that my friends or office are likely to have handy so I can extract information from my phone.
Store more then 10 incoming, outgoing, and missed calls. Store the start time, and the duration. Oh, and if I call the
If you read the actual mdadm code (I used v2.6.2 from http://www.cse.unsw.edu.au/~neilb/source/mdadm/). Looking at Monitor.c 487 of the Monitor function, it "polls" the file. Waiting 60 seconds (by default, you can set the delay lower). However, if you look at the mdstat_wait(int) call, you'll notice that it's doing a select on the "/proc/mdstat" file. It'll break out of the select if fairly quickly. Force a mirror break, and if you have it setup to get an e-mail, you'll have the e-mail with in your in box in seconds. There's no 1 minute or 15 minute lag.
I'd swear that at one point it used an ioctl to extract all of this information, but it's been ages since I actually read the mdadm tools source... Heck it might have been the precursor tools... mdtools or mdctl or whatever it was called before.
I've always said that FOSS is great for commodity areas. Thinks like standard infrastructure (the OS, simple routing, standard Office applications, web browsers, IM, e-mail, to a lesser extent database, HTTP, FTP, SSH, e-mail, DNS, etc and other servers). Just about anything that is standard out of the box functionality people expect from a computer is great to have as an Open Source software. In a lot of ways, even Microsoft software is free as in beer for most folks. Other then Office Software, most of above is "free" if you don't count what the OEM paid MS. The problem with the OEM fee is that it's like sales tax at the grocery store. It's hard to avoid, and few people consider it.
Other software can be FOSS, but in the general case unless your Sun, Microsoft, or IBM it doesn't make a lot of sense to compete over making infrastructure software. Oddly enough, even IBM and Sun are moving their major infrastructure pieces to Open Source. Where I work, I've been pushing us to start using Open technologies for all infrastructure in our software, if not using Open technologies for all our applications.
You seem to be describing a "calling convention", not an "ABI". Normally, specific hardware and languages had specific ways it expects to pass parameters. Most of the really odd looking things near the return types of a Win32 method are specifying the calling convention for passing arguments and return values to and from the caller. An ABI is something that a library, or specific piece of software has. Subversion has a binary compatible ABI and API guarantees that are tied to the version numbers. I believe that 1.2.x is forward and backward API compatible with 1.2.y. It is a forward API and ABI compatible from 1.X to 1.(X+n). When you change major version numbers, all bets are off.
A compatible API, says: "If this code compiles from source with this API, it will compile from source and behave the same will any compatible API". This is not discussing forward vs. backward compatible API's. A forward compatible API says that any newer version, a backwards compatible API, says you can re-compile the same source against an older version of the API. Generally most folks only worry about forward compatible API's, not backward compatible. Generally a backward compatible means you have a compatible ABI, or that something that was previously a macro became a function or vice versa.
A compatible ABI, says: "If you have a compiled binary, you can re-link against any compatible ABI and it will behave the same".
To illustrate the difference, you can change a function to a macro, and still be API compatible. However, unless you leave the function implementation in, it is not a compatible API. Or if you change ENOSUCHFILE from -12 to -10 (Linus has stated that he'd have changed specific error numbers to match SCO UNIX so that some of the emulation modes would have been easier, but he didn't realize it until it meant an ABI incompatiblity so he never did it). It's API compatible, but not ABI compatible. ABI compatibility is extremely hard to maintain, and keep the code maintainable (glibc and other libraries do some nasty tricks to ensure ABI compatiblity, and it makes things much harder to read). API compatibility isn't that hard if you try. In the general case, always add interfaces, never remove them. Generally speaking that is a fairly safe way to be forward compatible API's, and reasonable stable ABI's.
In general stable API's are good enough as long as you have source for everything. They are a good middle ground. Compatible ABI's have lots of value to users, but lots of costs to the maintainers. Linux has lots of userland ABI stability, but no internal ABI or API stability.
I believe the mdadm tools on Linux actually block the system call until on a change in the status. I didn't state that clearly. So you get immediate notification, RAID re-builds. I wrote something that checked every 2 minutes on all of our RAID devices (Nagios and scripting are great). Generally mdadm and it's alerts informed me once I had those setup, soon enough that could go in and disable the Nagios alerts prior to it noticing.
I'm shocked the raid tools for OpenBSD aren't better then that. Not a dig at it, OpenBSD generally prides itself on exceptional tools. OpenSSH, CARP (their replacement for VRRPD), their firewall tools and everything else. Linux has a system call that can be used to monitor the status of a RAID array. It can kick off an arbitrary command, including starting up recovery and/or e-mail alerts. Technically the system call doesn't, but the mdadm tools that use the system call can.
I really hope somebody replies telling me, I'm an idiot and that OpenBSD has exactly such tools. Well and they really exist, as opposed to the clever slashdot behavior of telling me I'm an idiot and be completely wrong.
Hmm... your history doesn't match up with Wikipedia's, read the last paragraph or two of the section of Depth fail. If you can cite sources for the version you present, that might be of interest in updating. I don't have an interest either way, but I dislike patents as currently implemented. Thus any abuses of them are of interest to me.
The Wikipedia article leads me to believe that Creative didn't do anything wrong in terms of current legal framework. It sounds like Creative presented it as a concept prior to Carmack inventing it.
Anyone who knows Red Hat can pick up Ubuntu in less than a day. And Slackware in another day. And Gentoo over a weekend. At which point, you pretty much know every distribution out there.
As a Linux user since the summer of 1996, I can assure that one cannot learn "Ubuntu" in a day. I'm a Linux power user. I use RedHat to the hilt. I can use RPM and YUM, and I find apt cumbersome and a pain. I haven't spent hours using it, but I probably spent 30-60 minutes playing with it. The differences to a professional Linux admin are significant. Yes, I can learn all the differences, but it'd take a lot longer then "a day" to get as good and competant on any old Linux distro as I am on a RedHat machine. I've used Slackware, RedHat, Caldera, Debian, Novell, Turbo Linux, along with LFS distributions. I've probably used others for brief periods. At some point around RedHat 7.2 or so, I decided that keeping them all straight was a total waste of time. It was a hassle to do security patches, it was a hassle to make configuration changes (I believe Slackware still doesn't use PAM, so all of the authentication had to be handled special). It was a huge pain to deal with various other configuration issues (Slackware used a totally different set of startup scripts then RedHat did). We standardized on exactly one Linux, RedHat Linux.
I used to regularly use advanced features like Kickstart to do unattended installs. I administered and ran about 30 machines at my last job, installing and running pretty much every service that comes stock on a RedHat Linux machine. I've reported problems and recommended solutions to the folks who run Whitebox Linux and CentOS. While I can actively use any old Linux box, there are numerous painful problems with switching distributions. Going from RedHat to Novell I find a pain. They are fairly similar, they use RPM's. The two use roughly the same kernel (on the versions I was using). However, all of the advanced management tools are different. It took me an hour or two to figure out how to have Novell's X utilities to auto-detect the video settings after swapping out monitors, I can do that on a RedHat machine without thinking. It took a while to figure out that YaST is how you do everything on Novell, so looking for "system-*" commands from RedHat was a waste of time. It's annoying that "chkconfig", and "service" aren't installed on all Linux machines. It's a pain that folks shuffle which files end up in/etc/, that/etc/profile.d isn't always in the same place. That not all of the network config is handled the system. That not all distributions use/etc/sysconfig the same way. In a serious emergency where time was money, I'd be in a serious trouble with Novell or Debian, where as with RedHat, I'd have no problems at all.
In the sense that you are using "can pick up in a day", I can go from a RedHat Linux distribution, to FreeBSD, OpenBSD, Solaris, or AIX and nearly as comfortable as I am when I work in a Debian or Ubuntu distribution. Also, it's nearly comical to see in this thread the assertion that any code that will run on one Linux distribution will run on another. I've had problems building the same source on two different versions of the same Linux distribution, yes compatibility libraries are very nice and they eventually solved my problem. But it's not like installing source and building it from scratch is a panacea, a lot of code doesn't like to run, or run well out of the box on different versions of RedHat due to different locations of various libraries and other fun stuff. There's a reason that LSB exists, and it ain't because all Linux's are just alike and moving from one to another is a snap. Just go look at all of the patches that RedHat applies to vanilla source in source RPM's, go look at how different they are then the ones that other distro's apply.
Sure, as long as they all run some form of X, and have a reasonable Window Manager or Desk
I believe your understanding to be very wrong. The GPL is always intended to play in the space of "I'm giving you my source code, if you use it, you have to give up yours if you link to mine". If you are correct the GPLv3 is wrong-headed and nobody should use it if that's not the space it's covering. Lots of folks who used the GPLv2 would be up in arms if GPLv3 isn't nearly identical in spirit. As most GPLv2 code will be allowed to update to GPLv3 automatically if they followed the FSF instructions for applying the license. If you believe I'm wrong, if you could cite the paragraph that leads you to believe that it's legal to link to GPLv3 code and not re-distribute the source, I'd be highly interested.
(LGPLv3, Apache, etc)
I'm unaware of an LGPLv3 (I believe there is a discussion about constructing one, but the most recent one I know of a public test available for is LGPLv2.1, which is essentially changing "Library" to "Lesser" from the LGPLv2).
is that most of the licenses in the third group (LGPLv3, Apache, etc)
Apache v1.1 or v2.0 are *NOT* in the third category as you imply. They are in the first category. You can take Apache 2.0 code modify it and redistribute it. No one has legal recourse to get you to stop distribution, or produce the source (at least according to my reading of the APLv2.0 Section 4). The only requirement Apache 2.0 has is some specific patent things, and that you distribute a copy of the license, that you leave the copyrights in tact, and that you acknowledge that you used that source. There is no provision for producing the pieces you modified. There is a provision for explicitly documenting that you did modify them if they are being distributed. With the LGPL, the copyright holder could sue you for copyright violation if you fail to produce the modified source to the modified binaries. The EPL and CPL have somewhat similar provisions.
I believe the GPLv3 has various options and provisions to avoid people writing the GPL + patent provisions, or GPL + attribution provisions. Thus allowing folks to make the GPLv3 compatible with other far more permissive licenses like the Apache 2.0 that have one small condition that the GPL didn't (which is an extra condition, and thus GPLv2 incompatible). The FSF has held that Apache 1.1 and Apache 2.0 are incompatible with the GPLv2, I believe the GPLv3 will make them compatible + have the GPL v2 provisions + have more patent and IP provisions). The Apache folks have said that GPL and Apache 1.1 and Apache 2.0 are compatible, that the additional provisions are insignificant.
The GPLv3 is about covering the second type of license.
Right now, there are roughly 3 types of OSS licenses.
Do what you want, but I own the copyright. (BSD, Apache 1 & 2)
I share my code, if you do anything with this you have to share yours the same way I share mine. (GPL)
I share my code, and if you re-use it you have to use it you have to tell people mine exists and thing you add you can pick the license. (Numerous licenses, including the CPL, EPL, LGPL).
The article states "look at the proliferation of licenses", as a sign that the GPL isn't filling a need. The simple facts are that the first two licenses are pretty much in the bag. Nobody writes new licenses that attempt to the accomplish the first two. Pretty much to a person, everyone uses Apache 2, BSD or the GPL to accomplish those goals. If you start looking down the list of other one-off licenses that are for OSS. Those are all about filling the need in the third item. If anything, it could be said that the LGPL is "failing". It isn't the "one true license" to accomplish the task. Essentially the proliferation of license's is about finding a "share and share alike" that can exist in a corporate environment. Where the core technology can be shared and developed by many folks, while the extensions and non-core pieces can be value-adds that are solve for money.
Greg Stein's a brilliant guy, and one hell of an engineer. But I think he's living in his own little world here. Lot's of folks like and enjoy writing software under the premise of the second type of license. Some folks do it under the first. In the end, the collaborative effort will virtually always win out. So in a lot of ways it doesn't matter if you use a license from the first or second group. That's why Apache has never been taken and had a closed competitor that is more used. Sure some commercial products are based on it, but none of them will ever quash Apache out of existence because they are so popular.
All the action is how to have an open source commercial license. The LGPL has a few terms that are a bit harsh on business, and have little to say with respect to patents or trademarks. In this day and age a license must address those.
Upon virtually every microsoft patch Firefox and Thunderbird require me to reset my default browser and e-mail client. My guess is that this is a simple security mechanism. Hooking into the startup of e-mail and web software would be a good place for a virus to hide... Then just startup the older software and life is good. It'd be a good place to hide malware startup that is outisde of the normal places that are checked for issues.
Just a point of order, if the US Constitution were not a living document, there'd be these 10 things right at the end that'd be missing. Most of us refer to them as "The Bill of Rights". The constitution was ratified *WITHOUT* those amendments if I remember my history correctly. We've got a long ways to before I'd say I'd rather have the original constitution as written without the first 10 amendments.... It's a very good thing that Constitution is a living document that can be changed, and re-interpreted as appropriate. A number of just vile interpretations have happened over the years, most of them were the first interpretation.
However, as a pretty damn safe rule of thumb, no system is going to run faster on equivalent hardware after being virtualized.
Oddly enough, that's not as true as you'd think. I know that HP and IBM both had projects to virtual hardware (not in exactly VMWare style virtualization, more like a JIT optimization) where software ran faster after being virtualized. By about 20% if I remember correctly. HP's project was Dynamo and IBM's was DAISY.
Virtualizing a CPU at runtime in a JIT-optimization fashion can actually lead to a significant speed up. Some of this might have been pulled into CPU's by now, but I'm always surprised that more technology isn't moving this direction. Ultimately optimization at runtime has more information then at compile-time, and thus can apply optimizations not otherwise available.
Slashdot Mirror
I'd have to go look up exactly when copyright was conceptually founded (I believe someone posted in a article a couple of months ago that it has existed since the days of the Romans conceptually that puts it back into at least 1000AD or so), but it is explicitly mentioned in the U.S. Constitution. So it's been around since the late 1700's. I believe John Locke wrote about it prior to the 1780's or so. Johnson and Johnson is currently suing the American Red Cross over a trade mark registered in the 1890's. The U.S. Patent Office has been around since around the time of the founding of the United States. For instance, Abraham Lincoln was proud of the fact that he was a patent holder.
So at least two such concepts pre-date things like Women's Sufferage, or the concept that African American's shouldn't be held as Slave's in the South in the United States. Given those dates, I'm reasonable confident there is no one alive who remembers before the three concepts of Intellectual Property existed (alright, there might be a handful alive from the trademark date I quote, but I think trademarks pre-date the early 1890's, I'm just too lazy to go find out when).
So while you refer to them as "new"... You can only mean new in comparison to concepts like "bipedal humans that walk upright" or "humans forming civiliations and moving from hunter gather to agricultural modes of survival", and still be intellectually honest (or grossly uninformed on the concepts).
We have the works of Shakespeare and Newton, because they eventually fell into the public domain. Now, if you want to argue that current U.S. copyright law is just stupid, I'll back you wholeheartedly. Unfortunately, as a citizen of the U.S. and the U.S. being a signer of the Berne Convention, means that Copyright Law can't be made to be sane. It could however be lowered to limits of the Berne convention, and then at least copyright would expire in 50 years after the work was published.
Assuming that Disney isn't successful forever, Mickey Mouse and Donald Duck will fall into the public domain. The entire music catalog of the Beetle's will be in the public domain in Britain within the next 20 years (they refused to increase copyright past 50 years recently). The original works of Einstien, Dirac, Godel, Turing, Plank, Hemmingway, Authur Miller, Tennesse Williams and others should eventually fall into the public domain (contemporary notable scientists and and playwrights). Hopefully within my life time (the next 40-60 years). The works will be preserved as long as it takes to get them into the public domain. The sad part is that 99% won't be. Only the things that were recognized as great works at the time will be. Who knows, maybe Shakespeare had a truely great pupil lost to the sands of time. It'd be far easier for libraries and other archivest to preserve if they didn't have to worry about copyright being an issue. It'd be easier to stand on the sholders of giants if I could use giants who were alive during my lifetime...
Kirby
Kirby
I believe your answer to number 3 is wrong according to MySQL AB (the corporation that owns the copyright of MySQL). They changed from the LGPL in version 2.x to GPL in version 3.x. See http://mmmysql.sourceforge.net/. I can't find the page right now, but at the time all this changed it was fairly clear that MySQL was tired of losing the licensing fees. Because the Connector was LGPL, you could embed MySQL in virtually any application and never need to purchase a license. The LGPL had little burden on a commercial use (essentially, ship the source, or offer a copy of the source to any one you shipped the connector to). The change to using the GPL should mean that your application should be bound by the GPL'ed. While your interpretation might be correct, the copyright holder disagrees with you. Even if your right, that could be very expensive.
As for 1, I'm not sure you can relicense it, I'd have to read the licenses closely, but you could release your stuff under GPLv3 and include any requirements the Apache license has (including the notice files, not removing the copyright notice, or the waivers of any warrantee). Effectively you can do what you want.
Kirby
Kirby
I'd suggest two things, first: it's not well defined what is and isn't copyright infringement with respect to source code. I suspect that's double true when the GPL was originally written (circa 1984 for the precursor licenses of the GPL, the Emacs and GNU C Compiler license, I think the GPL v1 was in 1987 or 89, while GPL v2 was 1991). Copyright infringement for books, pictures and music aren't completely well defined. If I took 5 lines from a book, is that infringement? If I took the basic plot, but re-wrote everything. How many seconds of sampled music is a copyright violation? How many long a sequence of notes can be re-used before it's copyright infringement. I'm reasonable sure each of those has to be decided on a case by case basis, and that the court is very inconsistent.
Second, it's very clear that they'll take you to court if you violate what they deem a derived work. My hunch is that their definition of infringement is vastly looser then any court in the U.S. would ever interpret it to be. If you even came close to any usage allowed by the GPL on code while using a "All Rights Reserved" license source, you'd be infringing on the copyright. I guarantee every interpretation of a derived work by a U.S. Court that is disallowed for GPL'ed code would also be disallowed for "All Rights Reserved" (which are essentially the only rights you have if you don't comply with the GPL). They spell out for the court the terms, so it's clear, if they don't follow the following terms, they have Fair Use rights. Just like I could license my code to say only guys born on days of the months that are prime can use it, they can dictate the terms of their license, and what they consider a derived work. As their definition is vastly looser then any standard definition used by a court, it's not like the court will use a stronger interpretation. It is also very difficult for someone to argue that they felt in compliance of the license, while using a different definition of "derived work".
If you feel that analysis is wrong, please point out something that would be allowed by a courts interpretation of "derived work" on code which was licensed as "All Rights Reserved", while being disallowed by the FSF definition of "derived work" in the GPL. Anything under fair use doesn't count, as that is allowed on any piece of source regardless of the license.
Kirby
It's a fairly reasonable assertion, they define the terms under which you are allowed to use their programs. It's not meant to line up with the legal term "derived work" that is fairly well established in most contexts (Software is one context where I'm not sure there is a well defined precedent for what is and isn't a derived work, see the bullshit look and feel lawsuits), but it's their working definition of "derived work". If you do what they consider a "derived work", you have none of the rights granted to you via the GPL (specifically the rights to modify or distribute the software, I believe technically you can still run it). It's not like you could ship their software or modify their software and distribute it without the consent of the Copyright Holder. The GPL merely explains explicitly what the copyright holder thinks is acceptable terms.
If they had spelled out that "here are the things we conditions that we consider not to be copyright violations", and spelled them out, and dropped the words "derived work", it'd all have the same legal weight. It's flat out copyright infringement to ship the Linux Kernel in binary or source form if you construct anything the GPL defines to be a "derived work" and distribute it. It's not like a court is going to find in favor of the VMware because of fair use rights. The only way VMware will win this is if they prove that it's a well separated piece of functionality that works outside of the standard Linux kernel, or only uses a well defined subset of the kernel API's that are common to numerous UNIX operating systems (what Linus would approve of essentially, according to several widely quoted public statements).
Kirby
First, RHEL did cause a fairly big splash. A number of people were fairly vocal, and called for the downfall of RedHat. However, there is one critical difference: RedHat freely gave away everything needed to build RHEL to anybody who wanted to show up and use it. It sure sounds like MySQL is not giving away the source or binaries to anyone. They are following the letter of the GPL (which is very good!). But they are not playing in an open space. To the best of my knowledge, RHEL didn't get particularly upset if you gave away the binaries (as long as you took the time to strip off their trademarked materials), and were clear that it wasn't RHEL. I don't really understand why MySQL doesn't just put the source out and let folks build their own binaries (which I think would cause most of this to go away, as there'd be a sourceforge project that distributed the binaries). In the end, their client license will ensure they get paid by folks who use their software in any non-Opensource project (They GPL their client, which is a problem for anybody who uses it in a proprietary application). That is however, their right, and I support them to do it. I just don't use their database at work for anything because of it. (I have a historical preference for PostGres over MySQL, and the change from LGPL to GPL on the client has made it a moot point to re-evaluate the technical merits).
Kirby
I'd suggest that the article didn't say this is "the only way to go past the industrial revolution", here merely pointed out that this might have been a driving factor behind the one in England. I'd say that Africa's situation is strong evidence that they haven't yet had to get over the hump of industrialization. Clearly there's some reason why England has, and a Third World nation hasn't. This line of reasoning seems at least plausible.
Just like historians who say that the Nazi's committed genocide against the Jews, aren't saying that every political party world wide has an interest in genocide. Which is the logical leap you seem to want to make.
You can read whatever you want into it, but it sure looks like your generalizing and projecting something that you wished it said. I see no implication that starts with what the author has to say, and ends where you suggest (everything after the word therefore in your post is you looking to pick a fight, not you being rational or logical).
Kirby
Kirby
Having life insurance or not in the general case, won't make a 5-10 year difference in your life span (over a large population I'd guess that is a reasonable estimate, I wouldn't be shocked to see that number be as high as 20-30 for specific at risk groups). Besides, health insurance I see as good thing for society and the quality of life. Life insurance is a good thing for your family in case of a serious accident. My life insurance pays off when I'm dead it doesn't really affect my quality of life in a measurable way. I've got lots of control over my choice of job, and quite a bit of my lifestyle. However, there are many things which I have no control over that are easy to measure, and statistically important. I'm fairly sure if you examine my DNA, you'll find I have several predispositions (no idea what they are, but I'm sure I have them). Smoking might be something you can say is a legitimate choice. Weight, I'm not so sure there are a number of legitimate medical conditions that make maintaining a healthy weight very difficult. I'm also really sure, it's none of my employers business how much I weight, or what various other specifics of chemical balance is. Shit it'd be a HIPPA violation if a nurse gave that information to anyone outside my immediate family. There are a lot of reasons why those values can be out of whack. Some of which are legitimate choices, some of which are the entire reason we have insurance. Until they are capable of differentiating between those fairly it sure seems like hurting the precise people that insurance is meant to help. I pay a disproportionate amount of a lot of things in taxes and other fees that I tolerate because they are good for society as a whole, and thus indirectly good for me. Life Insurance I don't feel that way about. It's directly good for a specific small group of people. If folks fall on hard times because of a death without insurance, society has other more general ways of helping such folks out.
Besides that, my other gripe about health insurance, is that insurance and litigation are the two things that drive the cost of health care up. The other is that medical providers charge crazy amounts for common things to cover up the costs of really expensive things insurance won't cover. I'm positive that there's no good reason for Aspirin to cost $50, but it does if you read a hospital bill closely. I'm highly unlikely to abuse my insurance or to sue anybody, yet I get to foot the bill for everybody who does. The pure amount of paperwork and headache insurance causes for my General Practitioner is why it costs ~$150-200 a visit. Where it's not uncommon to see a doctor for $30-40 if I have no insurance. I enjoyed an article about a move by a number of doctors to never accept insurance. It's a flat fee of $40-50 dollars, no insurance, they'll treat you just like they treat everybody else. Cash up front, if it takes 60 minutes and 2 tests great. If it takes 2 minutes and they send you out with a placebo, fine. But the fact that I'm billed $100 for the 5 minutes it took a nurse to flush my ears is assine. It's not a complex procedure, I'm fairly sure I could buy or fabricate the equipment used to do it for less then that. I'm fairly sure just about anybody could do the procedure. The only reason to charge so much is to cover the costs of other more expensive procedures, or the risk of a multi-million dollar lawsuit if something happens to my eardrum. Because it costs so much, nobody pays for anything out of their pocket to keep expenses insurance costs down. I like Health Savings Accounts, where you can put Tax Deferred money into an account up to the deductible amount, and have a high initial deductible of ~$2K-4K per year, folks who rarely need health care can carry that money over year to year, and saved a substantial amount on my premiums. It's essentially a relatively easy way to motivate consumers to choose efficient ways to deal with their medical costs, while still covering major costs in cases of disaster.
Health in
Have you considered a future in marketing or politics? Seriously, there's absolutely no difference between charging people extra who have more risk. If you frame it as a discount for good behavior or as a fee for bad behavior, it's doing absolutely the same thing. The only difference is how you present it. That is marketing, and that sort of thing is good for convincing people of an idea.
There's absolutely no way to differentiate between those two cases. I've got no problem with them basing things like this on a statistical basis, but at some point as a generally healthy person I'd just opt out. BMI is a bullshit measure, when I'm at the top weight that is considered "healthy" by BMI and you'd think I was a recovering from an eating disorder. I can't imagine what it'd take to get to the middle of that range (6'6" and 185 is pretty darn thin, I'd nearly have to give up eating to maintain that weight even while working out regularly). I weight about 60lbs more then anybody every guess I weight. Most of the folks in my family do the same thing. Nobody ever guesses close to what any of the men in my family weight.
Insurance is a system where we all pay in to help spread risk across everyone. This is the first step towards just having you pay in what you cost after they skim off their profits. If that's what they want to do, that's fine. Give me some catastrophic coverage for a major emergency, and let me pocket the ~$150/month difference in cost (and more like ~$600 once I negotiate with my employer to get the 75% of my insurance they pay for me). I'll happily invest that money, and pay for my own $4 antibiotic prescription I need once a year, along with the $40 for the doc visit. I'm generally very healthy. I don't have a wife or kids. I lose a tremendous amount of money to the insurance industry every year. I'm very low risk. Anything under $10K would be easy for me to pay cash, anything under about $50K that allowed me to continue working, I could easily finance. Out to about $100K I wouldn't be particularly worried about my ability to pay for it assuming I could continue to work or some form of short/long term disability covered a reasonable percentage my salary. Especially if I got another $600 a month to save for just such a situation. But then I'm freakishly good about saving for a rainy day.
Kirby
Hmmm, I live in Nebraska and I'm fairly sure that with the current tax and tuition situation would have left them with a number of other less then desirable options. I graduated from the University of Nebraska at Omaha. I believe both institutions have similar tuition rates and increases.
Currently tuition, is about $120/hr. When I started there in 1995 it was around $60/hr. All of those are the in-state prices. In all probability, this was seen as a way to fairly have the folks who are costing the most to absorb the costs more efficiently (Maybe not the business dept).
Currently the University of Nebraska has rules about who has to be admitted if you graduated from a state High School. Essentially, if you score a 20 on the ACT and graduated from an in-state HS, they have to admit you.
The costs associated with the University are going way up. I wouldn't be shocked if some of this has to do with the University's problem of hiring and retaining top Engineering, CS and MBA PhD's. They can easily command a great deal more money in the regular job market then say an English Professors salaries.
The alternative is to make it so that nobody but the rich can afford to send their kids to the state school they help subsidize.
Kirby
Next, what I do find very irritating about the phone companies lately is that they way overprice the phones. Then if you don't sign the contract, you get to pay them the profit up front on the phone. Then you get to pay extra every month, because they won't lock you in with a good rate. In my experience, they also build the phones to last about the length of a 2 year contract. I still have my circa 1997 cell phone. Damn thing still works, and runs like a champ. It's just the battery is long dead, and I can't find a replacement. It had a few other quirks (like you couldn't stop it from ringing without answering it). But the phone took an absolute beating for 5 years. I've had 5 phones since then, and I'm generally much easier on them. They die within a year or two.
Finally, I'd really like the phone makers to get together on a system of making phones fairly uniform in UI. I've had 6 phones. Each of them had 2-3 really nice features... But no phone has had the list of all of the features I wish it had. My current list:
If you read the actual mdadm code (I used v2.6.2 from http://www.cse.unsw.edu.au/~neilb/source/mdadm/). Looking at Monitor.c 487 of the Monitor function, it "polls" the file. Waiting 60 seconds (by default, you can set the delay lower). However, if you look at the mdstat_wait(int) call, you'll notice that it's doing a select on the "/proc/mdstat" file. It'll break out of the select if fairly quickly. Force a mirror break, and if you have it setup to get an e-mail, you'll have the e-mail with in your in box in seconds. There's no 1 minute or 15 minute lag.
I'd swear that at one point it used an ioctl to extract all of this information, but it's been ages since I actually read the mdadm tools source... Heck it might have been the precursor tools... mdtools or mdctl or whatever it was called before.
Kirby
I've always said that FOSS is great for commodity areas. Thinks like standard infrastructure (the OS, simple routing, standard Office applications, web browsers, IM, e-mail, to a lesser extent database, HTTP, FTP, SSH, e-mail, DNS, etc and other servers). Just about anything that is standard out of the box functionality people expect from a computer is great to have as an Open Source software. In a lot of ways, even Microsoft software is free as in beer for most folks. Other then Office Software, most of above is "free" if you don't count what the OEM paid MS. The problem with the OEM fee is that it's like sales tax at the grocery store. It's hard to avoid, and few people consider it.
Other software can be FOSS, but in the general case unless your Sun, Microsoft, or IBM it doesn't make a lot of sense to compete over making infrastructure software. Oddly enough, even IBM and Sun are moving their major infrastructure pieces to Open Source. Where I work, I've been pushing us to start using Open technologies for all infrastructure in our software, if not using Open technologies for all our applications.
Kirby
You seem to be describing a "calling convention", not an "ABI". Normally, specific hardware and languages had specific ways it expects to pass parameters. Most of the really odd looking things near the return types of a Win32 method are specifying the calling convention for passing arguments and return values to and from the caller. An ABI is something that a library, or specific piece of software has. Subversion has a binary compatible ABI and API guarantees that are tied to the version numbers. I believe that 1.2.x is forward and backward API compatible with 1.2.y. It is a forward API and ABI compatible from 1.X to 1.(X+n). When you change major version numbers, all bets are off.
A compatible API, says: "If this code compiles from source with this API, it will compile from source and behave the same will any compatible API". This is not discussing forward vs. backward compatible API's. A forward compatible API says that any newer version, a backwards compatible API, says you can re-compile the same source against an older version of the API. Generally most folks only worry about forward compatible API's, not backward compatible. Generally a backward compatible means you have a compatible ABI, or that something that was previously a macro became a function or vice versa.
A compatible ABI, says: "If you have a compiled binary, you can re-link against any compatible ABI and it will behave the same".
To illustrate the difference, you can change a function to a macro, and still be API compatible. However, unless you leave the function implementation in, it is not a compatible API. Or if you change ENOSUCHFILE from -12 to -10 (Linus has stated that he'd have changed specific error numbers to match SCO UNIX so that some of the emulation modes would have been easier, but he didn't realize it until it meant an ABI incompatiblity so he never did it). It's API compatible, but not ABI compatible. ABI compatibility is extremely hard to maintain, and keep the code maintainable (glibc and other libraries do some nasty tricks to ensure ABI compatiblity, and it makes things much harder to read). API compatibility isn't that hard if you try. In the general case, always add interfaces, never remove them. Generally speaking that is a fairly safe way to be forward compatible API's, and reasonable stable ABI's.
In general stable API's are good enough as long as you have source for everything. They are a good middle ground. Compatible ABI's have lots of value to users, but lots of costs to the maintainers. Linux has lots of userland ABI stability, but no internal ABI or API stability.
Kirby
I believe the mdadm tools on Linux actually block the system call until on a change in the status. I didn't state that clearly. So you get immediate notification, RAID re-builds. I wrote something that checked every 2 minutes on all of our RAID devices (Nagios and scripting are great). Generally mdadm and it's alerts informed me once I had those setup, soon enough that could go in and disable the Nagios alerts prior to it noticing.
Kirby
I'm shocked the raid tools for OpenBSD aren't better then that. Not a dig at it, OpenBSD generally prides itself on exceptional tools. OpenSSH, CARP (their replacement for VRRPD), their firewall tools and everything else. Linux has a system call that can be used to monitor the status of a RAID array. It can kick off an arbitrary command, including starting up recovery and/or e-mail alerts. Technically the system call doesn't, but the mdadm tools that use the system call can.
I really hope somebody replies telling me, I'm an idiot and that OpenBSD has exactly such tools. Well and they really exist, as opposed to the clever slashdot behavior of telling me I'm an idiot and be completely wrong.
Kirby
Hmm... your history doesn't match up with Wikipedia's, read the last paragraph or two of the section of Depth fail. If you can cite sources for the version you present, that might be of interest in updating. I don't have an interest either way, but I dislike patents as currently implemented. Thus any abuses of them are of interest to me.
The Wikipedia article leads me to believe that Creative didn't do anything wrong in terms of current legal framework. It sounds like Creative presented it as a concept prior to Carmack inventing it.
Kirby
As a Linux user since the summer of 1996, I can assure that one cannot learn "Ubuntu" in a day. I'm a Linux power user. I use RedHat to the hilt. I can use RPM and YUM, and I find apt cumbersome and a pain. I haven't spent hours using it, but I probably spent 30-60 minutes playing with it. The differences to a professional Linux admin are significant. Yes, I can learn all the differences, but it'd take a lot longer then "a day" to get as good and competant on any old Linux distro as I am on a RedHat machine. I've used Slackware, RedHat, Caldera, Debian, Novell, Turbo Linux, along with LFS distributions. I've probably used others for brief periods. At some point around RedHat 7.2 or so, I decided that keeping them all straight was a total waste of time. It was a hassle to do security patches, it was a hassle to make configuration changes (I believe Slackware still doesn't use PAM, so all of the authentication had to be handled special). It was a huge pain to deal with various other configuration issues (Slackware used a totally different set of startup scripts then RedHat did). We standardized on exactly one Linux, RedHat Linux.
I used to regularly use advanced features like Kickstart to do unattended installs. I administered and ran about 30 machines at my last job, installing and running pretty much every service that comes stock on a RedHat Linux machine. I've reported problems and recommended solutions to the folks who run Whitebox Linux and CentOS. While I can actively use any old Linux box, there are numerous painful problems with switching distributions. Going from RedHat to Novell I find a pain. They are fairly similar, they use RPM's. The two use roughly the same kernel (on the versions I was using). However, all of the advanced management tools are different. It took me an hour or two to figure out how to have Novell's X utilities to auto-detect the video settings after swapping out monitors, I can do that on a RedHat machine without thinking. It took a while to figure out that YaST is how you do everything on Novell, so looking for "system-*" commands from RedHat was a waste of time. It's annoying that "chkconfig", and "service" aren't installed on all Linux machines. It's a pain that folks shuffle which files end up in /etc/, that /etc/profile.d isn't always in the same place. That not all of the network config is handled the system. That not all distributions use /etc/sysconfig the same way. In a serious emergency where time was money, I'd be in a serious trouble with Novell or Debian, where as with RedHat, I'd have no problems at all.
In the sense that you are using "can pick up in a day", I can go from a RedHat Linux distribution, to FreeBSD, OpenBSD, Solaris, or AIX and nearly as comfortable as I am when I work in a Debian or Ubuntu distribution. Also, it's nearly comical to see in this thread the assertion that any code that will run on one Linux distribution will run on another. I've had problems building the same source on two different versions of the same Linux distribution, yes compatibility libraries are very nice and they eventually solved my problem. But it's not like installing source and building it from scratch is a panacea, a lot of code doesn't like to run, or run well out of the box on different versions of RedHat due to different locations of various libraries and other fun stuff. There's a reason that LSB exists, and it ain't because all Linux's are just alike and moving from one to another is a snap. Just go look at all of the patches that RedHat applies to vanilla source in source RPM's, go look at how different they are then the ones that other distro's apply.
Sure, as long as they all run some form of X, and have a reasonable Window Manager or Desk
I believe your understanding to be very wrong. The GPL is always intended to play in the space of "I'm giving you my source code, if you use it, you have to give up yours if you link to mine". If you are correct the GPLv3 is wrong-headed and nobody should use it if that's not the space it's covering. Lots of folks who used the GPLv2 would be up in arms if GPLv3 isn't nearly identical in spirit. As most GPLv2 code will be allowed to update to GPLv3 automatically if they followed the FSF instructions for applying the license. If you believe I'm wrong, if you could cite the paragraph that leads you to believe that it's legal to link to GPLv3 code and not re-distribute the source, I'd be highly interested.
I'm unaware of an LGPLv3 (I believe there is a discussion about constructing one, but the most recent one I know of a public test available for is LGPLv2.1, which is essentially changing "Library" to "Lesser" from the LGPLv2).
Apache v1.1 or v2.0 are *NOT* in the third category as you imply. They are in the first category. You can take Apache 2.0 code modify it and redistribute it. No one has legal recourse to get you to stop distribution, or produce the source (at least according to my reading of the APLv2.0 Section 4). The only requirement Apache 2.0 has is some specific patent things, and that you distribute a copy of the license, that you leave the copyrights in tact, and that you acknowledge that you used that source. There is no provision for producing the pieces you modified. There is a provision for explicitly documenting that you did modify them if they are being distributed. With the LGPL, the copyright holder could sue you for copyright violation if you fail to produce the modified source to the modified binaries. The EPL and CPL have somewhat similar provisions.
I believe the GPLv3 has various options and provisions to avoid people writing the GPL + patent provisions, or GPL + attribution provisions. Thus allowing folks to make the GPLv3 compatible with other far more permissive licenses like the Apache 2.0 that have one small condition that the GPL didn't (which is an extra condition, and thus GPLv2 incompatible). The FSF has held that Apache 1.1 and Apache 2.0 are incompatible with the GPLv2, I believe the GPLv3 will make them compatible + have the GPL v2 provisions + have more patent and IP provisions). The Apache folks have said that GPL and Apache 1.1 and Apache 2.0 are compatible, that the additional provisions are insignificant.
The GPLv3 is about covering the second type of license.
Kirby
Right now, there are roughly 3 types of OSS licenses.
The article states "look at the proliferation of licenses", as a sign that the GPL isn't filling a need. The simple facts are that the first two licenses are pretty much in the bag. Nobody writes new licenses that attempt to the accomplish the first two. Pretty much to a person, everyone uses Apache 2, BSD or the GPL to accomplish those goals. If you start looking down the list of other one-off licenses that are for OSS. Those are all about filling the need in the third item. If anything, it could be said that the LGPL is "failing". It isn't the "one true license" to accomplish the task. Essentially the proliferation of license's is about finding a "share and share alike" that can exist in a corporate environment. Where the core technology can be shared and developed by many folks, while the extensions and non-core pieces can be value-adds that are solve for money.
Greg Stein's a brilliant guy, and one hell of an engineer. But I think he's living in his own little world here. Lot's of folks like and enjoy writing software under the premise of the second type of license. Some folks do it under the first. In the end, the collaborative effort will virtually always win out. So in a lot of ways it doesn't matter if you use a license from the first or second group. That's why Apache has never been taken and had a closed competitor that is more used. Sure some commercial products are based on it, but none of them will ever quash Apache out of existence because they are so popular.
All the action is how to have an open source commercial license. The LGPL has a few terms that are a bit harsh on business, and have little to say with respect to patents or trademarks. In this day and age a license must address those.
Kirby
Upon virtually every microsoft patch Firefox and Thunderbird require me to reset my default browser and e-mail client. My guess is that this is a simple security mechanism. Hooking into the startup of e-mail and web software would be a good place for a virus to hide... Then just startup the older software and life is good. It'd be a good place to hide malware startup that is outisde of the normal places that are checked for issues.
Kirby
Kirby
Oddly enough, that's not as true as you'd think. I know that HP and IBM both had projects to virtual hardware (not in exactly VMWare style virtualization, more like a JIT optimization) where software ran faster after being virtualized. By about 20% if I remember correctly. HP's project was Dynamo and IBM's was DAISY.
Virtualizing a CPU at runtime in a JIT-optimization fashion can actually lead to a significant speed up. Some of this might have been pulled into CPU's by now, but I'm always surprised that more technology isn't moving this direction. Ultimately optimization at runtime has more information then at compile-time, and thus can apply optimizations not otherwise available.
Kirby