Re:Does certification actually matter?
on
JCert Is Dead
·
· Score: 4, Interesting
And with todays surplus of people wanting tech job every little bit that gives you a chance to get an interview helps
Exactly right. I'm a software developer, but in the course of all my Unix development experience I've also had to administer a lot of Unix boxes (just about every flavor). I've also maintained a couple of small networks. So after being unemployed as a developer for over a year, I brushed up on Solaris specifics (I have an ancient Sparc box doing server duty in the basement, and a Solaris X86 box beside my main Linux SMP box under my desk), got my Sun Cert for Solaris System Administrator, and am currently employed on contract to IBM as a Solaris sys admin (yes, IBM and Solaris -- managed services at a customer site, over 300 Sun boxes from Netras to SunFires plus other assorted Unix and Linux flavors).
(The pay isn't quite up to software engineer pay, but it beats working at Best Buy.;-)
Despite rewriting my resume with a heavy slant to my actual admin experience, I don't know that it would have been looked at without the cert. Not that Sparcstation and X86 experience helps much when you open up the cabinet doors on an SF 15000...
Is there really any kind of market for hardware that is programmed at the Assembly Language level anymore?
That's not really the point. The point is that assembly language programming will intimately acquaint you with the design flaws (and wins) of the processor. At the programming level, this is something that compiler writers worry most about. But it has performance (and price/performance) ramifications to everything based on that CPU.
Eg if a different architecture lets me add a few general purpose registers but still save on overall transistor count, I can have a lower power, lower cost chip that may well be faster (for a given clock freq) because the required code is simpler, hence smaller, hence less likely to blow the cache.
Lots of interacting variables there, of course, but the original point was that the x86 architecture blows chunks, and being able to run it at several gigaherz (at some ridiculously high power density) doesn't necessarily make up for that.
You only need to prove damages if you want to collect money. Even with no monetary damages you can force the copyright violator to cease and desist distribution of your software (or face contempt of court and criminal charges).
The last thing anyone shipping a money-making product wants is to have an injunction against shipping that product issued. Losing redistrubution rights because of an invalid license threatens exactly that.
(And that is why the GPL has never been seriously challenged in court. It's a lose-lose situation for the violator, since the odds of a judge declaring the GPL as equivalent to the public domain are essentially nil.)
It's all in the selection process. For humans (and other life forms), only the DNA of the ones that survive long enough to reproduce gets selected. You end up with mostly good DNA.
If you filtered/dev/random addresses through a selection process (run an SMTP check) you'd end up with good addresses. To speed up the process, generation random mutations in known good addresses and test those. (Which is indeed what some spammers do, they just skip the test phase.)
Hell no it doesn't. A little understanding may equal trust -- which leads to all the sorts of horrors that are routinely chronicled in the RISKS digest, and the kind of crap that Microsoft puts out.
The more experienced of us, with more understanding, know the many ways things can be screwed up (accidentally or deliberately) with a computer assist. As the saying goes, "to err is human; to really foul things up requires a computer".
I've been programming for 30 years, and I've worked on (among numerous other things) banking systems. That's why I still pay everything with paper cheques that get returned to me along with the statement.
For what (little) it's worth. I knew I shared Linus' birthday (or rather he mine, I've got a few years on him), but I didn't realize it was also Von Neumann's.
Immediately drop the electoral college in favor of total popular vote percentage across the country, as it's an implementation artifact from centuries ago,
Hell no it isn't, it's a design decision.
Think "republic", not "democracy".
Sure, the system could use some tweaking (or more likely, some undoing of previous tweaks), but don't even start until you grasp that fundamental.
There are cons that exploit people's generosity as well -- pretending to have car trouble and robbing someone who pulls over to help being a classic example
I don't think that I'd label that as a con -- confidence game -- so much as just simple thuggery. But point taken.
The actual premise of the transaction doesn't even sound legal. A banker needs to move money that isn't his by using an offshore account?
Yep. There's an old saying, "you can't con an honest man". Most big cons have some element of dishonesty (beyond the "getting something for nothing") because it helps to discourage the mark from checking on the legitimacy of the scenario in the first place.
Well, I was running 2.4.10 on a 2-way (P-III) SMP machine (my main home machine) with no problems that I recall, otoh SuSE may have backported some patches, it wasn't a stock kernel.
But hey, if -test10 is working for you, no need to upgrade yet.
My guess -- and it's only that -- is 'yes'. Early 2.4 had a lot of teething problems (what with VM and all) and didn't really settle down until about what, 2.4.10 or so? Those problems delayed the start of the 2.5 branch.
Since 2.6.0 already seems pretty stable (and went through 11 -test releases) we may well see 2.7 sometime early next year (January? Get Christmas and Linus's birthday (same day as mine) out of the way first).
No, no, you don't need to make a decision on that right now. Just try this one for a while and later when you've decided you like it we'll finish setting up the details.... </salespeak>
The article mentions one application that (currently) requires MS Office, there may be others.
The thing is, migrating the easy 80% gives OOo the dominant desktop position. Certainly any new apps will be written to be OOo-friendly, and there will be pressure to port the old ones. I'm sure they have their IT people looking at what it'd take to do that. Might be an opportunity there for developers or companies in the Austin area with migration/porting experience.
Unix's core value of writing programs for other programmers is precisely because, when the software is intended for the end user, the end user should never realize he's running Unix. The user shouldn't have to care what OS is running, it should just work, transparently. TiVo is an excellent example of this. Numerous point-of-sale systems are other examples. MacOS X for another -- although it makes its underlying Unix available to those that want it.
Way back in the early days of the BIX (BYTE Information Exchange) conferencing system, when Jerry Pournelle would sometimes go off on a rant about how hard Unix was to use, I would gently remind him that he was actually using a Unix system to post that rant and read my response. (That he was later able to use the Windows-based "Galahad" front-end (one of several) to BIX is a testament to the worth of the Unix value of text interfaces. Try slapping a different GUI over top of one you don't like!)
The original code may always be released under the BSD (or whatever original license), but that doesn't necessarily mean that it is available.
I get a copy of program 'bar', a proprietary extension of BSD-licensed program 'foo'. It does interesting things and I want to look at the original source to see how much of that is in there. How do I find the 'foo' source? The 'bar' author has no requirement or particular incentive to tell me. An internet search may turn up 'foo', or may turn up 18,467 programs also called 'foo' (and I have no idea which is the right one), or it may never have been posted anywhere -- the orginal version was distributed on a 2400' reel of 9-track magtape.
For that matter, I have no way of knowing that 'bar' is even based on 'foo' unless the distributer tells me. I might not know to even go looking for the source.
This argument about "oh, but the *original* code is still available" is common with the BSD defenders, and rather specious. Nothing in the BSD license guarantees that it is still available. The GPL at least guarantees that the source is available for as long as the binary is.
(That said, I agree that whoever writes the (original) code gets to pick the license. Just don't kid yourself about the real long-term effects of that choice.)
You are required to keep the copyright notice on subsequent copies of BSD-licensed works, not the original license. Even if that copyright notice includes the license terms, it only applies to the original content, not to your own additions nor to the combined work as a whole. (In fact, it doesn't. The license paragraph just requires you to retain "the above copyright notice" and "the following two paragraphs" (which are disclaimers of liability or warranty) -- not the license paragraph itself. That'd be viral;-).
Sure, if some subsequent recipient of the combined code can carefully separate out the original BSD'd code from the GPL'd combination and GPL'd additions, he may use that under the original BSD license. Good luck.
(For an expanded explanation, read the copyright page on any of the O'Reilly X Windows books, which contain text released under the X license, but also text and revisions copyright by O'Reilly, the net result of which "is that copying of this document is not allowed".)
When you use BSD code you can choose to licence the derivitive work under the GPL, which is to choose to continue the chain of life with fewer licensing rights than were previously had from where you got the software. It is a hastey generalization to say that the only thing you can do with BSD software is to make it proprietary. A presumption that is entirely wrong.
Funny, I don't recall making, much less expressing, such a presumption. However, the original author of the code (the one who chooses the license) has no guarantee one way or the other if he chooses the BSD license, and thus cannot guarantee those liberties to anyone except those to whom he personally hands the code. By choosing the GPL, he guarantees those rights to everyone who ever receives a copy -- in perpetuity, if that's the way copyright duration works out.
The GPL isn't about what the author wants only those he directly "sponsors" to be allowed, but also what he wants all those in depth (to borrow a multi-level marketing term) from those direct sponsors to be able to do. A slight restriction to guarantee wider reach of the original liberties. The BSD may permit such, but it cannot guarantee it.
I never said that legislation or case law agrees with that interpretation.
It would depend on how much (if any) of the original work (in terms of header files, etc) had to be incorporated into the "derived work" for the dynamic linkage to function, and even then it's not necessarily a clear case of copyright infringement.
Of course, if the violator has to ship a copy of the dynamic libraries along with his binary (vs having them preinstalled as part of a normal distro), then he's clearly redistributing them, and better follow the license. (Oh, he might win a lawsuit (as long as he's including source for the GPL'd libraries), but is it worth the risk?)
Yes, it is possible to accidentally use GPL'd code. Yes, you're screwwed if you do. No, screwwed does not mean you'll be executed. Yes, screwwed probably means there will be a cost to you: time, money, or both.
And this differs from using any other non-public domain software how? (Granted, if that non-PD software is, say, BSD licensed you're not so badly screwed -- unless you removed the BSD copyright notices -- OTOH if it was some proprietary vendor's code you're screwed even worse.) TANSTAAFL -- if you're casually incorporating pre-written code into a project you'd better darn well know where it came from and what the conditions are for its use.
I read his statement. Nowhere in it did he say for him to legally sell his work.
People sell stuff all the time without it being legal for them to do so. Unlike laws of nature, the laws of man are not self-enforcing.
But we agree on the main point -- for him to legally sell his work, he'd have to release source -- or take out the copyright bits that he doesn't hold copyright to.
The BSD license does grant the developer more liberty to use the code in any fashion they choose, including later restricting rights if they so choose. (Emphasis added.)
Exactly. The BSD may grant the first generation recipient more rights, at the potential cost of such to all subsequent generations of recipient. The point of the GPL is to guarantee liberties to all subsequent generations of recipient.
So to short-term thinkers, the BSD appears "freer", but in long-term reality the GPL is.
The concept of a derived work is not explicitly defined
Copyright law appies. The term "derived work" is pretty well defined both by legislation and case law.
shouldn't this carry over to any code which utilizes system calls in Linux?
There's a clear (and written) exception in the Linux license that permits user programs to make system calls without being affected by the GPL. That is, after all, the function of an OS. Binary modules go beyond normal system calls, however. (Furthermore, glibc exists for plenty of other kernels besides Linux, including BSD and proprietary.)
3. The GPL has many bizarre concessions and terms, such as requiring those who distribute GPL software to distribute it by mail at anyone's request, charging only the cost of media.
Lots of licenses have what seem to be bizarre terms. In this example, though, that offer need only be made if you don't distribute the source along with the binary -- and the term is not specifically the cost of the media, but "for a charge no more than your cost of physically performing source distribution,". I.e. it's okay to charge for shipping and handling too, just don't make it a profit center.
The beauty of the GPL is that there's nothing to "prove in court". Copyright law trumps, and if you distribute binaries of GPL'd works (or derivatives thereof) you are in violation of copyright law, unless you have the permission of the copyright holder (which you can obtain by complying with the License). (That the GPL has never been tried in court is an indication of its strength -- any sane lawyer up against it has advised his client to settle rather than fight.)
As to being forced to release source, clearly you cannot be, you can only be compelled to cease violating copyright. In the case of devices that embed GPLd code (like Linksys), that means to stop distributing the device (until the code is rewritten). Most manufacturers would rather not do that, being in the hardware rather than software business, so they are likely to agree to release source if they are able to do so, but that's their choice.
(Unless, of course, they got stuck with code they didn't realise was under the GPL from a 3rd party vendor -- in which case the hardware mfg goes after the vendor for violation of contract (at least) or possibly fraud, and also refers the copyright holder to that 3rd party for copyright infringement lawsuit. And stops distributing the device unless they can get dispensation from the copyright holder(s).)
Judges are going to have a hard time putting a dollar value on "failing to contribute back to the over all OSS experience."
Presumably there's new code involved -- kind of pointless to distribute an exact copy of something that's GPL'd under a proprietary license. Pretty easy to put a dollar value on that code, using industry accepted cost estimating models.
A judge might not actually set the valuation that high, but it leaves a lot of room for negotiation between that and zero.
Slashdot Mirror
And with todays surplus of people wanting tech job every little bit that gives you a chance to get an interview helps
;-)
Exactly right. I'm a software developer, but in the course of all my Unix development experience I've also had to administer a lot of Unix boxes (just about every flavor). I've also maintained a couple of small networks. So after being unemployed as a developer for over a year, I brushed up on Solaris specifics (I have an ancient Sparc box doing server duty in the basement, and a Solaris X86 box beside my main Linux SMP box under my desk), got my Sun Cert for Solaris System Administrator, and am currently employed on contract to IBM as a Solaris sys admin (yes, IBM and Solaris -- managed services at a customer site, over 300 Sun boxes from Netras to SunFires plus other assorted Unix and Linux flavors).
(The pay isn't quite up to software engineer pay, but it beats working at Best Buy.
Despite rewriting my resume with a heavy slant to my actual admin experience, I don't know that it would have been looked at without the cert. Not that Sparcstation and X86 experience helps much when you open up the cabinet doors on an SF 15000...
Is there really any kind of market for hardware that is programmed at the Assembly Language level anymore?
That's not really the point. The point is that assembly language programming will intimately acquaint you with the design flaws (and wins) of the processor. At the programming level, this is something that compiler writers worry most about. But it has performance (and price/performance) ramifications to everything based on that CPU.
Eg if a different architecture lets me add a few general purpose registers but still save on overall transistor count, I can have a lower power, lower cost chip that may well be faster (for a given clock freq) because the required code is simpler, hence smaller, hence less likely to blow the cache.
Lots of interacting variables there, of course, but the original point was that the x86 architecture blows chunks, and being able to run it at several gigaherz (at some ridiculously high power density) doesn't necessarily make up for that.
You only need to prove damages if you want to collect money. Even with no monetary damages you can force the copyright violator to cease and desist distribution of your software (or face contempt of court and criminal charges).
The last thing anyone shipping a money-making product wants is to have an injunction against shipping that product issued. Losing redistrubution rights because of an invalid license threatens exactly that.
(And that is why the GPL has never been seriously challenged in court. It's a lose-lose situation for the violator, since the odds of a judge declaring the GPL as equivalent to the public domain are essentially nil.)
It's all in the selection process. For humans (and other life forms), only the DNA of the ones that survive long enough to reproduce gets selected. You end up with mostly good DNA.
/dev/random addresses through a selection process (run an SMTP check) you'd end up with good addresses. To speed up the process, generation random mutations in known good addresses and test those. (Which is indeed what some spammers do, they just skip the test phase.)
If you filtered
understanding=trust
Hell no it doesn't. A little understanding may equal trust -- which leads to all the sorts of horrors that are routinely chronicled in the RISKS digest, and the kind of crap that Microsoft puts out.
The more experienced of us, with more understanding, know the many ways things can be screwed up (accidentally or deliberately) with a computer assist. As the saying goes, "to err is human; to really foul things up requires a computer".
I've been programming for 30 years, and I've worked on (among numerous other things) banking systems. That's why I still pay everything with paper cheques that get returned to me along with the statement.
For what (little) it's worth. I knew I shared Linus' birthday (or rather he mine, I've got a few years on him), but I didn't realize it was also Von Neumann's.
Many happy returns all round, then.
Immediately drop the electoral college in favor of total popular vote percentage across the country, as it's an implementation artifact from centuries ago,
Hell no it isn't, it's a design decision.
Think "republic", not "democracy".
Sure, the system could use some tweaking (or more likely, some undoing of previous tweaks), but don't even start until you grasp that fundamental.
There are cons that exploit people's generosity as well -- pretending to have car trouble and robbing someone who pulls over to help being a classic example
I don't think that I'd label that as a con -- confidence game -- so much as just simple thuggery. But point taken.
The actual premise of the transaction doesn't even sound legal. A banker needs to move money that isn't his by using an offshore account?
Yep. There's an old saying, "you can't con an honest man". Most big cons have some element of dishonesty (beyond the "getting something for nothing") because it helps to discourage the mark from checking on the legitimacy of the scenario in the first place.
Well, I was running 2.4.10 on a 2-way (P-III) SMP machine (my main home machine) with no problems that I recall, otoh SuSE may have backported some patches, it wasn't a stock kernel.
But hey, if -test10 is working for you, no need to upgrade yet.
My guess -- and it's only that -- is 'yes'. Early 2.4 had a lot of teething problems (what with VM and all) and didn't really settle down until about what, 2.4.10 or so? Those problems delayed the start of the 2.5 branch.
Since 2.6.0 already seems pretty stable (and went through 11 -test releases) we may well see 2.7 sometime early next year (January? Get Christmas and Linus's birthday (same day as mine) out of the way first).
No, no, you don't need to make a decision on that right now. Just try this one for a while and later when you've decided you like it we'll finish setting up the details....
</salespeak>
Actually with automatic update it might do just that. ;-)
The article mentions one application that (currently) requires MS Office, there may be others.
The thing is, migrating the easy 80% gives OOo the dominant desktop position. Certainly any new apps will be written to be OOo-friendly, and there will be pressure to port the old ones. I'm sure they have their IT people looking at what it'd take to do that. Might be an opportunity there for developers or companies in the Austin area with migration/porting experience.
Precisely. Mod parent up.
Unix's core value of writing programs for other programmers is precisely because, when the software is intended for the end user, the end user should never realize he's running Unix. The user shouldn't have to care what OS is running, it should just work, transparently. TiVo is an excellent example of this. Numerous point-of-sale systems are other examples. MacOS X for another -- although it makes its underlying Unix available to those that want it.
Way back in the early days of the BIX (BYTE Information Exchange) conferencing system, when Jerry Pournelle would sometimes go off on a rant about how hard Unix was to use, I would gently remind him that he was actually using a Unix system to post that rant and read my response. (That he was later able to use the Windows-based "Galahad" front-end (one of several) to BIX is a testament to the worth of the Unix value of text interfaces. Try slapping a different GUI over top of one you don't like!)
The original code may always be released under the BSD (or whatever original license), but that doesn't necessarily mean that it is available.
I get a copy of program 'bar', a proprietary extension of BSD-licensed program 'foo'. It does interesting things and I want to look at the original source to see how much of that is in there. How do I find the 'foo' source? The 'bar' author has no requirement or particular incentive to tell me. An internet search may turn up 'foo', or may turn up 18,467 programs also called 'foo' (and I have no idea which is the right one), or it may never have been posted anywhere -- the orginal version was distributed on a 2400' reel of 9-track magtape.
For that matter, I have no way of knowing that 'bar' is even based on 'foo' unless the distributer tells me. I might not know to even go looking for the source.
This argument about "oh, but the *original* code is still available" is common with the BSD defenders, and rather specious. Nothing in the BSD license guarantees that it is still available. The GPL at least guarantees that the source is available for as long as the binary is.
(That said, I agree that whoever writes the (original) code gets to pick the license. Just don't kid yourself about the real long-term effects of that choice.)
You are required to keep the copyright notice on subsequent copies of BSD-licensed works, not the original license. Even if that copyright notice includes the license terms, it only applies to the original content, not to your own additions nor to the combined work as a whole. (In fact, it doesn't. The license paragraph just requires you to retain "the above copyright notice" and "the following two paragraphs" (which are disclaimers of liability or warranty) -- not the license paragraph itself. That'd be viral ;-).
Sure, if some subsequent recipient of the combined code can carefully separate out the original BSD'd code from the GPL'd combination and GPL'd additions, he may use that under the original BSD license. Good luck.
(For an expanded explanation, read the copyright page on any of the O'Reilly X Windows books, which contain text released under the X license, but also text and revisions copyright by O'Reilly, the net result of which "is that copying of this document is not allowed".)
When you use BSD code you can choose to licence the derivitive work under the GPL, which is to choose to continue the chain of life with fewer licensing rights than were previously had from where you got the software. It is a hastey generalization to say that the only thing you can do with BSD software is to make it proprietary. A presumption that is entirely wrong.
Funny, I don't recall making, much less expressing, such a presumption. However, the original author of the code (the one who chooses the license) has no guarantee one way or the other if he chooses the BSD license, and thus cannot guarantee those liberties to anyone except those to whom he personally hands the code. By choosing the GPL, he guarantees those rights to everyone who ever receives a copy -- in perpetuity, if that's the way copyright duration works out.
The GPL isn't about what the author wants only those he directly "sponsors" to be allowed, but also what he wants all those in depth (to borrow a multi-level marketing term) from those direct sponsors to be able to do. A slight restriction to guarantee wider reach of the original liberties. The BSD may permit such, but it cannot guarantee it.
I never said that legislation or case law agrees with that interpretation.
It would depend on how much (if any) of the original work (in terms of header files, etc) had to be incorporated into the "derived work" for the dynamic linkage to function, and even then it's not necessarily a clear case of copyright infringement.
Of course, if the violator has to ship a copy of the dynamic libraries along with his binary (vs having them preinstalled as part of a normal distro), then he's clearly redistributing them, and better follow the license. (Oh, he might win a lawsuit (as long as he's including source for the GPL'd libraries), but is it worth the risk?)
Yes, it is possible to accidentally use GPL'd code. Yes, you're screwwed if you do. No, screwwed does not mean you'll be executed. Yes, screwwed probably means there will be a cost to you: time, money, or both.
And this differs from using any other non-public domain software how? (Granted, if that non-PD software is, say, BSD licensed you're not so badly screwed -- unless you removed the BSD copyright notices -- OTOH if it was some proprietary vendor's code you're screwed even worse.) TANSTAAFL -- if you're casually incorporating pre-written code into a project you'd better darn well know where it came from and what the conditions are for its use.
I read his statement. Nowhere in it did he say for him to legally sell his work.
People sell stuff all the time without it being legal for them to do so. Unlike laws of nature, the laws of man are not self-enforcing.
But we agree on the main point -- for him to legally sell his work, he'd have to release source -- or take out the copyright bits that he doesn't hold copyright to.
The BSD license does grant the developer more liberty to use the code in any fashion they choose, including later restricting rights if they so choose. (Emphasis added.)
Exactly. The BSD may grant the first generation recipient more rights, at the potential cost of such to all subsequent generations of recipient. The point of the GPL is to guarantee liberties to all subsequent generations of recipient.
So to short-term thinkers, the BSD appears "freer", but in long-term reality the GPL is.
The concept of a derived work is not explicitly defined
Copyright law appies. The term "derived work" is pretty well defined both by legislation and case law.
shouldn't this carry over to any code which utilizes system calls in Linux?
There's a clear (and written) exception in the Linux license that permits user programs to make system calls without being affected by the GPL. That is, after all, the function of an OS. Binary modules go beyond normal system calls, however. (Furthermore, glibc exists for plenty of other kernels besides Linux, including BSD and proprietary.)
3. The GPL has many bizarre concessions and terms, such as requiring those who distribute GPL software to distribute it by mail at anyone's request, charging only the cost of media.
Lots of licenses have what seem to be bizarre terms. In this example, though, that offer need only be made if you don't distribute the source along with the binary -- and the term is not specifically the cost of the media, but "for a charge no more than your cost of physically performing source distribution,". I.e. it's okay to charge for shipping and handling too, just don't make it a profit center.
But until the GPL is proven in court,
The beauty of the GPL is that there's nothing to "prove in court". Copyright law trumps, and if you distribute binaries of GPL'd works (or derivatives thereof) you are in violation of copyright law, unless you have the permission of the copyright holder (which you can obtain by complying with the License). (That the GPL has never been tried in court is an indication of its strength -- any sane lawyer up against it has advised his client to settle rather than fight.)
As to being forced to release source, clearly you cannot be, you can only be compelled to cease violating copyright. In the case of devices that embed GPLd code (like Linksys), that means to stop distributing the device (until the code is rewritten). Most manufacturers would rather not do that, being in the hardware rather than software business, so they are likely to agree to release source if they are able to do so, but that's their choice.
(Unless, of course, they got stuck with code they didn't realise was under the GPL from a 3rd party vendor -- in which case the hardware mfg goes after the vendor for violation of contract (at least) or possibly fraud, and also refers the copyright holder to that 3rd party for copyright infringement lawsuit. And stops distributing the device unless they can get dispensation from the copyright holder(s).)
Judges are going to have a hard time putting a dollar value on "failing to contribute back to the over all OSS experience."
Presumably there's new code involved -- kind of pointless to distribute an exact copy of something that's GPL'd under a proprietary license. Pretty easy to put a dollar value on that code, using industry accepted cost estimating models.
A judge might not actually set the valuation that high, but it leaves a lot of room for negotiation between that and zero.