One big advantage for public companies was the shifting of employee compensation from your income to the shareholders through employee stock options.
That's not really true, the market didn't value diluting the stock in that way (and, in many ways, still doesn't). So the companies went from paying employees to not having to do so (or paying them much less anyway). It's like if you throw away a million dollar painting, and I "pay" for a local housing project with it (wow, I'm a nice guy:)... you didn't "lose" a million dollars, or lose a bunch of houses and the IRS aren't going to allow you to put either down as a TAX break:).
Of course, you'd then be free not to throw anything away... just in case it might be worth something, even though you'd still only have a worthless old painting if you hadn't thrown it away. And some people seem to be applying similar reasoning to Stock Options.
You have a decent SCM application, that does the right thing. Or you re-diff against the previous version. Or you look through by hand and take the obvious bits. Or you have a minor fork which is "what apple has", and then you can compare the results of both builds. Or you do nothing, but in that case you haven't lost anything. You've just not used what they gave you.
No obligation, but it is stupid on Apple's part to not keep code in sync.
Possibly, possibly not... but if it's supposedly so bad for them to do what they are doing, why are you complaining.
They'll learn better from their own mistakes, if they are mistakes. All the complaining does is piss them off, and blind them to any problems they are creating because they won't want to admit the annoying people are right.
And as I said in the previous story, there is a big difference between "what customers want" and "what upstream wants"... sometimes requiring two seperate patches on seperate deadlines. As long as they obey the GPL, I really couldn't care if they never send patches back... they aren't required to do so, and if the KDE people wanted that they should have used some other license. You could argue that it's "nice" to do so, and part of the "community" spirit... but given their entire GUI foundation is proprietary I don't see that they'd benifit much.
In many ways this is like saying that when OpenBSD develop new code X, they should be required to backport it to NetBSD... no, sorry, it doesn't work that way... and sometimes that part can take 10x as long as writting the code, if you do it, so I certainly would never use any source that required it.
Any library which isn't actually used, isn't even loaded.
No, what really isn't loaded are the bits of code that aren't actually loaded. Maybe you meant eventaully paged out, and not paged back in... but then you'd just be mostly wrong.
If you think a win32/linux server with a GUI loaded and a monitor turned off is "just as well off" as a linux server that never loaded the GUI in the first place, you are nieve at best.
Tracking structures have to be allocated, even if everything is paged out (so it can be paged back in)... this is eating kernel memory. Pretty much all code allocates resources (esp. GUI code), at best that's swapped to disk and is just screwing your swap and taking more kernel tracking space... but things like file descriptors etc. are just more unswappable kernel memory. At the best of times that GUI code is actually doing something (thus keeping "a bit" of it's RAM footprint in core), and god help you if someone enabled a screensaver (luckily win32 comes with one turned on by fdefault for the login screen -- yeh). Any activity at all is also affecting the scheduler and VM, as it has to deal with the tasks that wouldn't be there otherwise.
To be fair Red Hat developers are using the same tools and same OS as upstream, Apple developers aren't. This means Red Hat just needs to know "does it work" but apple needs to know "does it work" and does it the ported version work. Also, all the commercial Linux companies are carrying around huge patchsets on upstream... sometimes because upstream is slow/dead... but also because "making it work for my customer tomorrow" is very different than "getting it into upstream", and even with the same tools/OS require two completely seperate patches.
So, while I have no love for Apple and think the 5th Free Unix is going to do as well as the 6th[1] in the long run, I can't fault them for what I see them doing with KHTML.
[1] Yes, sun fan boys, that's a random stab in the dark at you.
I think it will be useful to have a system whereby drivers can be loaded without requiring the entire kernel to be compiled.
You can do this. RHEL, for instance, has had driver update CDs... where you couldn't install normally, so you insert the driver update CD before your install CD and now you can. Just like XP/Solaris/OSX/whatever-comercial-OS.com.
configure his web server to report itself as IIS in the headers it returns. That's the only real way to know what a web server is running
I'm trying to decide if your post was supposed to be a "Troll Tuesday" one, or if indeed you are just clueless. I'm thinking, mostly, it's just cluessness but I'm hoping not.
No they're not. Candy costs about 55 cents a bar now, when it used to be about 33 cents when I was a kid. Does this mean if I were to shoplift a candy bar, it's the store's fault? Or the candy manufacturer?
While I wouldn't advocate that you violate the copyright protection on a DVD, doing so is vastly different than walking into a store and stealing a copy off the shelf. By the same measure copyright violation cannot be compared to stealing a candy bar, and I assume you know this and are intentionally lying. A more accurate comparson would be to not being able to afford the candy and so going home and making your own candy using a recipe for said store candy that you "aquired" from somewhere.
And even that isn't a great comparison, but at least both involve breaking the same law.
Point to an instance of Microsoft discriminating against gays - point to a change in their internal policies that says they're going to. Microsoft is one of the few companies I know that offers domestic partner benefits - are they changing that? NO!
Err, removing their support of this bill? Not speaking is an action. I find it hard to believe that they think the govt. will move to Linux (basically the only stick they have) over this. Apart from that I really don't see what anyone could do to them to stop them doing the right thing.
I guess I can see it if the the govt. hinted at restarting a monopoly charge, but I'd have expected it to happen a while ago.
OSI fully recognizes that Sun is acting quite freely.
Which is why OSI have publicly stated they want to cut down the number of licenses they approve, because saying the CDDL is fine is just making them irrelevant.
Your sentiment is extremely GPL-centered.
When Solaris source code is available, it'll be the 7th major unixlike OS with source code freely available (Linux, {Net, Free, Open, DFly}BSD, Darwin) and Solaris will be incompatible with all the others. Worse than that is that every person even vaguely qualified has advised me to not even look at the Solaris source, as you are likely to become tainted and unable to contribute to GPL source code.
And before you speak out of ignorance, yes, Linux dual licenses bits of code and so shares with the other 5 above.
For Schwartz, or random Sun Fan boys, to say "It's not the CDDL's problem, it's the GPL's" is at best ignorant and more likely just lies.
Has anyone done any benchmarks I've missed that show apache-httpd is actually faster using threads, for real platforms? Yes, I realise win32 requires them... I don't care.
As an example, Redhat announced a while back that they'd support a given version of RHEL for, what, 3 years after its release?
Over double that, it's 7 years from release. Which means that box you don't want to have to care about can tick over serving DNS for about 6 years all the while getting errata, but you can also have the latest postgresql on another box.
My only complaint is that arch doesn't do anything to ensure its own files stay in a permissions state that'll be acceptable to all people committing to the tree. So I occasionally run into "permissions hell" when working on the same branch with another developer.
Intersting... I'd have assumed for multiple core developers you'd just have a release repo. and one developer repo. each pulling changesets into each others repo. and doing a single pull into the core repo. every now and then (probably with a testing repo. in between). Did you try that?
That's kind of how I work, as a single core developer... the release repo. is just a branch with remote tags
A comma delimited file doesn't tell the parser what each of the fields means, neither does a straight binary data file. At best they annouce their filetype in a header.
...making it just as obvious what each column is for. Actually it's often much better, I can write about 3 lines of perl that will always correctly parse the CSV file and put the values into a hash... God knows how much time you'd have to spend on the XML version.
Because, of course, the XML file is really an "XML subset" or the parser needs to deal with the fact that the "record" has 10 extra child nodes (whitespace) and will often have extra nodes or changed nodes like having only a nickname, not having a nickname, having the nickname as a child of firstname or having the firstname, lastname and nickname inside another "person" node -- because, hey, it's XML so it's magic.
And if you went the XML subset route, you are screwed because people will assume otherwise.
Sendmail -- most BSDs use it... craps itself under load, and is a major PITA to cleanup the queue.
Attach the disks, have multiple procs read/writting data as fast as possible (you don't need much disk space, writting small files is fine).
Using recursive SCM_RIGHTS to use huge numbers of PF_LOCAL sockets (it isn't accounted after SCM_RIGHTS).
Find a simple setuid() app. and fork/exec bomb to that.
BIND is fairly simple to DOS attack from the local machine.
HTTP is also likely to be easy to DOS, due to needing 1 proc per connection... where you, as the attacker, don't need anything like that.
portmap almost always doesn't do any auth checks on assign X service to Y port calls... so that would easily take out NFS, FAM and/or any other RPC services you are running (re-direct the ports to HTTP for even more fun).
Run something like: (while:; do; mkdir a; cd a; done)... takes out more than a few directory walking applications (often including backups).
There's probably some more obvious ones, if I thought about it a bit.
I guess I wouldn't mind if Fedora came with defauts that stopped a forkbomb on my box, if I was stupid enough to run one... but if they fucked it up and set it too low I guarantee I'd be very pissed.
Realistically it doesn't have to be a product that someone actually wants for them to be convinced that they want to buy it.
As the article said, you can make a restaurant work with mediocre food... but it's much easier to have good food. But let's see the reasons for what you are saying...
We were building an application for a niche market of government accounting employees
Happily ignoring the "don't sell to govt. from a startup" that the article mentioned (and has been said elsewhere).
I have to admit that it was very frustrating to see the marketing people from SAP go to these small governments and convince them to spend ten times as much...
And happily ignoring the "don't compete directly with much larger businesses" that the aritcle mentioned (and has been said elsewhere).
In today's world, being able to actually sell something is more important than making something useful
As far as I'm concerned, if someone copies and uses something I created (a painting or drawing for example) without permission or regard for my rights as its creator, even if it doesn't technically "cost" me anything monetarily, it is still theft. This has happened to me before, and I'll be damned if I didn't feel like I'd been stolen from.
Yeh, I know how you feel... you just stole a couple of minutes of my life with your moronic ramblings. And unlike you, with your pretty IP, I can't get them back... THIEF.
Of course I've now used and abused something (your text) without permission or regard for your rights... Oh my GOD, I'm a THIEF too.
Of course, we're both entitled to STEAL in this way... but it's ok, no point in making up different words for different things one word is more than enough. Oh, and give me my air back... THIEF.
What would (minus any syscall overhead) ultimately be one mov instruction to give something the kernel has always known to userland turns into literally THOUSANDS of instructions
Of course that mov breaks horribly with HT CPUs, then you might want to know about multicores... and god forbid you need anything like a numa layout.
Oh, and I've yet to see an application that needs this kind of information in a tight loop... and if you do most people would just call sysconf(_SC_NPROCESSORS_CONF) or sysconf(_SC_NPROCESSORS_ONLN) once, and put it in a variable.
A lot of the "bloat" I see is from people saying "Oh, we can't possibly link to a decent string library that's way too much bloat for what we need"... and so, of course, everyone reimplements 65% of one badly. Repeat X times for all the other common non-problems.
OSI seem to be based on the idea that there is an ideal that we can derive from the aether that will cover everything.
I doubt it... much more likely is that OSI recognize that as an OpenSource developer I'm not a lawyer. And if they have 50 "approved" licenses that are incompatible then any kind of OSI branding is worthless and people will start ignoring them.
For instance the fact that OSI "approved" Sun's CDDL was worth nothing to me, and I've had advise from several other places that I should avoid seeing CDDL code just as much if it were still fully proprietary.
Slashdot Mirror
You need to specify "display: inline-table;" for the table element.
That's not really true, the market didn't value diluting the stock in that way (and, in many ways, still doesn't). So the companies went from paying employees to not having to do so (or paying them much less anyway). It's like if you throw away a million dollar painting, and I "pay" for a local housing project with it (wow, I'm a nice guy :) ... you didn't "lose" a million dollars, or lose a bunch of houses and the IRS aren't going to allow you to put either down as a TAX break :).
Of course, you'd then be free not to throw anything away ... just in case it might be worth something, even though you'd still only have a worthless old painting if you hadn't thrown it away. And some people seem to be applying similar reasoning to Stock Options.
You have a decent SCM application, that does the right thing. Or you re-diff against the previous version. Or you look through by hand and take the obvious bits. Or you have a minor fork which is "what apple has", and then you can compare the results of both builds. Or you do nothing, but in that case you haven't lost anything. You've just not used what they gave you.
Possibly, possibly not ... but if it's supposedly so bad for them to do what they are doing, why are you complaining.
They'll learn better from their own mistakes, if they are mistakes. All the complaining does is piss them off, and blind them to any problems they are creating because they won't want to admit the annoying people are right.
And as I said in the previous story, there is a big difference between "what customers want" and "what upstream wants" ... sometimes requiring two seperate patches on seperate deadlines. As long as they obey the GPL, I really couldn't care if they never send patches back ... they aren't required to do so, and if the KDE people wanted that they should have used some other license. You could argue that it's "nice" to do so, and part of the "community" spirit ... but given their entire GUI foundation is proprietary I don't see that they'd benifit much.
In many ways this is like saying that when OpenBSD develop new code X, they should be required to backport it to NetBSD ... no, sorry, it doesn't work that way ... and sometimes that part can take 10x as long as writting the code, if you do it, so I certainly would never use any source that required it.
No, what really isn't loaded are the bits of code that aren't actually loaded. Maybe you meant eventaully paged out, and not paged back in ... but then you'd just be mostly wrong.
If you think a win32/linux server with a GUI loaded and a monitor turned off is "just as well off" as a linux server that never loaded the GUI in the first place, you are nieve at best.
Tracking structures have to be allocated, even if everything is paged out (so it can be paged back in) ... this is eating kernel memory. Pretty much all code allocates resources (esp. GUI code), at best that's swapped to disk and is just screwing your swap and taking more kernel tracking space ... but things like file descriptors etc. are just more unswappable kernel memory. At the best of times that GUI code is actually doing something (thus keeping "a bit" of it's RAM footprint in core), and god help you if someone enabled a screensaver (luckily win32 comes with one turned on by fdefault for the login screen -- yeh). Any activity at all is also affecting the scheduler and VM, as it has to deal with the tasks that wouldn't be there otherwise.
To be fair Red Hat developers are using the same tools and same OS as upstream, Apple developers aren't. This means Red Hat just needs to know "does it work" but apple needs to know "does it work" and does it the ported version work. Also, all the commercial Linux companies are carrying around huge patchsets on upstream ... sometimes because upstream is slow/dead ... but also because "making it work for my customer tomorrow" is very different than "getting it into upstream", and even with the same tools/OS require two completely seperate patches.
So, while I have no love for Apple and think the 5th Free Unix is going to do as well as the 6th[1] in the long run, I can't fault them for what I see them doing with KHTML.
[1] Yes, sun fan boys, that's a random stab in the dark at you.
You can do this. RHEL, for instance, has had driver update CDs ... where you couldn't install normally, so you insert the driver update CD before your install CD and now you can. Just like XP/Solaris/OSX/whatever-comercial-OS.com.
I'm trying to decide if your post was supposed to be a "Troll Tuesday" one, or if indeed you are just clueless. I'm thinking, mostly, it's just cluessness but I'm hoping not.
I'm inclined to agree ... however that, IMNSHO, shows how open they want to be. You can have any color you want, as long as it's black.
While I wouldn't advocate that you violate the copyright protection on a DVD, doing so is vastly different than walking into a store and stealing a copy off the shelf. By the same measure copyright violation cannot be compared to stealing a candy bar, and I assume you know this and are intentionally lying. A more accurate comparson would be to not being able to afford the candy and so going home and making your own candy using a recipe for said store candy that you "aquired" from somewhere.
And even that isn't a great comparison, but at least both involve breaking the same law.
Err, removing their support of this bill? Not speaking is an action. I find it hard to believe that they think the govt. will move to Linux (basically the only stick they have) over this. Apart from that I really don't see what anyone could do to them to stop them doing the right thing.
I guess I can see it if the the govt. hinted at restarting a monopoly charge, but I'd have expected it to happen a while ago.
Even though we're talking about Solaris, we're not restricted to using the output formats that are also supported on Solaris.
Which is why OSI have publicly stated they want to cut down the number of licenses they approve, because saying the CDDL is fine is just making them irrelevant.
When Solaris source code is available, it'll be the 7th major unixlike OS with source code freely available (Linux, {Net, Free, Open, DFly}BSD, Darwin) and Solaris will be incompatible with all the others. Worse than that is that every person even vaguely qualified has advised me to not even look at the Solaris source, as you are likely to become tainted and unable to contribute to GPL source code.
And before you speak out of ignorance, yes, Linux dual licenses bits of code and so shares with the other 5 above.
For Schwartz, or random Sun Fan boys, to say "It's not the CDDL's problem, it's the GPL's" is at best ignorant and more likely just lies.
So now you have a somewhat opaque dep. on iptables ... and it allows all local users to also bind to the web server port. Not exactly an improvement
SELinux doesn't even really help, you just have to start the httpd as root and "know" it'll just bind to the port/chroot/whatever and then drop privs.
I know my web server does that, and I apache-httpd seems to (although I'm not sure how easy it is to chroot it).
Has anyone done any benchmarks I've missed that show apache-httpd is actually faster using threads, for real platforms? Yes, I realise win32 requires them ... I don't care.
Over double that, it's 7 years from release. Which means that box you don't want to have to care about can tick over serving DNS for about 6 years all the while getting errata, but you can also have the latest postgresql on another box.
Intersting ... I'd have assumed for multiple core developers you'd just have a release repo. and one developer repo. each pulling changesets into each others repo. and doing a single pull into the core repo. every now and then (probably with a testing repo. in between). Did you try that?
That's kind of how I work, as a single core developer ... the release repo. is just a branch with remote tags
Well, I'd say it's pretty much required to use "tla changes --diffs" instead of without the option. And I'd have to add:
tla addtla tree-lint
tla mv
tla rm
The later two, not so much ... but still. Saying that, I've been pretty happy with Arch.
Errr... sorry to break it to you but CSV files will often have a header that looks like:
...making it just as obvious what each column is for. Actually it's often much better, I can write about 3 lines of perl that will always correctly parse the CSV file and put the values into a hash ... God knows how much time you'd have to spend on the XML version.
Because, of course, the XML file is really an "XML subset" or the parser needs to deal with the fact that the "record" has 10 extra child nodes (whitespace) and will often have extra nodes or changed nodes like having only a nickname, not having a nickname, having the nickname as a child of firstname or having the firstname, lastname and nickname inside another "person" node -- because, hey, it's XML so it's magic.
And if you went the XML subset route, you are screwed because people will assume otherwise.
There's probably some more obvious ones, if I thought about it a bit.
I guess I wouldn't mind if Fedora came with defauts that stopped a forkbomb on my box, if I was stupid enough to run one ... but if they fucked it up and set it too low I guarantee I'd be very pissed.
As the article said, you can make a restaurant work with mediocre food ... but it's much easier to have good food. But let's see the reasons for what you are saying...
Happily ignoring the "don't sell to govt. from a startup" that the article mentioned (and has been said elsewhere).
And happily ignoring the "don't compete directly with much larger businesses" that the aritcle mentioned (and has been said elsewhere).
If you ignore all other advise, sure.
Yeh, I know how you feel ... you just stole a couple of minutes of my life with your moronic ramblings. And unlike you, with your pretty IP, I can't get them back ... THIEF.
Of course I've now used and abused something (your text) without permission or regard for your rights ... Oh my GOD, I'm a THIEF too.
Of course, we're both entitled to STEAL in this way ... but it's ok, no point in making up different words for different things one word is more than enough. Oh, and give me my air back ... THIEF.
Of course that mov breaks horribly with HT CPUs, then you might want to know about multicores ... and god forbid you need anything like a numa layout.
Oh, and I've yet to see an application that needs this kind of information in a tight loop ... and if you do most people would just call sysconf(_SC_NPROCESSORS_CONF) or sysconf(_SC_NPROCESSORS_ONLN) once, and put it in a variable.
A lot of the "bloat" I see is from people saying "Oh, we can't possibly link to a decent string library that's way too much bloat for what we need" ... and so, of course, everyone reimplements 65% of one badly. Repeat X times for all the other common non-problems.
Hint: This is false for both FreeBSD and Linux.
I doubt it ... much more likely is that OSI recognize that as an OpenSource developer I'm not a lawyer. And if they have 50 "approved" licenses that are incompatible then any kind of OSI branding is worthless and people will start ignoring them.
For instance the fact that OSI "approved" Sun's CDDL was worth nothing to me, and I've had advise from several other places that I should avoid seeing CDDL code just as much if it were still fully proprietary.