The other question that this brings up, is since any Windows DLL is dynamically-linked into a Windows.exe, does that mean that GPL software that uses/windows/*.dll can't be released under the GPL? And who is violating the GPL in those cases? The person who wrote the software, or Microsoft?
All of this fascinates me, since the interoperation of different executables and dynamic libraries is often cascading (Linking to one library pulls in it's dependents, etc.) and different from one case to the next, so at what level can you link to proprietary code and still be okay with the GPL?
Depending on how weaselly the lawyer, that may open the loophole for drivers anyway. The Linux kernel is capable of running headless. Therefore, it is not intended to REQUIRE a video card driver. It can run silent, so it's not intended to require sound card drivers, either. Etc, etc.
In all honesty, I see a drastic difference between your basic runtime dynamic libraries and device drivers, from a userland perspective.
From a hard-code perspective I can see their point, since changing the library changes functionality.
So I guess it comes to this: At what level does the GPL apply? On a per source file basis (in which case, static-linking is okay), a per-compiled binary basis (static-linking is not okay, dynamic okay), or an as-run binary basis (no static or dynamic linking at all.) Is there a choice?
If not, I would personally suggest they add (cc)'s option system. "GPL Source", "GPL Dynamic-OK", and "GPL Hardline".
nVidia and ATi can't open their drivers completely. Last I checked there was portions of both drivers written by other companies that they don't own the rights to.
So first those companies would have to agree to GPL'ing their libraries that are used, then nVidia/ATi would have to go over their code to make sure they didn't license anything else.....
It would be far easier to continue offering the closed-source drivers, and then offer the specifications to the OSS community for a from-scratch driver made from the appropriate documentation. But then ATi has nVidia's specs and vice-versa. And there's the rub.
Proprietary drivers should never have been allowed to link to the linux kernel - doing so makes them a derivitive (yes, even those drivers that predate the linux kernel).
So what you're saying is using the module interface constitutes a derivative work?
I thought modifying code / re-using non-standard libraries = derivative work.
Attaching to an "open" interface = Attaching to an "open" interface.
So the Samba client, since it attaches to Microsoft's SMB service, is a derivative work? That kind of thinking could potentially open the OSS community to some serious legal issues.
I'm all for Open Source, but each company/coder/developer has the right to choose for themselves what type of release is best for them. GPL people want share and share alike. As long as there isn't GPL (or similar) code in the closed drivers, I don't see where the issue is.
The problem there is that it doesn't require 100% market share and no competitors to be a Monopoly.
You fall afoul of the Monopoly/Anti-Trust laws when you use your market share to bully another company out of the market. The best example of this would be Microsoft requiring OEM's to sell Windows with every PC they shipped or they would be charged higher rates for Windows.
With the Intel/Skype issue, I don't think it's so much that Intel said "We'll pay you to make a version that does 10 lines on ours only" as it is how they advertised it. According to everything I've read, (Including this.) it's not a "Intel is partnering with Skype to bring you 10-way Conference Calls, only on Intel D!", it's a "Intel is the only CPU powerful enough to power this!" which should fall afoul of truth-in-marketing laws, not anti-trust. But with the anti-trust suit already pending, it seems silly to start a seperate suit.
With Universal Binaries, why was the laptop switch not made to Intel, and maintain G5 PowerMacs as a line of systems too? (iMac/Mac Mini/etc. could be whichever was cheaper at the time.)
Of course, the major question is, with there being an IBM contract still in place for G5s, is this going to be the case? (Steve said they were switching to Intel, but I don't remember reading the key "All Models" phrase...)
For those who aren't fond of the changes to the GPL from v.2 to v.3, why not just vet your complaints about v.3 with a structured rebuttal, and then go on developing v.2 software until it's fixed or something better comes along?
Also, there are plenty of "Open-Source" Licences. (MPL, LGPL, BSD, CopyCat, etc.) Is there something the GPL v.3 does that the others (GPL v.2 included) don't do?
How much play time have the members of the dev team put into playing other MMO's, either to look for features that may benefit WoW, or to look for stumbling blocks to avoid?
For example, unclear quest instructions like in FFXI or the staggered leveling system in City of Heroes, etc.
The only reason for us going Win98 > Win2000 was roaming profiles and "real" domain logins, and it's the smallest change of UI to make things simpler on those less-than computer literate.
XP may have an "optimized" kernel, but it also adds more new code as well. NX-bit support in the kernel, additional GUI code, feature creep, etc.
Each generation will have optimisations as well as added code. While I can't speak to XP in particular, unless there's a drive for optimization, the two normally tend to balance out at best during a normal development cycle. Otherwise, there would be no reason to upgrade hardware, as Windows XP would run faster on a Pentium II 266 with 128MB RAM than NT 3.5. (Since, after all, the XP kernel is 3 generations of optimizations.)
Actually, we bought Windows 2000 OEM copies with the parts for the new systems we needed. One copy per stick of RAM. Cost was the same as the last time upgrade licenses were available.
Actually, I've never seen anyone quote where this is in the license agreement, nor have I found it myself. Can you point me to where that is? It'll really help for in the future when we add more machines.
Slashdot Mirror
The other question that this brings up, is since any Windows DLL is dynamically-linked into a Windows .exe, does that mean that GPL software that uses /windows/*.dll can't be released under the GPL? And who is violating the GPL in those cases? The person who wrote the software, or Microsoft?
All of this fascinates me, since the interoperation of different executables and dynamic libraries is often cascading (Linking to one library pulls in it's dependents, etc.) and different from one case to the next, so at what level can you link to proprietary code and still be okay with the GPL?
Depending on how weaselly the lawyer, that may open the loophole for drivers anyway. The Linux kernel is capable of running headless. Therefore, it is not intended to REQUIRE a video card driver. It can run silent, so it's not intended to require sound card drivers, either. Etc, etc.
In all honesty, I see a drastic difference between your basic runtime dynamic libraries and device drivers, from a userland perspective.
From a hard-code perspective I can see their point, since changing the library changes functionality.
So I guess it comes to this: At what level does the GPL apply? On a per source file basis (in which case, static-linking is okay), a per-compiled binary basis (static-linking is not okay, dynamic okay), or an as-run binary basis (no static or dynamic linking at all.) Is there a choice?
If not, I would personally suggest they add (cc)'s option system. "GPL Source", "GPL Dynamic-OK", and "GPL Hardline".
nVidia and ATi can't open their drivers completely. Last I checked there was portions of both drivers written by other companies that they don't own the rights to.
So first those companies would have to agree to GPL'ing their libraries that are used, then nVidia/ATi would have to go over their code to make sure they didn't license anything else.....
It would be far easier to continue offering the closed-source drivers, and then offer the specifications to the OSS community for a from-scratch driver made from the appropriate documentation. But then ATi has nVidia's specs and vice-versa. And there's the rub.
Proprietary drivers should never have been allowed to link to the linux kernel - doing so makes them a derivitive (yes, even those drivers that predate the linux kernel).
So what you're saying is using the module interface constitutes a derivative work?
I thought modifying code / re-using non-standard libraries = derivative work.
Attaching to an "open" interface = Attaching to an "open" interface.
So the Samba client, since it attaches to Microsoft's SMB service, is a derivative work? That kind of thinking could potentially open the OSS community to some serious legal issues.
I'm all for Open Source, but each company/coder/developer has the right to choose for themselves what type of release is best for them. GPL people want share and share alike. As long as there isn't GPL (or similar) code in the closed drivers, I don't see where the issue is.
Could be worse.
Here we can have AT&T or Time-Warner RoadRunner. (Or pay Earthlink for them to pay TWC for RoadRunner, or pay AOL to pay TWC........)
So I caught, but not until after I replied.
So it was my reply's reply that I hoped you saw, you see.
Oops! Missed parent post.
When one pleads the fifth, they're normally declaring their intent to use their 5th Amendment Rights.
So I would presume pleading the second is the same, but for the 2nd Amendment Rights.
Well, if it's not stable for you, could you list the bugs and issues you're having?
Either here or in the Firefox Bugzilla would be great, and don't forget your TPS cover sheet!
The problem there is that it doesn't require 100% market share and no competitors to be a Monopoly.
You fall afoul of the Monopoly/Anti-Trust laws when you use your market share to bully another company out of the market. The best example of this would be Microsoft requiring OEM's to sell Windows with every PC they shipped or they would be charged higher rates for Windows.
With the Intel/Skype issue, I don't think it's so much that Intel said "We'll pay you to make a version that does 10 lines on ours only" as it is how they advertised it. According to everything I've read, (Including this.) it's not a "Intel is partnering with Skype to bring you 10-way Conference Calls, only on Intel D!", it's a "Intel is the only CPU powerful enough to power this!" which should fall afoul of truth-in-marketing laws, not anti-trust. But with the anti-trust suit already pending, it seems silly to start a seperate suit.
Hence why it's only GPL violators who fall afoul of the law.
Since they're not including source in re-distributed GPL code, the IP ownership isn't being reported properly.
With Universal Binaries, why was the laptop switch not made to Intel, and maintain G5 PowerMacs as a line of systems too? (iMac/Mac Mini/etc. could be whichever was cheaper at the time.)
Of course, the major question is, with there being an IBM contract still in place for G5s, is this going to be the case? (Steve said they were switching to Intel, but I don't remember reading the key "All Models" phrase...)
For those who aren't fond of the changes to the GPL from v.2 to v.3, why not just vet your complaints about v.3 with a structured rebuttal, and then go on developing v.2 software until it's fixed or something better comes along?
Also, there are plenty of "Open-Source" Licences. (MPL, LGPL, BSD, CopyCat, etc.) Is there something the GPL v.3 does that the others (GPL v.2 included) don't do?
Actually, animation from France is also considered "anime".
It's mostly just to deliniate "Import" from "Domestic".
How much play time have the members of the dev team put into playing other MMO's, either to look for features that may benefit WoW, or to look for stumbling blocks to avoid?
For example, unclear quest instructions like in FFXI or the staggered leveling system in City of Heroes, etc.
Nah.
My Very Eager Mother Just Served Us Nine Pizzas Sans Elephants.
(Mercury, Venus, Earth, Mars, Jupiter, Saturn, Uranus, Neptune, Pluto, Sedna, Eris.)
Exactly.
The only reason for us going Win98 > Win2000 was roaming profiles and "real" domain logins, and it's the smallest change of UI to make things simpler on those less-than computer literate.
Optimized means that some part of the kernel has better performance.
My point is that while I'm sure there were some optimizations, there was enough extra added that the optimizations may be a moot point.
Plus there's a lot more to an operating system than just the kernel.
XP may have an "optimized" kernel, but it also adds more new code as well. NX-bit support in the kernel, additional GUI code, feature creep, etc.
Each generation will have optimisations as well as added code. While I can't speak to XP in particular, unless there's a drive for optimization, the two normally tend to balance out at best during a normal development cycle. Otherwise, there would be no reason to upgrade hardware, as Windows XP would run faster on a Pentium II 266 with 128MB RAM than NT 3.5. (Since, after all, the XP kernel is 3 generations of optimizations.)
Many thanks. I've been looking for official confirmation of that for a while.
Well, re-read your original post, because you typoed. The original post says Quantitative.
And I'm going to drop the weasel words claim, as it can only degrade into a flame war.
"Quantative Speed Analysis" = "Benchmark Results"
(Yes, Mods, I know I'm duping myself.)
Actually, we bought Windows 2000 OEM copies with the parts for the new systems we needed. One copy per stick of RAM. Cost was the same as the last time upgrade licenses were available.
Actually, I've never seen anyone quote where this is in the license agreement, nor have I found it myself. Can you point me to where that is? It'll really help for in the future when we add more machines.
What performance hit from Win98 -> XP were you expecting? Could you post the quantitative speed analysis numbers, please?
/ evaluation/sysreqs/default.asp
/ sysreqs.mspx
http://www.microsoft.com/windows2000/professional
64MB RAM Minimum, 133MHz CPU Minimum.
http://www.microsoft.com/windowsxp/pro/evaluation
128MB RAM Minimum (Though it'll install on 64) and 300MHz CPU.
So if 2000 needs less to function, that leaves more for the rest of our software. And stop with the Weasel Words.