But to push an OS to users on the grounds that it's written in $LANGUAGE is, IMO, just silly.
That they used C++ was just one point in their description. This has more relevance when describing an open-sourced OS because it invites C++ developers to take a look.
You're attributing the success of all 3rd party developers to Microsoft's business practices. Had any other company's OS become dominant, 3rd party companies would probably have had at least the same success; possibly more if Microsoft's APIs weren't the crap they are.
The GNU/Linux community has no control of how.NET is implemented, and that leaves us in a vulnerable position.
What the hell are you talking about? The CLR aspect of.NET -- the part Mono deals with -- is a specification submitted to a standards body (Mono doesn't mess with Passport/Hailstorm). Ximian, being the implementors of this specification, make every single decision about how it is implemented. Wait a minute...RMS didn't found the EMCA...they can't be trusted!
If you believe that Ximian is not part of the "GNU/Linux community", someone else from the beloved community could implement the specification and by doing so, will have full control over how it is done.
A marathon has an end. The battle for desktop dominance does not. You're going to have to try and pass the leader at some point, but there's no finish line to tell you when to do it.
The "slipstream" thing might make sense KDE was strapped for resources (developers) but the very nature of OSS doesn't allow this to happen (not now, at least). There's practically no shortage that would limit the number of new ideas that can be tested out. KDE should try to pull ahead because it definitely has the ability to do so (it probably is trying already, though).
"you can't rid the world of cannibalism by eating all the cannibals."
Sure, but it just might work in this situation: Ximian uses MS's practices and puts MS out of business (stop laughing). Ximian goes out of business due to lack of funding. Problem solved.
Do you really think casual users of this kind of craft will have the skill to pull off an auto rotation maneuver?
Sure...if their life depended on it. But autorotation could be programmed into backup systems on the craft...may not be as effective as a trained pilot, but better than a 16-year old with a learner's permit.
Dude...there are people who want to write GPL'd software. Some of those people also like(d) Delphi. The free version of Kylix is a great thing.
Though it's unlikely, Borland could conceivably discontinue the free version (gradually, by dumbing it down with each version) and get people to buy it instead. But Kylix will probably never get enough users for Borland to get away with that.
Was borland.com slashdotted? Couldn't get to it. Thought they could handle a decent amount of traffic.
What if I wish to modify and redistribute my own version of kylix?
Sure, it's a nice concept...being able to change something you don't like. But how often do you think you will need to modify the Kylix source to get the job done? What's wrong with just using a program?
Sure, in the military sense, you could argue that a fortified location is easier to defend than to break in to. But communications programs aren't the same as forts. Communications programs have other stuff to do. Security is secondary to getting useful things done. A fort has to purpose but to defend itself so security is a lot easier.
You are less likely to be carjacked if you park in a large parking lot.
Sure, but carjacking is a much more involved and risky task than attacking an unsecured computer system.
First you say:
It really won't matter in the end, from a security standpoint, because well-designed security frameworks are undefeatable through principle, not through secrecy.
then (saying you don't need security):
Of course I wouldn't [expose network info] , because my network is far from secure in terms of code quality (it runs Linux), but then again, I don't run a network containing sensitive information that could cause some serious damage...
then (saying that you need security):
No, I don't need schmucks using my wireless T-1 for DDoSing EFNet. They have enough problems as it is.
then (relying on obscurity again):
Wrong again. The fact that there are millions of high bandwidth connections on much less secure systems (like all of those home broadband users who don't know that an Internet exists outside of email and the web) make it much less of a problem to me than if I was securing critical data.
Ripping CD's has more to do with media speed than the CPU.
Could an MP3 or Ogg encoding program really outpace even a 32x CD-ROM drive? Doesn't 32x mean 32 times normal playback speed -- allowing it to go through the entire contents of a CD in under 3 minutes (74min / 32)? Can PC chips today encode a whole CD in 3 minutes? Or am I confused about the 32x thing?
I thought CISC code generation was more difficult than RISC code generation. CISC has more instructions to deal with while RISC (I've only seen MAL vs. HC11 asm) seems much easier to write a compiler for. RISC instructions are more consistent and complete and come in every useful permutation while x86 asm, with it's weird and different-sized registers, requires a little more thought (when coding by hand).
I know that GCC output has been heavily optimized for x86 chips but is the program's code itself optimized too? The extra time spent making GCC itself run faster would be better spent optimizing output code. Does anybody know if GCC itself is optimized at a lower level (lower than plain algorithm/data-structure optimization)?
cygwin is better than command.com but even better is the bunch of utilities that have been ported with the DJGPP compiler. As far as I know they run without any emulation layer and the speed difference is very visible ('ls' takes forever in cygwin).
DJGPP Home Mirror list Good place to start - (get the.zip files that end with the letter 'b'. You can usually guess the program from the first few letters. Ones you probably want are bsh, grep, find, gwk, gzip, tar, lss, sed)
Not that I'm pro-MS or anything, but would precompiled code be much of a problem? Couldn't the objects be stored in memory the same way? This would allow objects to be passed among different platform since it's only the class's code that is natively compiled...or are there situations where you have to pass the class code around too? Maybe when sending an implementation for an interface...?
It is accepted that extra effort is needed to make something cross-platform. Microsoft could then easily make extensions that are difficult to implement on other platforms simply by providing low-level functions that don't have an equivalent in other OSes (could result in lower speeds or inability to run on Linux).
Java may not have invented VMs but it definitely helped prove the feasibility of an semi-compiled, garbage-collecting language. It also popularized heirarchical packaging, combining interface & implementation and the use of interfaces instead of allowing multiple inheritance.
Though Java is pretty popular now, it's not like everyone started using it right away. Microsoft's has an unfair advantage that is reflected in the fact that there were lots of publications and so much momentum behind it even before it was released. So many drones are flocking to C# just because of Microsoft marketing, not because of the merits of the language.
So...you say that there are a variety of languages that can be compiled to CLI and that Sun was displaying 'stupidity' by making a new language and throwing out the old ones? Sun felt that existing weren't as clean as Java for the environment they wanted to provide. If Microsoft's approach works with so many other languages, why did they make C#? Maybe because they realized that Java was a very good language for.NET but they needed one of their own. C# is just Java with different keywords/symbols.
Slashdot Mirror
That they used C++ was just one point in their description. This has more relevance when describing an open-sourced OS because it invites C++ developers to take a look.
The name really is nothing to do with Atheism, or God, or anything else.
Except that Athena is the Greek god of war and wisdom.
You're attributing the success of all 3rd party developers to Microsoft's business practices. Had any other company's OS become dominant, 3rd party companies would probably have had at least the same success; possibly more if Microsoft's APIs weren't the crap they are.
The GNU/Linux community has no control of how .NET is implemented, and that leaves us in a vulnerable position.
What the hell are you talking about? The CLR aspect of .NET -- the part Mono deals with -- is a specification submitted to a standards body (Mono doesn't mess with Passport/Hailstorm). Ximian, being the implementors of this specification, make every single decision about how it is implemented. Wait a minute...RMS didn't found the EMCA...they can't be trusted!
If you believe that Ximian is not part of the "GNU/Linux community", someone else from the beloved community could implement the specification and by doing so, will have full control over how it is done.
A marathon has an end. The battle for desktop dominance does not. You're going to have to try and pass the leader at some point, but there's no finish line to tell you when to do it.
The "slipstream" thing might make sense KDE was strapped for resources (developers) but the very nature of OSS doesn't allow this to happen (not now, at least). There's practically no shortage that would limit the number of new ideas that can be tested out. KDE should try to pull ahead because it definitely has the ability to do so (it probably is trying already, though).
"you can't rid the world of cannibalism by eating all the cannibals."
Sure, but it just might work in this situation: Ximian uses MS's practices and puts MS out of business (stop laughing). Ximian goes out of business due to lack of funding. Problem solved.
Do you really think casual users of this kind of craft will have the skill to pull off an auto rotation maneuver?
Sure...if their life depended on it. But autorotation could be programmed into backup systems on the craft...may not be as effective as a trained pilot, but better than a 16-year old with a learner's permit.
Will my SkyCar turn into a suitcase?
Dude...there are people who want to write GPL'd software. Some of those people also like(d) Delphi. The free version of Kylix is a great thing.
Though it's unlikely, Borland could conceivably discontinue the free version (gradually, by dumbing it down with each version) and get people to buy it instead. But Kylix will probably never get enough users for Borland to get away with that.
Was borland.com slashdotted? Couldn't get to it. Thought they could handle a decent amount of traffic.
What if I wish to modify and redistribute my own version of kylix?
Sure, it's a nice concept...being able to change something you don't like. But how often do you think you will need to modify the Kylix source to get the job done? What's wrong with just using a program?
if some loser with a .45 comes along
Hmm...doesn't seem like the guy with the .45 is the 'loser' in this situation. :)
Sure, in the military sense, you could argue that a fortified location is easier to defend than to break in to. But communications programs aren't the same as forts. Communications programs have other stuff to do. Security is secondary to getting useful things done. A fort has to purpose but to defend itself so security is a lot easier.
You are less likely to be carjacked if you park in a large parking lot.
Sure, but carjacking is a much more involved and risky task than attacking an unsecured computer system.
First you say:
then (saying you don't need security):
then (saying that you need security):
then (relying on obscurity again):
I'll take your system apart and just steal the hard drive and I'm gone with your data...
Possibly encrypted data.
Imagine how much more would be saved if people stopped using anything that consumed electricity. Buying a G4 from Apple is just not an option.
A G4 tower takes up the space of 5 or so of these.
Dude, the thing is 2"x6"x9". You can fit at least 15 of them in the space of a G4 Tower.
Ripping CD's has more to do with media speed than the CPU.
Could an MP3 or Ogg encoding program really outpace even a 32x CD-ROM drive? Doesn't 32x mean 32 times normal playback speed -- allowing it to go through the entire contents of a CD in under 3 minutes (74min / 32)? Can PC chips today encode a whole CD in 3 minutes? Or am I confused about the 32x thing?
I thought CISC code generation was more difficult than RISC code generation. CISC has more instructions to deal with while RISC (I've only seen MAL vs. HC11 asm) seems much easier to write a compiler for. RISC instructions are more consistent and complete and come in every useful permutation while x86 asm, with it's weird and different-sized registers, requires a little more thought (when coding by hand).
I know that GCC output has been heavily optimized for x86 chips but is the program's code itself optimized too? The extra time spent making GCC itself run faster would be better spent optimizing output code. Does anybody know if GCC itself is optimized at a lower level (lower than plain algorithm/data-structure optimization)?
So you think that in other discussion groups (ones without karma), people don't defend their messages?
cygwin is better than command.com but even better is the bunch of utilities that have been ported with the DJGPP compiler. As far as I know they run without any emulation layer and the speed difference is very visible ('ls' takes forever in cygwin).
DJGPP HomeMirror list
Good place to start - (get the
Not that I'm pro-MS or anything, but would precompiled code be much of a problem? Couldn't the objects be stored in memory the same way? This would allow objects to be passed among different platform since it's only the class's code that is natively compiled...or are there situations where you have to pass the class code around too? Maybe when sending an implementation for an interface...?
It is accepted that extra effort is needed to make something cross-platform. Microsoft could then easily make extensions that are difficult to implement on other platforms simply by providing low-level functions that don't have an equivalent in other OSes (could result in lower speeds or inability to run on Linux).
Java may not have invented VMs but it definitely helped prove the feasibility of an semi-compiled, garbage-collecting language. It also popularized heirarchical packaging, combining interface & implementation and the use of interfaces instead of allowing multiple inheritance.
Though Java is pretty popular now, it's not like everyone started using it right away. Microsoft's has an unfair advantage that is reflected in the fact that there were lots of publications and so much momentum behind it even before it was released. So many drones are flocking to C# just because of Microsoft marketing, not because of the merits of the language.
So...you say that there are a variety of languages that can be compiled to CLI and that Sun was displaying 'stupidity' by making a new language and throwing out the old ones? Sun felt that existing weren't as clean as Java for the environment they wanted to provide. If Microsoft's approach works with so many other languages, why did they make C#? Maybe because they realized that Java was a very good language for .NET but they needed one of their own. C# is just Java with different keywords/symbols.
Remember, nobody's complaining about Google's method. Even the complaint letter cites Google's method as a valid alternative if payment is involved.