90% of unix sysadmin is accidentailly arcane, and nobody has ever bothered to fix it because it has indirectly brought about profitable support contracts and the job security and loyalty of sysadmins (who incidentally tie up their mastery of Unix with their personal self-worth).
By the same argument, Unix permissions are so complicated nobody (to a first approximation) uses them well either. Although the front-emd of Unix-style might be more streamline, having to create an "arbitrary" number of directories and groups doesn't sound less complex to me.
Well, there's a difference between "I can get by with what I know just fine, thanks" and "Most of the complicated queries you might do are better expressed in a proper programming language"...
The problem is not that their not straight ASCII -- they are not straight ISO 8859-1. Windows uses several reserved characters for things like curly quotes, and these characters are not implemented in Unix character sets.
I would imagine if you had a MS True Type font, it would look OK. (Or, Slashdot could shut everyone up by implementing a curly quote to HTML & code replacer in it's submit processor.)
Lay off the coffee, spaz. For the record I have no hatred, nor am I part of the linux community.
Since you seem to know so much about the situation, how about letting us know what Troll's profits are, before and after KDE got popular. (I brought up ActiveX component companies because there's several hundred of them, which must indicate that it's a slightly profitable business at best. See http://www.componentshopper.com/)
Troll Tech was merely business savvy and opportunistic.
Troll Tech wasn't savvy at all -- they were just lucky.
They were in the business of selling a GUI toolkit for client Unix systems -- a market that is pretty much dwindling to nothing, and what's left is heavily standardized on Motif.
Doesn't sound like the wisest market to be in. There's probably a half-dozen companies that make VisualBasic widgets that are larger and more profitable than TrollTech.
The only thing that saved them is that KDE chose to use their freeware stuff, and KDE turned out to be such a polished product that Linux users were willing to hold their nose over the licencing issue and develop apps for Qt. If KDE didn't exist, or was buggy/defective, TrollTech would still be two guys in a basement somewhere.
SQL (as a language, not a feature set) is designed for set processing. A particular SQL implementation is heavily optimised for the data store implementation used in it's RDBMS.
I'd think you'd be crazy not to write as much of your applicaiton in SQL as possible, even though it's ugly lookin' and it will never be more than a portion of your code base. Ignoring a powerful and efficent tool like that seems foolhardy. (Not that I haven't used Postgresql or mysql, so perhaps their dialects are too limited to be as useful.)
Trying to write the loops to manage set processing in a "proper" language will almost always be more inefficent and error prone. Not to mention the insanity of running that code interpreted in Perl/PHP/ASP/Whatever on your webserver.
The thing is, if you are doing a reasonable upgrade (like once every two years to get a 100% clock speed boost), it makes more sense to buy a new system. Realistically, you probably don't want the video card and hard drive from your PII-333 in your brand new PIII-700 system.
But, certain people out of concern for penis envy or gaming performance or whatever upgrade far more often. These guys might have a GeForce2 and a DMA66 drive in their Celeron 300A @ 450, and when they upgrade to an AMD 800, it makes sense to move the parts over. At that point the old computer is pretty much worthless with no drive and no video, and it makes sense to recycle the case.
Maximum bytes per row: MS-SQL 6.5: 1962 MS-SQL 7.0: 8060
Note that this limitation is pretty easy to work around with foreign key relationships. I have a feeling that MySQL users might percieve this problem to be greater than it actually is because they are used to designing larger 'flat file' databases.
Ironically, DHCP was actually introduced to help the little guy who didn't want the overhead of manually tracking the address of every device on their network.
Some might recall the FUD surrounding TCP/IP on corporate networks, which was pretty much killed by "autoconfiguration" (and Netscape, of course). Protocols like IPX were seen to be easier to manage because they managed themselves. (Ironically, IP6 is moving back to a IPX-like NIC addressing scheme...)
It's too bad that you have to deal with hosting on a dynamic address, but that's something a few extra dollars a month will take care of. DHCP was intended to simplify client management -- complicating server management is a side effect. (The 'little guy' or the 'guy who refuses to pay full freight?)
Microsoft and Intel were betting on a big x86 market. That's why they both cut such sweetheart deals with IBM.
(MS non-exclusively licenced DOS to IBM for a pittance -- $50K IIRC. Why would they give DOS to IBM essentially for free if they didn't think they could sell it someone else?)
So maybe MS didn't forsee that "100% IBM PC AT compatible" would be the standard of the future, but they certainly knew that for the PC industry to grow, it had to grow out of the incompatible mess of the CPM/Apple days.
Actually, I believe that requiring licences to write for your platform is illegal in the US (Atari versus Activision, and Sega versus EA, both long ago).
The reason Sega/Nintendo/Sony get away with it is that it's legal in Japan, and Japan is a huge market. Of course, it would be almost impossible to produce a game for a complex system like a N64 or Playstation 2 without the official development kits, so that plays into it too.
On the other hand, the XBox is commodity hardware, produce primarily for the US (true?), and uses publically documented APIs. It will be interesting to see how Microsoft handles the licencing issue.
The thing has firewire so that it can connect directly to a Sony-brand HDTV set (coming soon).
Sony is slowly pushing iLink across all of its product line as what it wants to be the digital interconnect standard. (A "Sony" plug instead of an "RCA" plug -- Consumers are apparently afraid of a big bunch of interconnect jacks, and they can be replaced by one firewire daisychain, which of course also buys Sony a copy protection system.)
The real problem, though, is client/server based programs... , I can take a GPL X11 program, say Emacs
First of all, I'm not sure if running Emacs over X11 is really a 'client/server' relationship -- X is client/server, but Emacs is oblivious to what display it's running on.
So, this doesn't seem to be too much of a problem. The licence file is clearly called COPYING, and not RUNNING. (And don't give me any nonsense about "RUNNING = COPYING into RAM", because if that's the intent the GPL is essentially a big lie.)
So the database example is a more appropriate confusion -- Can my proprietary VisualBasic front end talk (via ODBC) to the GPL MySQL DBMS? Is RPC OK? What about a simpler protocol like SOAP which is similar to text I/O?
It gets more confusing when you throw additional layers like CORBA in -- What if you had a GPL program which talked to a proprietary CORBA ORB which then talked to a GPL ORB (maybe on the same machine) which then talked to a proprietary back-end. Let's throw Java in there - did it come with your OS, or did you download it? Are you legal? (Better call a lawyer, who will probably tell you to play it safe and don't do that.)
Since KDE and Gnome are implementing COM-like in-process component support, there's going to be some interesting flamewars ahead. Is it "linking" (not exactly)? Is it a seperate "program" (no fork - not exactly either)?
The fallback answer seems to be "original intent" -- Are you using technology just to try to get around the GPL? I'm just not sure if that angle will hold up in court,
I'm not an expert in any of the legal stuff, so maybe I'm off-base. Just trying to point out that we live in an increasing interconnected world, and the software technology to go along with it. Meanwhile we have a licence designed in the old mini-computing days, and seems to have many of the assumptions of those days to go along with it.
Well, Sun used to sell lots of workstations to people that weren't excelusively programmers. If only programmers used Unix, Unix would have died long ago.
(So, I guess Linux is the perfect OS for people who think Netscape-style HTML is anywhere near a replacement for an office suite. Well put.)
Don't get too worried -- most people who advocate Linux on the desktop don't need to use any applications more complex than web/mail/news/sound players, and like to do these (rather generic) tasks in a Unix environment.
Unlike any other OS, Linux's desktop "Killer Apps" seems to be programming tools. Not good if you need to get any work done, but presumably all of those programming tools will eventually produce programs that do something other than assist in writing programs.
Pretty much every review of WordPerfect for Windows since version 6 reads pretty much the same: "Choose WordPerfect if you haven't standard on a word processor - If you've standardized on Word, it's not worth the effort to switch."
Not good when Word has had an 80%+ marketshare since 1993 or so.
The nice thing about Windows Networking is that it has already essentially been banned from the Internet -- most larger ISPs make a point of blocking ports 137-139 (probably missing the new W2K SMP ports), and I would imagine that Universities block it at their border. So, unlike telnet/ftp which has traditionally been open to the entire Internet on campus networks, Windows filesharing is an internal problem.
My question is: Did Microsoft or OEMs ship a version of Windows 9x so that it shares drives by default? (I know NT has it's admin shares.) Would they really be that stupid?
(I've seen users that can't figure out how to print, but yet somehow have filesharing turned on, but everytime I've installed Windows, it seems like you need to take 3 extra steps to get it working.)
Without copyright, GNU could offer a standard contract -- If you want to see my code, you must sign this contract mandating what you do with it and who you show it to.
That would still be legal even if copyrights were not. It would also totally change the distribution mechinism of open source software, (as would it with commercial software).
Actually, that's a bunch of crap. You fight a war for territory or resources because there is a finite amount of those things, so it's worth fighting for.
The desktop computer market, on the other hand, has been doubling in size every few years, and has been doing so for almost ten years. Every day there is a bewildering array of new applications and new markets opening up, and Microsoft can't cover all of them. It's really just a series of lucky historical accidents that allowed Microsoft to capture most of the growth of the last decade.
The key to staying alive or thriving is to capture your share of new growth plus a little more. Apple's been doing this just fine lately, and now they have a healthy 10% of a growing market. They don't need to 'beat' Microsoft, because there is plenty of money in what they've got.
As for Linux, the theory is that if it gets good enough, it can capture some of this new desktop growth. It's all potential on the client-side right now. On the other hand, the server market has been growing faster than even the client market, and Linux is mopping up there.
This "only one winner" crap on Slashdot is juvenilia at it's finest. AMD is doing great -- Intel must be going down (not true). Linux and Sun are selling -- Microsoft must be having a hard time (not true either). All boats are rising in this internet tidal wave.
Cadillacs actually started going down hill in 1973, when they started adding emission controls. In 1974 fuel economy restrictions came in.
Not that emissions/fuel economy is a bad thing, it's just that GM responded by detuning the bloated 472 ci engines they used. As a result reliablity went into the toliet, and you had a 2.5 ton car with only 170 HP.
The early eightes was another disaster as they f-ed up retooling for newer engines and front wheel drive. Cars have been getting slowly better since about 1986
By the time a large corporation pays someone to spec the part, order it, track the orders, report to their manager, take some smoke breaks, fill up their cube with little dolls, and then finally open up a three year old machine and install the part, you've more than spent the replacement value of the machine.
Meanwhile, Dell (or whoever) will installe extra RAM now for free. Of course the RAM part is more expensive now than in three years, but it will still save you money.
You know, in the low-end space where Microsoft has been playing, interoperability has never been that much of a concern for anyone. Apple? Novell? WordPerfect? Unlike in the midrange, there hasn't been a great "open system" tradition for the desktop/file server market, so Microsoft is of course making it up as they go along. The results are the "Microsoft Way" you speak of -- more of a defacto religion than a planned one.
(I'll ignore your "dark ages" stuff which just seems to take away from your point.)
Now they are trying to get into the big market datacenters, of course their 'Program Manager' sales droids are running into actual real world Unix and Mainframe people. These are they kind of folks who understand these sorts of issues and are actually worried about little details like MS Kerberos interoperability. Microsoft is probably finding that they can't dictate solutions, because their customers already have similar solutions in place, and they simply won't buy Microsoft unless MS plays ball. This is going to have to bring about a cultural change within MS.
It's true that Apple doesn't have a 'BIOS', but they do have undocumented chips which you would need to engineer around. There's a reason they need to patch the MacOS everytime a new machine ships.
On top of that, you would need to include a retail copy of MacOS with every machine, even if Apple would sell you a copy and the licence would allow you to install it.
If a Crusoe-powered Mac laptop is true, there's only one rational scenario -- Apple is making it. (However, with the G3 running at 500+Mhz one wonders if it is worth the engineering effort.)
Slashdot Mirror
Perhaps this is more accurate:
90% of unix sysadmin is accidentailly arcane, and nobody has ever bothered to fix it because it has indirectly brought about profitable support contracts and the job security and loyalty of sysadmins (who incidentally tie up their mastery of Unix with their personal self-worth).
By the same argument, Unix permissions are so complicated nobody (to a first approximation) uses them well either. Although the front-emd of Unix-style might be more streamline, having to create an "arbitrary" number of directories and groups doesn't sound less complex to me.
Well, there's a difference between "I can get by with what I know just fine, thanks" and "Most of the complicated queries you might do are better expressed in a proper programming language" ...
The problem is not that their not straight ASCII -- they are not straight ISO 8859-1. Windows uses several reserved characters for things like curly quotes, and these characters are not implemented in Unix character sets.
I would imagine if you had a MS True Type font, it would look OK. (Or, Slashdot could shut everyone up by implementing a curly quote to HTML & code replacer in it's submit processor.)
Lay off the coffee, spaz. For the record I have no hatred, nor am I part of the linux community.
Since you seem to know so much about the situation, how about letting us know what Troll's profits are, before and after KDE got popular. (I brought up ActiveX component companies because there's several hundred of them, which must indicate that it's a slightly profitable business at best. See http://www.componentshopper.com/)
Troll Tech was merely business savvy and opportunistic.
Troll Tech wasn't savvy at all -- they were just lucky.
They were in the business of selling a GUI toolkit for client Unix systems -- a market that is pretty much dwindling to nothing, and what's left is heavily standardized on Motif.
Doesn't sound like the wisest market to be in. There's probably a half-dozen companies that make VisualBasic widgets that are larger and more profitable than TrollTech.
The only thing that saved them is that KDE chose to use their freeware stuff, and KDE turned out to be such a polished product that Linux users were willing to hold their nose over the licencing issue and develop apps for Qt. If KDE didn't exist, or was buggy/defective, TrollTech would still be two guys in a basement somewhere.
SQL (as a language, not a feature set) is designed for set processing. A particular SQL implementation is heavily optimised for the data store implementation used in it's RDBMS.
I'd think you'd be crazy not to write as much of your applicaiton in SQL as possible, even though it's ugly lookin' and it will never be more than a portion of your code base. Ignoring a powerful and efficent tool like that seems foolhardy. (Not that I haven't used Postgresql or mysql, so perhaps their dialects are too limited to be as useful.)
Trying to write the loops to manage set processing in a "proper" language will almost always be more inefficent and error prone. Not to mention the insanity of running that code interpreted in Perl/PHP/ASP/Whatever on your webserver.
The thing is, if you are doing a reasonable upgrade (like once every two years to get a 100% clock speed boost), it makes more sense to buy a new system. Realistically, you probably don't want the video card and hard drive from your PII-333 in your brand new PIII-700 system.
But, certain people out of concern for penis envy or gaming performance or whatever upgrade far more often. These guys might have a GeForce2 and a DMA66 drive in their Celeron 300A @ 450, and when they upgrade to an AMD 800, it makes sense to move the parts over. At that point the old computer is pretty much worthless with no drive and no video, and it makes sense to recycle the case.
Maximum bytes per row:
MS-SQL 6.5: 1962
MS-SQL 7.0: 8060
Note that this limitation is pretty easy to work around with foreign key relationships. I have a feeling that MySQL users might percieve this problem to be greater than it actually is because they are used to designing larger 'flat file' databases.
Ironically, DHCP was actually introduced to help the little guy who didn't want the overhead of manually tracking the address of every device on their network.
Some might recall the FUD surrounding TCP/IP on corporate networks, which was pretty much killed by "autoconfiguration" (and Netscape, of course). Protocols like IPX were seen to be easier to manage because they managed themselves. (Ironically, IP6 is moving back to a IPX-like NIC addressing scheme...)
It's too bad that you have to deal with hosting on a dynamic address, but that's something a few extra dollars a month will take care of. DHCP was intended to simplify client management -- complicating server management is a side effect. (The 'little guy' or the 'guy who refuses to pay full freight?)
Microsoft and Intel were betting on a big x86 market. That's why they both cut such sweetheart deals with IBM.
(MS non-exclusively licenced DOS to IBM for a pittance -- $50K IIRC. Why would they give DOS to IBM essentially for free if they didn't think they could sell it someone else?)
So maybe MS didn't forsee that "100% IBM PC AT compatible" would be the standard of the future, but they certainly knew that for the PC industry to grow, it had to grow out of the incompatible mess of the CPM/Apple days.
Actually, I believe that requiring licences to write for your platform is illegal in the US (Atari versus Activision, and Sega versus EA, both long ago).
The reason Sega/Nintendo/Sony get away with it is that it's legal in Japan, and Japan is a huge market. Of course, it would be almost impossible to produce a game for a complex system like a N64 or Playstation 2 without the official development kits, so that plays into it too.
On the other hand, the XBox is commodity hardware, produce primarily for the US (true?), and uses publically documented APIs. It will be interesting to see how Microsoft handles the licencing issue.
The thing has firewire so that it can connect directly to a Sony-brand HDTV set (coming soon).
Sony is slowly pushing iLink across all of its product line as what it wants to be the digital interconnect standard. (A "Sony" plug instead of an "RCA" plug -- Consumers are apparently afraid of a big bunch of interconnect jacks, and they can be replaced by one firewire daisychain, which of course also buys Sony a copy protection system.)
The real problem, though, is client/server based programs ... , I can take a GPL X11 program, say Emacs
First of all, I'm not sure if running Emacs over X11 is really a 'client/server' relationship -- X is client/server, but Emacs is oblivious to what display it's running on.
So, this doesn't seem to be too much of a problem. The licence file is clearly called COPYING, and not RUNNING. (And don't give me any nonsense about "RUNNING = COPYING into RAM", because if that's the intent the GPL is essentially a big lie.)
So the database example is a more appropriate confusion -- Can my proprietary VisualBasic front end talk (via ODBC) to the GPL MySQL DBMS? Is RPC OK? What about a simpler protocol like SOAP which is similar to text I/O?
It gets more confusing when you throw additional layers like CORBA in -- What if you had a GPL program which talked to a proprietary CORBA ORB which then talked to a GPL ORB (maybe on the same machine) which then talked to a proprietary back-end. Let's throw Java in there - did it come with your OS, or did you download it? Are you legal? (Better call a lawyer, who will probably tell you to play it safe and don't do that.)
Since KDE and Gnome are implementing COM-like in-process component support, there's going to be some interesting flamewars ahead. Is it "linking" (not exactly)? Is it a seperate "program" (no fork - not exactly either)?
The fallback answer seems to be "original intent" -- Are you using technology just to try to get around the GPL? I'm just not sure if that angle will hold up in court,
I'm not an expert in any of the legal stuff, so maybe I'm off-base. Just trying to point out that we live in an increasing interconnected world, and the software technology to go along with it. Meanwhile we have a licence designed in the old mini-computing days, and seems to have many of the assumptions of those days to go along with it.
Well, Sun used to sell lots of workstations to people that weren't excelusively programmers. If only programmers used Unix, Unix would have died long ago.
(So, I guess Linux is the perfect OS for people who think Netscape-style HTML is anywhere near a replacement for an office suite. Well put.)
From technet -- Direct hosted "NetBIOS-less" SMB traffic uses port 445 (TCP and UPD).
Don't get too worried -- most people who advocate Linux on the desktop don't need to use any applications more complex than web/mail/news/sound players, and like to do these (rather generic) tasks in a Unix environment.
Unlike any other OS, Linux's desktop "Killer Apps" seems to be programming tools. Not good if you need to get any work done, but presumably all of those programming tools will eventually produce programs that do something other than assist in writing programs.
Pretty much every review of WordPerfect for Windows since version 6 reads pretty much the same: "Choose WordPerfect if you haven't standard on a word processor - If you've standardized on Word, it's not worth the effort to switch."
Not good when Word has had an 80%+ marketshare since 1993 or so.
The nice thing about Windows Networking is that it has already essentially been banned from the Internet -- most larger ISPs make a point of blocking ports 137-139 (probably missing the new W2K SMP ports), and I would imagine that Universities block it at their border. So, unlike telnet/ftp which has traditionally been open to the entire Internet on campus networks, Windows filesharing is an internal problem.
My question is: Did Microsoft or OEMs ship a version of Windows 9x so that it shares drives by default? (I know NT has it's admin shares.) Would they really be that stupid?
(I've seen users that can't figure out how to print, but yet somehow have filesharing turned on, but everytime I've installed Windows, it seems like you need to take 3 extra steps to get it working.)
Without copyright, GNU could offer a standard contract -- If you want to see my code, you must sign this contract mandating what you do with it and who you show it to.
That would still be legal even if copyrights were not. It would also totally change the distribution mechinism of open source software, (as would it with commercial software).
Actually, that's a bunch of crap. You fight a war for territory or resources because there is a finite amount of those things, so it's worth fighting for.
The desktop computer market, on the other hand, has been doubling in size every few years, and has been doing so for almost ten years. Every day there is a bewildering array of new applications and new markets opening up, and Microsoft can't cover all of them. It's really just a series of lucky historical accidents that allowed Microsoft to capture most of the growth of the last decade.
The key to staying alive or thriving is to capture your share of new growth plus a little more. Apple's been doing this just fine lately, and now they have a healthy 10% of a growing market. They don't need to 'beat' Microsoft, because there is plenty of money in what they've got.
As for Linux, the theory is that if it gets good enough, it can capture some of this new desktop growth. It's all potential on the client-side right now. On the other hand, the server market has been growing faster than even the client market, and Linux is mopping up there.
This "only one winner" crap on Slashdot is juvenilia at it's finest. AMD is doing great -- Intel must be going down (not true). Linux and Sun are selling -- Microsoft must be having a hard time (not true either). All boats are rising in this internet tidal wave.
Cadillacs actually started going down hill in 1973, when they started adding emission controls. In 1974 fuel economy restrictions came in.
Not that emissions/fuel economy is a bad thing, it's just that GM responded by detuning the bloated 472 ci engines they used. As a result reliablity went into the toliet, and you had a 2.5 ton car with only 170 HP.
The early eightes was another disaster as they f-ed up retooling for newer engines and front wheel drive. Cars have been getting slowly better since about 1986
By the time a large corporation pays someone to spec the part, order it, track the orders, report to their manager, take some smoke breaks, fill up their cube with little dolls, and then finally open up a three year old machine and install the part, you've more than spent the replacement value of the machine.
Meanwhile, Dell (or whoever) will installe extra RAM now for free. Of course the RAM part is more expensive now than in three years, but it will still save you money.
You know, in the low-end space where Microsoft has been playing, interoperability has never been that much of a concern for anyone. Apple? Novell? WordPerfect? Unlike in the midrange, there hasn't been a great "open system" tradition for the desktop/file server market, so Microsoft is of course making it up as they go along. The results are the "Microsoft Way" you speak of -- more of a defacto religion than a planned one.
(I'll ignore your "dark ages" stuff which just seems to take away from your point.)
Now they are trying to get into the big market datacenters, of course their 'Program Manager' sales droids are running into actual real world Unix and Mainframe people. These are they kind of folks who understand these sorts of issues and are actually worried about little details like MS Kerberos interoperability. Microsoft is probably finding that they can't dictate solutions, because their customers already have similar solutions in place, and they simply won't buy Microsoft unless MS plays ball. This is going to have to bring about a cultural change within MS.
It's true that Apple doesn't have a 'BIOS', but they do have undocumented chips which you would need to engineer around. There's a reason they need to patch the MacOS everytime a new machine ships.
On top of that, you would need to include a retail copy of MacOS with every machine, even if Apple would sell you a copy and the licence would allow you to install it.
If a Crusoe-powered Mac laptop is true, there's only one rational scenario -- Apple is making it. (However, with the G3 running at 500+Mhz one wonders if it is worth the engineering effort.)