Absolutely. MySQL's ability to handle VSAM and ISAM files on mainfames and seamless integrate it with features like PSF report generation will make this transition easy for DB2 customers.:-)
No, MySQL is not a risk to DB2. Totally different customer bases with totally different needs.
Symphony is based on the OO engine. I'd assume the direction will be to move OO to context sensitive menus and tabs and unify the product lines.
DB2 and MySQL don't really compete, they serve mostly entirely different functions and customer bases. So probably just leave them alone and have them develop independently of one another. Besides DB2 is merging with Informix (another database they bought).
As for Solaris vs. AIX. That one is tough. I'd say most likely pull Solaris technologies into Linux, AIX and maybe even i/OS (OS/400) kind of cannibalize, but who knows?
I'd say on the app server support both and merge technologies slowly.
Very well put. OSX gives you convenient access to Unix applications and offers Unix interfaces to itself. But in no way is it a traditional Unix.
As for X it is kind of the same thing. OSX users have very high GUI expectations. They like Aqua apps and they like having access to X apps. That's different than liking X apps.
I'm not sure I'm decrying I was in the WYSIWYM camp. With OO I'd say we've lost even on Linux.
Heck I've been nagging my wife for 15 years that the enormous variable formatting problems she faces would go away almost instantly if she switched to WYSIWYM with no apparent effect.
Absolutely no one likes maintaining different code bases. No one likes smelly garbage. Saying something is bad is not a criticism. Saying:
X is bad X exists because of A,B and C There is a solution to the problems of X, Y Y introduces problems and costs D,E,F Here is why D,E,F is not as bad as A,B,C means you have thought about the issue.
If you don't understand why Window, MacOSX and Linux use different driver models then you don't understand the problem nor the reason for the problem. Just saying you wish the problem didn't exist isn't really criticism.
Oh yeah I'm talking about changes that took a decade or more. And yeah it was bad the text editor crowd it was worse with the WYSIWYM crowd because they had even better reasons why word processors (OO really didn't exist, but Star Office did) were inferior to their document systems.
It isn't just "its good enough for me". Often the Linux people realize this is a real issue. With limited resources only a small fraction of the issues can be addressed. Addressing expensive ones is going to be rare and it requires time to be convincing.
Which could be accomplished by expanding Cygwin's compatibility too...
Cygwin is so-so. Used to be better actually. For that matter Windows Services for Unix by Microsoft is so-so. Neither is remotely close though to really getting most Unix apps to run.
2) They want an easier to maintain system when things go wrong
Easier in what way? A copy of Norton Ghost (or another backup solution) and a monthly/weekly backup will make my system "easy" to fix in the event of catastrophic h/d failure or data corruption problem... anything else is just "go replace the failed component" or "well shit, time to reinstall the OS" anyways.
Not my experience at all. I've made use of/dev/proc many times to diagnose and/or bypass hardware problems. It is fantastic to be able to pass messages directly to hardware, for example force a bus to init a drive it doesn't see.
As far as software and OS, reinstall is usually out of the question. A desktop + all the apps takes about 100 hours to configure the right way.
4)They want scripting heavily integrated with apps and the system
Again, what sort of scripting? "Integrated scripting" isn't half as important as the purpose you want each individual script to accomplish.
Integrated is not with each other but with apps and the OS. Frequently scripts need to act between multiple apps. Applescript allows you to do this on Mac, but it is limited. There really is no substitution for Unix apps having good command line controls and universal scripting languages.
5)They want to distribute their desktop i.e. network transparency
I think we've gone beyond the bounds of the average home user, or the average corporate network user, or even the average developer at this point.
Most companies that switched to Linux back in 2000 did so for this reason. End users may not know the term "Network transparency" but they love the ability to pop on any machine in the corporation and have it act the same. They love being able to pop windows to other users "Jay you need to fill these 3 lines out then pass it back to me".
Absolutely that requires their IT guys to set it up. In terms of home users, network transparency is essentially what all the TV interface stuff is about. Also VPN....
OK Windows XP and Vista (32) both. Vista 64 probably still edges ahead but that is getting very close.
As for evidence, look at any list of random devices. Take scanners for example where support on Linux is very thin.
Re:Let me be the first critic
on
Linux Needs Critics
·
· Score: 2, Informative
1) They want to use Unix apps 2) They want an easier to maintain system when things go wrong 3) They want more control over apps and the system at all levels 4) They want scripting heavily integrated with apps and the system 5) They want to distribute their desktop i.e. network transparency
This is a pretty good example of why criticism isn't taken seriously. Do you actually know anything about the driver models on Windows, OSX and Linux and how they contrast? Why they are different? What stops them from using each other's code?
Linux has non geek users. Generally they don't have a problem. Where Linux has problems is with semi-geeks who feel comfortable with Windows but don't really know what they are doing.
In the mid 90s there was a widespread belief among Linux people that GUIs were a bad thing. A better approach to computing was simple window management + graphical apps. The KDE movement and later Gnome changed that attitude.
In the 80s there was a belief that compiler breath and standardization was more important than compiled code speed. GCC was not going to have all sorts of thousands of cases to squeeze an extra few percentage points out, like the commercial compilers. GCC has done a complete 180 on that issue, especially with PowerPC and 8086 architectures.
Up until about 2000 the majority of Linux users thought that WYSIWYG was a terrible way to handle document authoring and the primary document authoring systems should be WYSIWYM. Again a 180.
People do get listened to, and change does happen.
Free makes sense for OEMs. $20 x 100k+ units makes it worth dealing with lots of hassles. Server software can be expensive. $3000 x 20 units makes hassles worth you time. Saving a hundred on Windows doesn't make it worth it for an individual, never has.
On the other hand, compared to the other Unixes of the time Linux was worth the hassles. A SCO (back in the 1990s) used to run over a grand. You could get a Linux system up for 2-3k which was comparable to a $7k low end Sun workstation.
Windows desktop is cheap, free as in beer doesn't make much sense for windows. OTOH many of Linux other competitors were not cheap and free as in beer made a big difference. Where Linux has an edge over Windows is free as in freedom, or free as in beer applications.
The Linux community is fine with well thought out criticism. For example we had a criticism about a slow down the person tracked it down to a particular kernel module and then provided a comparative benchmark. People started working on it and it will be fixed in a month.
He didn't just say "Linux has gotten slower". Or the Linux developers should wave the magic driver wand and get more drivers.
RedHat gave up on the desktop when they dropped the RedHat line for Fedora. Arguably they gave up even a few years before that when they focused their energy on servers, say around 1996.
But they also want the Unix/model to be adopted. I've had this discussion a lot on package management. Linux users like the distribution model: software vendor -> distribution -> end user
rather than the windows model: distribution -> software vendor -> end user
So it is not just wider adoption but they want systematic change.
When Linux gets anywhere close to critical mass people will buy systems with Linux preinstalled or Linux compatibility. They will get hardware configured by the OEM, who will work with distribution and the hardware manufacturer.
What happens with Linux along with every OS is that end users are forced to conduct this discussion themselves.
Nonsense. People understand they have to replace their cell phone addons when they replace cell phones. If people are motivated to switch the few hundred dollars in hardware costs are doable. If they are starting to consider switching seriously, they will start buying hardware which is compatible. They certainly don't buy Sun only hardware.
The OS with the best hardware support far and away is Windows. Linux is 2nd but they ain't neck and neck. BSDs are further back at 3rd....
From any perspective other than Windows, Linux hardware support is excellent. The "I should be able to buy any piece of random windows crap and have it work" is an unrealistic expectation. Microsoft spends billions to get that to happen for Windows.
What Linux has managed to achieve is
1) Most hardware works 2) The vast majority of hardware kinda works 3) Older less supported hardware technologies also work well.
Sorry but if you wan:
-- The vast majority of hardware works very well.
There really is only one OS. Which is fine, Windows does have genuine advantages and that is one of them. It was OS/2 not Linux that claimed to be a "better Windows than windows".
There are multiple versions of both Firefox 2 and 3 which run under eCS, so you can certainly test how well it handles such things.
I'm not sure that is fair to OS/2. Firefox is not taking advantage of OS/2s built in advantages and those advantages haven't been expanded upon in many years. That's why I thought a "what if" version makes more sense. Unless you are saying that OS/2 with a decade of very light development still holds up well, which would be great but I find it unlikely.
What I think was cool, is something that Apple is just starting to move back into supporting the notion of forking code off to different types of chips. In the 1990s there were the i486/i980 motherboards, SCO was actually the leader in supporting that multi CPU stuff. Basically the i486 handled most code while the i980 acted like a very high power vector math chip (a lot of what a 3D GPU does today). This chip split was far less than what the RS/6000s did (OTOH the 486/990 were much cheaper). I know OS/2 was moving towards supporting this not sure if it got there, but IMHO supporting different kinds of CPUs is a big step up from dual.
Anyway I don't disagree that OS/2 2.1 ~ Windows 2000 in terms of overall quality.
As for which API I have no idea. I was never an OS/2 developer just a 1.3-3.0 end user.
I agree OS/2 was preemptive. It wasn't cooperative in the Windows 286, GEOS sense. What was different was the fact that apps organized themselves into threads and processes rather than just threads:
Slashdot Mirror
Hold on a second here. You are very very different concepts.
1) Is OSX traditional?
2) Is traditional defined by a specification?
3) Is traditional good and non-traditional bad?
I agreed the answer to #2 was no. You seem to be arguing that since there isn't a spec:
not #2 implies not #3, which I think is a stretch.
But even if I accepted it, that still has nothing to do with #1.
There is no specification. Its a feeling thing. Fuzzy (GP) did a pretty job describing the differences.
Absolutely. MySQL's ability to handle VSAM and ISAM files on mainfames and seamless integrate it with features like PSF report generation will make this transition easy for DB2 customers. :-)
No, MySQL is not a risk to DB2. Totally different customer bases with totally different needs.
Symphony is based on the OO engine. I'd assume the direction will be to move OO to context sensitive menus and tabs and unify the product lines.
DB2 and MySQL don't really compete, they serve mostly entirely different functions and customer bases. So probably just leave them alone and have them develop independently of one another. Besides DB2 is merging with Informix (another database they bought).
As for Solaris vs. AIX. That one is tough. I'd say most likely pull Solaris technologies into Linux, AIX and maybe even i/OS (OS/400) kind of cannibalize, but who knows?
I'd say on the app server support both and merge technologies slowly.
Very well put. OSX gives you convenient access to Unix applications and offers Unix interfaces to itself. But in no way is it a traditional Unix.
As for X it is kind of the same thing. OSX users have very high GUI expectations. They like Aqua apps and they like having access to X apps. That's different than liking X apps.
I'm not sure I'm decrying I was in the WYSIWYM camp. With OO I'd say we've lost even on Linux.
Heck I've been nagging my wife for 15 years that the enormous variable formatting problems she faces would go away almost instantly if she switched to WYSIWYM with no apparent effect.
Absolutely no one likes maintaining different code bases. No one likes smelly garbage. Saying something is bad is not a criticism. Saying:
X is bad
X exists because of A,B and C
There is a solution to the problems of X, Y
Y introduces problems and costs D,E,F
Here is why D,E,F is not as bad as A,B,C means you have thought about the issue.
If you don't understand why Window, MacOSX and Linux use different driver models then you don't understand the problem nor the reason for the problem. Just saying you wish the problem didn't exist isn't really criticism.
Oh yeah I'm talking about changes that took a decade or more. And yeah it was bad the text editor crowd it was worse with the WYSIWYM crowd because they had even better reasons why word processors (OO really didn't exist, but Star Office did) were inferior to their document systems.
It isn't just "its good enough for me". Often the Linux people realize this is a real issue. With limited resources only a small fraction of the issues can be addressed. Addressing expensive ones is going to be rare and it requires time to be convincing.
1) They want to use Unix apps
Which could be accomplished by expanding Cygwin's compatibility too...
Cygwin is so-so. Used to be better actually. For that matter Windows Services for Unix by Microsoft is so-so. Neither is remotely close though to really getting most Unix apps to run.
2) They want an easier to maintain system when things go wrong
Easier in what way? A copy of Norton Ghost (or another backup solution) and a monthly/weekly backup will make my system "easy" to fix in the event of catastrophic h/d failure or data corruption problem... anything else is just "go replace the failed component" or "well shit, time to reinstall the OS" anyways.
Not my experience at all. I've made use of /dev /proc many times to diagnose and/or bypass hardware problems. It is fantastic to be able to pass messages directly to hardware, for example force a bus to init a drive it doesn't see.
As far as software and OS, reinstall is usually out of the question. A desktop + all the apps takes about 100 hours to configure the right way.
4)They want scripting heavily integrated with apps and the system
Again, what sort of scripting? "Integrated scripting" isn't half as important as the purpose you want each individual script to accomplish.
Integrated is not with each other but with apps and the OS. Frequently scripts need to act between multiple apps. Applescript allows you to do this on Mac, but it is limited. There really is no substitution for Unix apps having good command line controls and universal scripting languages.
5)They want to distribute their desktop i.e. network transparency
I think we've gone beyond the bounds of the average home user, or the average corporate network user, or even the average developer at this point.
Most companies that switched to Linux back in 2000 did so for this reason. End users may not know the term "Network transparency" but they love the ability to pop on any machine in the corporation and have it act the same. They love being able to pop windows to other users "Jay you need to fill these 3 lines out then pass it back to me".
Absolutely that requires their IT guys to set it up. In terms of home users, network transparency is essentially what all the TV interface stuff is about. Also VPN....
OK Windows XP and Vista (32) both. Vista 64 probably still edges ahead but that is getting very close.
As for evidence, look at any list of random devices. Take scanners for example where support on Linux is very thin.
1) They want to use Unix apps
2) They want an easier to maintain system when things go wrong
3) They want more control over apps and the system at all levels
4) They want scripting heavily integrated with apps and the system
5) They want to distribute their desktop i.e. network transparency
etc...
This is a pretty good example of why criticism isn't taken seriously. Do you actually know anything about the driver models on Windows, OSX and Linux and how they contrast? Why they are different? What stops them from using each other's code?
They aren't converging they are diverging.
Linux has non geek users. Generally they don't have a problem. Where Linux has problems is with semi-geeks who feel comfortable with Windows but don't really know what they are doing.
I don't agree let me give several examples.
In the mid 90s there was a widespread belief among Linux people that GUIs were a bad thing. A better approach to computing was simple window management + graphical apps. The KDE movement and later Gnome changed that attitude.
In the 80s there was a belief that compiler breath and standardization was more important than compiled code speed. GCC was not going to have all sorts of thousands of cases to squeeze an extra few percentage points out, like the commercial compilers. GCC has done a complete 180 on that issue, especially with PowerPC and 8086 architectures.
Up until about 2000 the majority of Linux users thought that WYSIWYG was a terrible way to handle document authoring and the primary document authoring systems should be WYSIWYM. Again a 180.
People do get listened to, and change does happen.
Free makes sense for OEMs. $20 x 100k+ units makes it worth dealing with lots of hassles. Server software can be expensive. $3000 x 20 units makes hassles worth you time. Saving a hundred on Windows doesn't make it worth it for an individual, never has.
On the other hand, compared to the other Unixes of the time Linux was worth the hassles. A SCO (back in the 1990s) used to run over a grand. You could get a Linux system up for 2-3k which was comparable to a $7k low end Sun workstation.
Windows desktop is cheap, free as in beer doesn't make much sense for windows. OTOH many of Linux other competitors were not cheap and free as in beer made a big difference. Where Linux has an edge over Windows is free as in freedom, or free as in beer applications.
The Linux community is fine with well thought out criticism. For example we had a criticism about a slow down the person tracked it down to a particular kernel module and then provided a comparative benchmark. People started working on it and it will be fixed in a month.
He didn't just say "Linux has gotten slower". Or the Linux developers should wave the magic driver wand and get more drivers.
RedHat gave up on the desktop when they dropped the RedHat line for Fedora. Arguably they gave up even a few years before that when they focused their energy on servers, say around 1996.
I think most Linux people want wider adoption.
But they also want the Unix/model to be adopted. I've had this discussion a lot on package management. Linux users like the distribution model:
software vendor -> distribution -> end user
rather than the windows model:
distribution -> software vendor -> end user
So it is not just wider adoption but they want systematic change.
When Linux gets anywhere close to critical mass people will buy systems with Linux preinstalled or Linux compatibility. They will get hardware configured by the OEM, who will work with distribution and the hardware manufacturer.
What happens with Linux along with every OS is that end users are forced to conduct this discussion themselves.
Nonsense. People understand they have to replace their cell phone addons when they replace cell phones. If people are motivated to switch the few hundred dollars in hardware costs are doable. If they are starting to consider switching seriously, they will start buying hardware which is compatible. They certainly don't buy Sun only hardware.
Right now they aren't motivated switchers.
The OS with the best hardware support far and away is Windows. Linux is 2nd but they ain't neck and neck. BSDs are further back at 3rd....
From any perspective other than Windows, Linux hardware support is excellent. The "I should be able to buy any piece of random windows crap and have it work" is an unrealistic expectation. Microsoft spends billions to get that to happen for Windows.
What Linux has managed to achieve is
1) Most hardware works
2) The vast majority of hardware kinda works
3) Older less supported hardware technologies also work well.
Sorry but if you wan:
-- The vast majority of hardware works very well.
There really is only one OS. Which is fine, Windows does have genuine advantages and that is one of them. It was OS/2 not Linux that claimed to be a "better Windows than windows".
Well its been a while since this was posted, Every OS sucks.
There are multiple versions of both Firefox 2 and 3 which run under eCS, so you can certainly test how well it handles such things.
I'm not sure that is fair to OS/2. Firefox is not taking advantage of OS/2s built in advantages and those advantages haven't been expanded upon in many years. That's why I thought a "what if" version makes more sense. Unless you are saying that OS/2 with a decade of very light development still holds up well, which would be great but I find it unlikely.
What I think was cool, is something that Apple is just starting to move back into supporting the notion of forking code off to different types of chips. In the 1990s there were the i486/i980 motherboards, SCO was actually the leader in supporting that multi CPU stuff. Basically the i486 handled most code while the i980 acted like a very high power vector math chip (a lot of what a 3D GPU does today). This chip split was far less than what the RS/6000s did (OTOH the 486/990 were much cheaper). I know OS/2 was moving towards supporting this not sure if it got there, but IMHO supporting different kinds of CPUs is a big step up from dual.
Anyway I don't disagree that OS/2 2.1 ~ Windows 2000 in terms of overall quality.
As for which API I have no idea. I was never an OS/2 developer just a 1.3-3.0 end user.
I agree OS/2 was preemptive. It wasn't cooperative in the Windows 286, GEOS sense. What was different was the fact that apps organized themselves into threads and processes rather than just threads:
http://www.ibm.com/developerworks/linux/library/l-osmig1.html
I'd like the core wars to keep going with more unboard cache. I want 64 cores.