What advancements is Linux developing?

[gflores]

I guess I'm a big noob when it comes to Linux, but I'll be installing Mandrake once school starts to become more familiar with it. Anyway, the reason I ask, is that Microsoft and Apple appear to be creating innovating features, like file storage and searching, in their upcoming operating systems, but I never really hear anything about Linux. So... please enlighten me.

 

[drag]

Innovations in Linux is pretty good. It's developement seems faster then either OS X or Windows.

However what you see coming from MS in terms of Longhorn and such is still mostly fantasies, as time goes on you'll see the promises begin to mellow out and features are going to be dropped. Like lots of the features of the WinFS filesystem have been dropped.

OS X is a bit more realistic.

However in both cases your going to see more hype then anything else. Trying to get people excited and so forth.

Most of the advancements that are going on can realy only be appreicieted by a geek. Mostly stuff that is to boring for most folks.

You have stuff like gstreamer that provides a way to pipe multimedia from application to application. Much like you can pipe stdout into stdin and create complex scripts and such. Stuff like the low-latency jack plugins, LADSPA, and the adoption of ALSA sound drivers allow Linux to be usefull for high quality music creation. Now Linux has a sound system that is competative with it's Windows and Apple counterparts and the free software developements like ardour.

We've had stuff like XFS filing system, which allows roboust handling of large files. It's more stable, safe, and faster then NTFS. Stuff like that has been around for a while. ReiserFS has been developing new technologies and whatnot that will probably seem similar to lots of things that WinFS is promising.

Linux has been 64bit for ages and ages, which is something that MS still can't seem to do. They had stable versions of AMD64 Linux running and being tested even before AMD64 was released to the open market. WinXP-64 is still pretty much non-existant.

Linux is almost infinately scalable.
You have version running on the smallest memory footprints imaginable. Running in many sohpisticated embedded applications and stuff like that.

Yet on the flip side Linux runs the fastest and most powerfull computers in existance. It is used heavily in research such as protein combinations and gene research. Drug research, nuclear simulation testing, astronomy. Anything that requires massive number crunching abilities. It has removed much of the need for the old-school type super computers and made doing similar stuff a fraction of the cost (although supercomputers are still needed for specific tasks definately).

Out of the top 20 most powerfull computers in the world, as of the last top500 poll (the 24th)
#2, #5, #7, #9, #10, #11, #12, #15, #16, #17, #19 ranked computers run Linux.

And every month new developements push Linux to better perform on all sorts of different hardware platforms.

Then you have other things like Mono, which is Linux's .NET implimentation. Which is progressing at a very fast rate. It's conceviable that a Linux OS will provide a better .NET platform then even Windows. (or at least until longhorn comes out)

Of course I could go on for hours. But normally this sort of stuff isn't going to make the cover of PC Magazine or anything like that.

 

[RyanLM]

It's conceviable that a Linux OS will provide a better .NET platform then even Windows. (or at least until longhorn comes out)

Umm, wouldnt they have to actually get it all working first? While many of your comments fair, this one was just silly. There is no reason to think that .Net would run better on linux, especially when you consider the GUI side of it.

 

[groovin]

drag has some great points. most of the advancements in the works for linux wont get the average persons blood going... but they will add up to things the end user can appreciate... a snappier GUI, better sound support, etc.

the things i am excited about are more developments in SMP , clustering, and reiserfs4.

 

[Nothinman]

Umm, wouldnt they have to actually get it all working first? While many of your comments fair, this one was just silly. There is no reason to think that .Net would run better on linux, especially when you consider the GUI side of it.

The only problem I see with the GUI side of .NET is that MS didn't make Windows.Forms part of the standard they submitted which means .NET will have major problems being cross-platform for anything that requires a GUI. I believe the Mono guys are implementing Windows.Forms but we'll see how well it actually works.

But in general I agree with drag unless MS does a bunch of other bad things with .NET that make it a PITA to make real cross-platform apps. If Java and the NetWare admin tools that were done in java any indication, as long as all the libraries are there, apps won't even have to be written with Linux in mind to run better than they do on Windows. Some of the Java NetWare tools needed a little 'tweaking' to run but once they did they were noticably faster on Linux than on Windows.

 

[RyanLM]

When looking at things like Java and .Net it is all about the implementation on a given platform. Bringing something from one os to another that "isnt" supposed to be OS dependent shouldnt get a speed boost, at least not one that is "noticable", uless one particular implementation is better than the other.

In the case of .Net, I really doubt Mono implmentation of the runtime will beat out MS's version.

The comes of course the GUI side of things, which is alway going to be a kludge off windows. Mono is also ditching its Wine approach (so I have heard) because of problems, which isnt too suprising.

 

[n0cmonkey]

Originally posted by: drag
We've had stuff like XFS filing system, which allows roboust handling of large files.

Is this really worth mentioning when the OP used the word "innovation?"

ReiserFS has been developing new technologies and whatnot that will probably seem similar to lots of things that WinFS is promising.

ReiserFS is closer to innovation though.

Linux has been 64bit for ages and ages, which is something that MS still can't seem to do.

Still not innovative, and I believe there has been a 64bit Windows out there for a while.

They had stable versions of AMD64 Linux running and being tested even before AMD64 was released to the open market.

How can they claim stability without the ability to test it?

Linux is almost infinately scalable.
You have version running on the smallest memory footprints imaginable. Running in many sohpisticated embedded applications and stuff like that.

There is an embedded Windows too, and from what I've heard it ain't bad.


Of course, I'd love to see some innovation out of Redmond or Cupertino. I don't think there has been a whole lot lately. Most of these wizbang features came from towns other than these two.

 

[Nothinman]

When looking at things like Java and .Net it is all about the implementation on a given platform. Bringing something from one os to another that "isnt" supposed to be OS dependent shouldnt get a speed boost, at least not one that is "noticable", uless one particular implementation is better than the other.

Well then, considering that in my Java example the JREs used were both from Sun so they should be roughly equal Linux must have an advantage somewhere else.

In the case of .Net, I really doubt Mono implmentation of the runtime will beat out MS's version.

Depends on what the criteria are, I know of people doing development in C# in 100% Mono right now.

The comes of course the GUI side of things, which is alway going to be a kludge off windows. Mono is also ditching its Wine approach (so I have heard) because of problems, which isnt too suprising.

Why would it be a kludge on non-Windows platforms? All Windows.Forms is, is an API that interacts with the windowing system, the fact that MS' implementation is done in MFC (or whatever they used) and the Mono one is done in GTK is irrelevant. And since GTK runs on Windows it wouldn't be inconceivable that we could have a Windows.Forms implementation on Windows that uses GTK.

How can they claim stability without the ability to test it?

Because it was running on the emulators AMD was using before there was real silicon produced.

 

[n0cmonkey]

Originally posted by: Nothinman

How can they claim stability without the ability to test it?

Because it was running on the emulators AMD was using before there was real silicon produced.

Ahh, it was one of those "sure it works! Test? Yeah, we've uh tested it." kind of things instead of a legitimate claim.

 

[drag]

Originally posted by: n0cmonkey

Originally posted by: drag
We've had stuff like XFS filing system, which allows roboust handling of large files.

Is this really worth mentioning when the OP used the word "innovation?"

It is a innovation, it may be old news. However It's certainly better then any other filesystem out there that I am aware of that gets used by Windows vs OS X vs Linux.

ReiserFS has been developing new technologies and whatnot that will probably seem similar to lots of things that WinFS is promising.

ReiserFS is closer to innovation though.

ReiserFS IS innovation. They are always working on something new, and is about the closest you can get to cutting edge in a mostly usable form. What they have in the kernel right now may not be ready for full time production, but it does incorporate very advanced ideas. Now weither or not they are fully correct ideas, they are certainly innovative.


Now these are just a couple of examples of what people have and are working on. I have others if you want, no were near scraping the bottom of the barrel.

Take for instance Redhat's Global filing system., or the work they've done to get a robust LVM implimentation.

How about Google's Filing system? The "other" GFS.

Linux has been 64bit for ages and ages, which is something that MS still can't seem to do.

Still not innovative, and I believe there has been a 64bit Windows out there for a while.

There was a beta version of W2k that was somewhat 64bit compatable, never made it to market and they scrapped it shortly after the beta was released. They had a version of Windows NT/w2k that ran on Alpha 64 chips, but that was in 32bit mode.

They had stable versions of AMD64 Linux running and being tested even before AMD64 was released to the open market.

How can they claim stability without the ability to test it?

Because Suse worked hand in hand with AMD to get a AMD-64 compatable OS developed and ready. They had access to special emulation tools and prototype and pre-production hardware.

AMD knew that there was no real need for 64bit in the desktop and MS wasn't going to release WinXP-64 to coincide with their release. By having a usable Linux implimentation they knew that there would be a marketable reason to use AMD64 chips over Xeons in the server room.

So they helped Suse to make sure that there was a OS ready to use the chips when time came to start selling large amounts of them. Otherwise there is no real reason to choose them over the fast Xeons.

Linux is almost infinately scalable.
You have version running on the smallest memory footprints imaginable. Running in many sohpisticated embedded applications and stuff like that.

There is an embedded Windows too, and from what I've heard it ain't bad. [/quote]

Didn't say it wasn't. It's just that with the introduction of 2.6 it made it more usefull for embedded platforms. Lots of the internal code was seperated and cleaned up. They also removed more of the x86 assumptions to make it better able to be adapted to more simplistic and non-compatable archatectures.

Also there is a lot of work going to into specialized versions of Linux. To make it more realtime-like in it's behavior. Numerious projects, forks, and patches are created by developers trying to make Linux more suitable for situations were more full-flavored OSes like embedded Windows or traditional Linux are unsuitable. Progress is progressing, and that's basic innovation and why I mentioned it.

Of course, I'd love to see some innovation out of Redmond or Cupertino. I don't think there has been a whole lot lately. Most of these wizbang features came from towns other than these two.

I wouldn't know. They keep that stuff a secret and won't let me play around with it like I can with Linux.

Umm, wouldnt they have to actually get it all working first? While many of your comments fair, this one was just silly. There is no reason to think that .Net would run better on linux, especially when you consider the GUI side of it.

No, but neither does MS have a it all actually working either. Not until your going to get longhorn is when your going to see the real potential of this stuff.

My point wasn't that Linux will be better, or faster, or whatever metric you use to determine "greatness" in software. It's that the Mono team has a fairly decent chance of doing it first.

The nice thing about Linux is that it's easily adaptable. Everything is loose, peices are easily interchangable.

For instance, I could if I wanted to remove X.org for XFree86. I could, or any other decent Linux geek, do that in a short amount of time. Less then a couple days I could have a decently running OS with a different X server.

I could even replace it with a commercial X varient such as Accelerated-X and get better 2d and 3d performance out of my laptop. Every peice is interchangable, even by non-developers.

So it's very likely IF mono gets stable Windows.forms and stuff like that, (Mono is already 1.0 and consdidured usable right now)and IF the developement community warms up to it (unlike Java), you could have a very tight integration of Mono (.NET) into the Linux desktop in a short order.

It all depends on how quick MS get it's butt in gear. If longhorn is out 2005, then MS definately beat Linux. If Longhorn gets out 2006, they will still beet Mono/Linux. However Linux will still have a good chance because it will take a couple years, if more, before people will begin to switch over to Longhorn.

Now if Longhorn isn't going to be out, until, say 2008. Then Linux will have it beat.

But that's just time, MS will still have massive userbase, still vastly oversized Linux. Probably for the next 20 years even. Who knows?

But we are talking about innovation, not market realities. MS just never realy was that innovative. It just wasn't. They have a lot of money invested in that stuff now, but who knows? Stuff like that takes more then just money.

Hell, lots of there most popular products, such as Exchange, were created by other companies before they were bought out by MS.

But this is not MS vs Linux. It's just who does what and what innovations Linux does.

MS is doing some crazy stuff. Stuff that burned Mac OS.

For instance Longhorn has a all new API. All Win32 programs, all developers still developing Win32 apps will be isntantly obsolete. Longhorn will be able to develope and run Win32 apps, but you can do that in Linux nowadays, too.

Hell it's possible to write entire windows programs in Linux just by using Wine and tools like GCC with it's cross platform compiling capabilities. As long as you stick to documented Windows features Win32 apps are nearly as stable in Linux as they are in Windows. It's the hidden or mis-documented features that causes problems. Of course nothing beats running Windows apps then Windows itself, buy a long shot. It's just a for-instance.

here check out Mono

They have .Net enviroments for Windows, Linux, Linux AMD64, OS X, and others. It's quite interesting, especially if you considure the possiblities. GTK# is suppose to be pretty snazzy.

Also they have a IDE for Mono that you can check out.

I am no developer though. I am just repeating things I've read other places.

 

[drag]

Originally posted by: n0cmonkey

Originally posted by: Nothinman

How can they claim stability without the ability to test it?

Because it was running on the emulators AMD was using before there was real silicon produced.

Ahh, it was one of those "sure it works! Test? Yeah, we've uh tested it." kind of things instead of a legitimate claim.

I said they had it working before the chips were released to the public. AMD64 had emulators that they gave suse and freinds to use, but also they had pre-production CPU's and motherboards to use.

If you'd had gone and looked at Nvidia's website, for instance, you'd notice that they had AMD64 drivers for Linux months before anybody in the general public could possibly use it.

Linux is a fairly dominate OS in the server world. Windows is still more popular, about a 5 to 10 ratio (like is about 25%-30% to Windows 50%)

But I figured that AMD was hoping to get Opterons popular in the server world. It's main advantage over Intanium is it's price per performance and it's main advantage over Xeons was it's 64bit-ness.

But what is the point of that if you can only use 32bit compatability mode?

 

[n0cmonkey]

Originally posted by: drag

It is a innovation, it may be old news. However It's certainly better then any other filesystem out there that I am aware of that gets used by Windows vs OS X vs Linux.

Then give credit where it is due: SGI came up with it. SGI ported it to Linux. SGI released it under an acceptable license. Not the Linux developers.

There was a beta version of W2k that was somewhat 64bit compatable, never made it to market and they scrapped it shortly after the beta was released. They had a version of Windows NT/w2k that ran on Alpha 64 chips, but that was in 32bit mode.

Or you can get Windows-64bit.

Because Suse worked hand in hand with AMD to get a AMD-64 compatable OS developed and ready. They had access to special emulation tools and prototype and pre-production hardware.

I am aware of that, but until the product is out there, there is no way to tell if anything running on it will be stable.

AMD knew that there was no real need for 64bit in the desktop and MS wasn't going to release WinXP-64 to coincide with their release. By having a usable Linux implimentation they knew that there would be a marketable reason to use AMD64 chips over Xeons in the server room.

So they helped Suse to make sure that there was a OS ready to use the chips when time came to start selling large amounts of them. Otherwise there is no real reason to choose them over the fast Xeons.

They helped a lot more than just SuSE, but yes, I understand. But they can't claim stability without having something solid to show.

Didn't say it wasn't. It's just that with the introduction of 2.6 it made it more usefull for embedded platforms. Lots of the internal code was seperated and cleaned up. They also removed more of the x86 assumptions to make it better able to be adapted to more simplistic and non-compatable archatectures.

Also there is a lot of work going to into specialized versions of Linux. To make it more realtime-like in it's behavior. Numerious projects, forks, and patches are created by developers trying to make Linux more suitable for situations were more full-flavored OSes like embedded Windows or traditional Linux are unsuitable. Progress is progressing, and that's basic innovation and why I mentioned it.

It's not innovation.

I wouldn't know. They keep that stuff a secret and won't let me play around with it like I can with Linux.

Make sure no one has messed with your host file, or your firewall or something. You shouldn't be blocked from most of the sites that actually have information on recent (or even old) innovations.

EDIT: Screwed up formatting, and forgot the Windows 64bit link.

 

[drag]

Make sure no one has messed with your host file, or your firewall or something. You shouldn't be blocked from most of the sites that actually have information on recent (or even old) innovations.

Oh, that's cute.

You show me the link to were I can download Microsoft's CVS (or whatever revisioning system they use) latest snapshot of their Avalon, or next generation printing API for longhorn.

Oh, and BTW if you could give me a link to were I can get the latest OS X beta so I can setup Pear on my Linux box and play around with it.

Then I will see if I can work on my firewall issue.

Because I can find lots of articles that talk about it, but nothing that I can realy use or play around with

What is the point of reading about stuff that may or may not exist in the near, or distant, future? Except for intellectual curiosity.

 

[n0cmonkey]

Originally posted by: drag

Make sure no one has messed with your host file, or your firewall or something. You shouldn't be blocked from most of the sites that actually have information on recent (or even old) innovations.

Oh, that's cute.

It was intended as a silly way to say "read my comment again, I think you misunderstood." My point was that both Apple and Microsoft have been pretty quiet in the innovation department lately.

Apple has been doing some neat stuff recently. Not necessarily technilogically innovative, but neat.

What I heard about WinFS just reminded me of BeFS. The "new" security posture Microsoft is promoting was taken from OpenBSD, and they ADMIT it. Some of their new security features have been available in other OSes for a while (atleast Linux and OpenBSD).

You show me the link to were I can download Microsoft's CVS (or whatever revisioning system they use) latest snapshot of their Avalon, or next generation printing API for longhorn.

I couldn't even provide a link to a description of Avalon.

Oh, and BTW if you could give me a link to were I can get the latest OS X beta so I can setup Pear on my Linux box and play around with it.

Here's the core. Seriously.

Most of the techy innovations are definitely coming from the FOSS community. But most people aren't too interested in them.

 

[RyanLM]

Well then, considering that in my Java example the JREs used were both from Sun so they should be roughly equal Linux must have an advantage somewhere else.

Without knowing specifics, it is hard to say where the advantage/difference lies. Just because the product is made by the same company doesnt mean it is all equal. Considering Suns hate for MS, nothing would surprise me. In all seriousness, low level performance is damn near equal on both systems.

Depends on what the criteria are, I know of people doing development in C# in 100% Mono right now.

That is fine, but that doesnt mean the mono runtime is as good as MS's or as complete. I am not arguing that mono is useless, not at all. I was arguing the assumtion that mono is somehow automatically/soon to be superior to MS's implementation on windows because its on linux. Doing a quick google - http://geri.cc.fer.hr/~ivoras/web2/papers/monoperf.html it seems that in this small demo, MS's runtime not only outpaced Mono by a large amount, it also out paced gcc.

Why would it be a kludge on non-Windows platforms?

Because it deviates from the standard. And creates no set replacement standard. First Wine, now GTK - and what tomorrow?

No, but neither does MS have a it all actually working either. Not until your going to get longhorn is when your going to see the real potential of this stuff.

MS has 1.1 working just fine. It has plenty of potential even without the advancements of .Net 2.0. MS already has Sun playing Catchup, which isnt bad for 1.0.

It's that the Mono team has a fairly decent chance of doing it first

I guess I just dont see how since they are still playing catchup. MS already is on betas of .Net 2.0

Not to mention that one of the biggest pluses for .Net is Visual Studio, an application which Mono Develope cant even come close to matching, let alone passing.

Now if Longhorn isn't going to be out, until, say 2008. Then Linux will have it beat.

Beat it to what though? It has currently already won, hasnt it? It is already out there, in place, and plenty of market penetration. I am reading the rest of your post, and I am sort of understanding what you are saying. I am a developer, I work with .Net on a daily basis, and I am familiar with Mono - which is why I know it is a good start, but it is NOT up to production level.

I agree linux is a very pliable system, and you can bend it almost any way, however that is rarely the "best" tool for job. There is always the right tool for a job, and most of the time a jack of all trades is a master of none. In this case, I consider most things .Net still under the windows domain, and will for at least the next couple years.

However, I do love competition, and OS debate for that matter. As far as Innovation goes, Longhorn has me far more interested than OSX or Linux. Speaking as someone who has actually used the 3d enabled version of Avalon first hand, seen some of the yet to be released UI elements and innovations - It is jaw dropping, and fundamentally amazing that it uses so little CPU. The way the programinig model blurs the line between application and website is to me the most innovative. A freaking application in an XML file that is totally dynamic and visual, sserved by a website is amazing to witness first hand. But I am comming from a developers standpoint, I spend my day in VS.net and I have yet to find any other IDE that even compares to a fraction of the feature list, let alone the ease of use. It is hard to get excited about other platforms when I feel like I have to take a step back in time to develope on them.

 

[drag]

Originally posted by: n0cmonkey

Originally posted by: drag

Make sure no one has messed with your host file, or your firewall or something. You shouldn't be blocked from most of the sites that actually have information on recent (or even old) innovations.

Oh, that's cute.

It was intended as a silly way to say "read my comment again, I think you misunderstood." My point was that both Apple and Microsoft have been pretty quiet in the innovation department lately.

Apple has been doing some neat stuff recently. Not necessarily technilogically innovative, but neat.

What I heard about WinFS just reminded me of BeFS. The "new" security posture Microsoft is promoting was taken from OpenBSD, and they ADMIT it. Some of their new security features have been available in other OSes for a while (atleast Linux and OpenBSD).

OK. Sometimes Text-based conversations leave a lot to be desired. (I was agreeing with you in my reply to your first comment about lack of info.

You show me the link to were I can download Microsoft's CVS (or whatever revisioning system they use) latest snapshot of their Avalon, or next generation printing API for longhorn.

I couldn't even provide a link to a description of Avalon.

Well here is what I got with "I feel lucky" in google's search of Avalon Windows API

Realy interesting read.

Oh, and BTW if you could give me a link to were I can get the latest OS X beta so I can setup Pear on my Linux box and play around with it.

Here's the core. Seriously.

Most of the techy innovations are definitely coming from the FOSS community. But most people aren't too interested in them.

[/quote]

Ahhhhh, isn't it nice though?

FLOSS: The geek's choice.

 

[drag]

edit:

ARGGGG double post!

Ohwell.

Here is what I'll say instead of a double post:

But I am comming from a developers standpoint, I spend my day in VS.net and I have yet to find any other IDE that even compares to a fraction of the feature list, let alone the ease of use. It is hard to get excited about other platforms when I feel like I have to take a step back in time to develope on them.

Some people say that IDE's in general are the wrong direction, though. some discussion. I've read stuff that with experienced people moving from a IDE to a customized enviroment with a advanced text editor (such as vi or Emacs) makes them actually more productive then having to deal with a IDE.

But what do I know? (answer: Not much).

I think that one thing about open source or free software IDE's, like Eclipse, is that there is just not a real demand for them in the Linux/Unix world. Lots of developers used to operating with shells and highly customized setups see little reason for sacrificing it for a IDE. Of course it's critical for attracting Windows developers to flip over to the strange side, but any project isn't probably going to attract much old-school talent. For better or for worse.

 

[n0cmonkey]

Innovation coming out of the FOSS community:
Priv Sep
FreeBSD jails
Usermode Linux
Advanced memory protection (W^X and stackghost(?) for OpenBSD, PaX for Linux)
Better tcp port randomization in OpenBSD (not really innovation, it makes perfect sense, but no one seems to do it )
Systrace for OpenBSD, NetBSD, Linux, and OS X (while I know this wasn't the first time someone thought of intercepting and restricting syscalls, it's the first implimentation I can think of and definitely the best out there)

Recent Apple innovation:
Airtunes
spotlight is interesting. (description: Tiger?s Spotlight can find email messages, calendars and contacts along with documents, movies, images ? any kind of file ? all at once.)

 

[drag]

Don't forget about Fedora's impimentation of Execsheild (Pax and Exec sheild are still both better then just NX)if your looking into security stuff, or the clunky, but effective SELinux stuff.

 

[n0cmonkey]

Originally posted by: drag
Don't forget about Fedora's impimentation of Execsheild (Pax and Exec sheild are still both better then just NX)if your looking into security stuff,

There are a number of technologies that came before PaX and W^X, but I only mentioned the good ones.

or the clunky, but effective SELinux stuff.

I won't really consider this innovative, until they make it usable. I guess I should mention grsecurity though. A little bit different, and a bit more usable.

 

[drag]

I thought pax came before execsheild....

I figure they use execsheild because it's less likely to break stuff. But oh well.

 

[n0cmonkey]

Originally posted by: drag
I thought pax came before execsheild....

I figure they use execsheild because it's less likely to break stuff. But oh well.

No, I think exec shield came first. There was a link to exec shield on the pax site, with a mention of getting some logic from there or something. Not positive though.

 

[Nothinman]

Without knowing specifics, it is hard to say where the advantage/difference lies. Just because the product is made by the same company doesnt mean it is all equal. Considering Suns hate for MS, nothing would surprise me. In all seriousness, low level performance is damn near equal on both systems.

Right, because Sun is going to sabotage their product on the most widely used platform out there out of spite.

That is fine, but that doesnt mean the mono runtime is as good as MS's or as complete. I am not arguing that mono is useless, not at all. I was arguing the assumtion that mono is somehow automatically/soon to be superior to MS's implementation on windows because its on linux. Doing a quick google - http://geri.cc.fer.hr/~ivoras/web2/papers/monoperf.html it seems that in this small demo, MS's runtime not only outpaced Mono by a large amount, it also out paced gcc.

Noone ever said gcc was the fastest compiler, just that it was the most portable. And I believe on win32 console output is noticably slower than on unix systems, it would be interesting to see if redirecting that output to a file changed the speed at all. And if you noticed that Mono bench was done on FreeBSD, not Linux which could make a difference. And you have to rememeber that MS has had longer to work on their .NET implementation than the Mono people, I'm sure things will get better after Mono has been out for a while and more people have had time to mess with it. But frankly I'd rather run a slow Mono app on Linux than a fast Mono app on Windows.

Because it deviates from the standard. And creates no set replacement standard. First Wine, now GTK - and what tomorrow?

What standard? Windows.Forms will use MFC now and Avalon later (or whatever it's called on Longhorn) so is it a kludge too? I believe they tried to use Wine originally to save time and later realized it was too much of a PITA to do properly since Wine is so much more than that because of all of the because of all of the sh!t it has to deal with trying to implement a full win32 system. Implementing it in GTK will be much cleaner and better overall and I happen to like GTK more than MFC or whatever anyway.

And frankly since Windows.Forms is a higher level library than MFC, GTK, etc it makes sense that it doesn't matter at all what window kit is used to draw the windows, just that the Windows.Forms API is fully implemented and works.

Not to mention that one of the biggest pluses for .Net is Visual Studio, an application which Mono Develope cant even come close to matching, let alone passing.

Yea it's a huge plus that you have to buy the version of VS that matches the version of .NET that you want to use, if you were unlucky enough to buy 2002 there's no way to use the latest .NET runtime in there. Way to go.

 

[DarphB]

Right, because Sun is going to sabotage their product on the most widely used platform out there out of spite.

I wouldnt put it past Sun But, based on the numbers in the link provided it seems Java runs a bit faster on Windows looping through that code than on *nix. More importantly, .Net runs faster than just about everything else, ON WINDOWS.

And I believe on win32 console output is noticably slower than on unix systems, it would be interesting to see if redirecting that output to a file changed the speed at all.

If it was "noticably" slower on windows, wouldnt the windows scores be worse than the unix machines?

And if you noticed that Mono bench was done on FreeBSD, not Linux which could make a difference.

Could but shouldnt. Not enough to make up the difference in any case.

But frankly I'd rather run a slow Mono app on Linux than a fast Mono app on Windows.

So you dont like using the best tool for the job? In this particular case the managed app is faster/as fast as the normal compiled app. Not to mention kills java's performance. Why on earth would you pick it? Are you a Zealot?

What standard? Windows.Forms will use MFC now and Avalon later (or whatever it's called on Longhorn) so is it a kludge too?

They actually are made to work together, and Windows Forms will not go away like Wine support might. Also, you will have consistant conventions all throught the framework, which cannot be garranteed with the flavor of the week implementations others will come up with.

Implementing it in GTK will be much cleaner and better overall and I happen to like GTK more than MFC or whatever anyway.

It is my understnading that WinForms are a GDI+ abstraction, not MFC. Either way, the framework built around the graphics classes is the best I have seen yet.

Yea it's a huge plus that you have to buy the version of VS that matches the version of .NET that you want to use, if you were unlucky enough to buy 2002 there's no way to use the latest .NET runtime in there. Way to go.

And an exiting product is supposed to magically support all of the new frameworks and have designers, tools, and optmizations for future releases? BTW, you can MS releases many version of the Dev tools, some free, some $100, some more, its hardly an issue if coding is what you do.

Bottom line, VS.net is a time saver, and the right tool.

 

[Nothinman]

Could but shouldnt. Not enough to make up the difference in any case.

You never know, FreeBSD has a completely different kernel, libc and thread library.

So you dont like using the best tool for the job? In this particular case the managed app is faster/as fast as the normal compiled app. Not to mention kills java's performance. Why on earth would you pick it? Are you a Zealot?

I like using the tools I like, whether other people consider them the best for the job or not. I could probably be considered a zealot, but in most applications the performance difference won't be that big of a deal unless it's that bad across the board. And even if it is say 4x across the board it'll only be noticable in things that take more than a second or two. I mean, if opening a menu takes 1ms on Win32 and 4ms on Linux, who cares as long as it works? Eventually it'll get sped up so I'd rather stick it out on Linux than boot Windows.

They actually are made to work together, and Windows Forms will not go away like Wine support might. Also, you will have consistant conventions all throught the framework, which cannot be garranteed with the flavor of the week implementations others will come up with.

Ah so the only reason it's a kludge is because MS designed MFC and Windows.Forms.

It is my understnading that WinForms are a GDI+ abstraction, not MFC. Either way, the framework built around the graphics classes is the best I have seen yet.

So Windows.Forms on Windows implements all of the widgets from scratch instead of using what MS already had done? So you'd rather the Mono people do the same and ditch GTK and come up with a new widget set and use straight xlib calls to avoid being a kludge?

And an exiting product is supposed to magically support all of the new frameworks and have designers, tools, and optmizations for future releases? BTW, you can MS releases many version of the Dev tools, some free, some $100, some more, its hardly an issue if coding is what you do.

Not magical, it wouldn't be hard to have the class definations in a file that VS.NET loads or even just parse the header files to get the attributes that popup in the code editor. The changes to existing classes should be small since they want to maintain compatibility and new ones should be easy to pickup via things like control registration. But they probably don't want to make VS any bigger and slower than it is already. IIRC the only difference between VS 2002 and 2003 was support for .NET 1.2 and the cheap upgrade period only lasted like a month.

Bottom line, VS.net is a time saver, and the right tool.

VS.NET isn't bad but it's big, slow and the MSDN search tool sucks ass.