MP3, WMA, etc playback, probably from CD-R(W)s as well.
Built-in ethernet and ability to hook up to a Windows PC for multimedia access (see above). Don't forget Win2k (and WinME I think) have Connection Sharing built-in...
MS buys chunk of broadcast time from 2nd&3rd rate cable channels in the wee hours and starts piping encrypted exclusive programming for the X-Box. Who needs Time Warner when the Golf Channel will do?
With built-in ethernet, and broadband coming to the home, Video over DSL is around the corner. MS buys Covad/Northpoint and uses them for on-demand video to the XBox.
MS aint dumb folks. That machine has a lot of horsepower that would be wasted on just games. Plus, it's built on a well-known, widely-supported architecture. Getting multimedia capabilities, networking, etc. on this thing will be trivial compared to what Sega, Nintendo or Sony have to go through.
No, it should really read: IBM kills Dynix. IBM owns Sequent (them of the Dynix OS) and has been selling it pretty agressively as an enterprise platform (and it is a good one too). Well, Sequent has always used x86 processors --Monterey was supposed to unite AIX and Dynix, breathing life into the Sequent line post-IA64.
So basically they're killing Dynix (it was supposed to die) and substiting it with AIX RL (Linux). I.e. Linux is getting its first, official mainframe/micro line --yeah you can run it on RS/6000s, but you can also run plain AIX. In a coupla years you will *only* be able to run AIX RL on IA-64 Sequents...
engineers never lie; we just approximate the truth.
Aren't y'all sick of rehashing the same rumors over and over? Let me offer some potential alternatives that would keep the pagecounts here high and the/. crew happy:
New Amiga-based PDA announced.
Microsoft Funds "KDE Foundation".
Linus Torvalds admits: "I run Win2K".
Pat Buchanan to nix Copyright Law: Geeks vote for Pat!
UI Expert proves: Vi is better than Emacs.
Finally free: the Linux Kernel is BSD'ed
etc, etc, et-freaking-cetera...
engineers never lie; we just approximate the truth.
I've seen plenty of comments mentioning Java's shortcomings but none mentioning C++'s: name-mangling and the lack of an industry standard.
Case in point: I work for a company that deploys a large C++ application on 5 platforms: Solaris, AIX, HPUX, NT and Dynix. For part of the functionality we rely on licensed, commercial, dynamically linked C++ binaries. The vendor of those binaries builds them with the native compiler on each platform. However, our code is heavily templated and performance-dependent and can only be compiled with 1-2 compilers, Kai's being the only one that works on all platforms... since Kai uses a different name-mangling structure than each native compiler, we've had to write C wrappers for everything.
Now, it seems that an ECMA-compliant, standardized language (if C# becomes that) will fix a lot of this nonsense...
engineers never lie; we just approximate the truth.
You can get an advanced unix shell for NT --I've seen a zsh port, and zsh can expand spaces in filenames (by quoting the path automatically) or you can shell out the money for 4NT (pardon the pun) which can also do that.
engineers never lie; we just approximate the truth.
Go to Ericsson's site and take a look at this baby. Approx. 2x4 in for a *global* (Europe/US) GSM phone. If that phone had some of the functionality of a PDA, I wouldn't want anything else.
But hold on; what else do you need? a decent screen, an easy input system (e.g. Graffiti), a calendar, etc. Well, what if the screen and input part was a separate device, a "dumb PDA" if you will, about as big as that cell phone, but all the processing, memory, etc, were in the cellphone itself? So, e.g. you could have access to your calendar without the "dumb PDA" auxiliary device, but when you needed some better I/O, you could lug the tiny "dumb" terminal with you, or even use your laptop instead?
That's exactly the promise of Bluetooth: a "Personal Area Network", where naturally, the cellphone becomes the CPU (because more people are likely to lug around a cell than a Palm) and any other device (a PDA, a laptop, a printer, even your Casio wristwatch:-) are terminals to the data stored there. Nokia, Ericsson and Motorola will take us there; they are all part of the Bluetooth SIG, and they are behind Symbian, which will probably end up kill the Palms --better OS, better industry support.
engineers never lie; we just approximate the truth.
I have been administering a couple of high-end Linux clusters, and let me tell you: NFS for distributed writes is a disaster under Linux.
After days and days of testing we eventually came to the conclusion that linux NFS cannot deal with multiple writes to the same server at the same time. We were using quad Xeon IIIs and gigabit networking (cLAN by Giganet) on a dedicated, hardware RAID 5 array, so hardware was definitely not an issue (unless the RAID 5 was tripping us up). I have heard the same complaints from other Beowulfers, so this seems to be indeed a major problem.
I would suggest you look to a Solaris server with maybe Linux clients, if you indeed want to deploy Linux (I am assuming, since this is/. after all). I haven't had any experience with such a setup, I just assume Solaris can do NFS the best:-)...
engineers never lie; we just approximate the truth.
Hear, hear, you're right on the money... this thing is totally unfeasible commercially at least. Plus, to add to your comments: I doubt they can get SONAR to work at all, at least from a supersonic version. It being faster than sound coming from the SONAR and all:-)...
engineers never lie; we just approximate the truth.
Well, I got me an AE degree too; such vehicles exist already --they are called wave-riders, riding the shockwave their blunt body is creating. Look at a buncha X Planes, and most significantly the X-33.
Also, all modern ICBM designs induce shock waves at their nose to push air aside and reduce overheating of the rocket body --at some point in the Cold War that was such a huge breakthrough that the US had fake, wooden, sharp noses attached over the blunt noses when moving the ICBMs around...
engineers never lie; we just approximate the truth.
Nit to pick: I doubt they're considering 340m/s as Mach 1. Underwater, that's probably gonna be closer to 1.5 Km/s --speed of sound changing with medium density and all that...
engineers never lie; we just approximate the truth.
That's not a bad idea... but the big difference is that car companies don't hand out their 'mules' (pre-production prototypes) to regular drivers. I.e. if the car press gets a whiff of how fast the new Corvette goes, or how it feels to drive one, it has to be because GM leaked the info.
Another fundamental difference of 'hardware' and software engineering: in hardware your components have already been tested to death, you're usually only testing the integration and any new technologies (e.g. how many different engines did the Corvette have in its 50-yr history? 2? 4? 10? how many where also available in other GM cars? every single one).
OTOH, software has to exhaustively test every thing as much as they can, as so much crap is interdependent and people keep reinventing the wheel over and over again. More new code->bigger QA dept->need for 'public' or NDA-based beta programs.
engineers never lie; we just approximate the truth.
I am pretty sure you can squeeze an MPEG-2 stream (DVD) over an 1MBps connection; there are a coupla companies already working on video over DSL, so it has to.
engineers never lie; we just approximate the truth.
My understanding (and this is second hand, so take it with a few grains of salt) is that PHP4 actually compiles to bytecode now, and that the compiler will just help you distribute that bytecode. Of course, if this is true, I don't understand why Zend doesn't have PHP Compiler ready now...
Now, about JSP: instead of me trying to convince you why it should be thrown away, read this great article from Servlets.com --hardly the place for Java critics...
engineers never lie; we just approximate the truth.
I said I havent tried the Java-thru-PHP approach. I am very curious to try it, so it might happen. But anyways:
You are correct, JSP is translated to Java and then compiled as a servlet. Now, what does that mean for the end-developer? * Is JSP faster than PHP/ASP? No, not out of the box. Both PHP and ASP are compiled to bytecode, so loading time between a PHP/ASP page and a JSP servlet are not far off. The true advantage of JSP comes from the ability to communicate with Java beans that can do very nifty tricks (like DB connection pooling, persistent objects, etc). Other than that, a JSP servlet is not static (i.e. always loaded); it has to communicate with static beans/Java classes to get those advantages. Now, this was true a few months ago with JServ and GnuJSP, Tomcat might have changed things a bit. But what if PHP can do the same trick as JSP (i.e. communicate with persistent Java objects, so you don't have to replicate them with each page invocation)? I think some benchmarks would shed some light on this one. * JSP is syntactically ugly. Both ASP and PHP have an advantage there. * JSP is nowhere as widely used as either PHP or ASP. The 1.0 and 1.1 implementation where so very different, forcing people to port stuff over. I.e. there's a much larger libary of ready-made code for either ASP or PHP over JSP. * Most importantly, (to me at least): JSP is Java; as much as I love Java for serious programming, it sucks for scripting: it's a rigid, serious programming language. PHP/ASP OTOH have all the nifty shortcuts to make code easier to develop and read (e.g: variables within strings, automatic conversion between strings and numbers, etc, etc).
As I said, I love servlets. For serious applications, servlets kick ass. For RAD, servlets are a pain and a half. PHP/ASP Java connectivity can bridge that gap and get rid of that half-baked technology Sun came up with to be in the server-side-scripting market (i.e JSP)...
engineers never lie; we just approximate the truth.
Yeah, but this test was comparing apples to oranges: * C was given the choice of two not-very-good compilers: GCC and MSVC. From experience, I have seen the same code (especially math or array-intensive code) execute an order of magnitude faster when compiled with Kai CC or Portland Group CC. OTOH, Java was using the top of the line compilers and JVM (e.g. MS's JVM is well known to be much faster than even Sun's in Solaris...) * Java had the advantage of run-time optimization. If you go to Ars Technica and read up on HP's Dynamo, you'll see how run-time optimization *alone* can give you a 15-20% improvement in speed with *compiled* binaries. Granted, run-time optimization is 'in the box' for the Java platform while, besides Dynamo, C/C++ are stuck without it.
Even if you dismiss the run-time optimization advantage as an integral part of the test, the choice of compilers *did* have a speed effect...
At any rate, I *am* a Java fan --I am just curious to see some true, fair benchmarks.
engineers never lie; we just approximate the truth.
ASP is nice; I am a PHP man myself, but I recently had to look into ASP for a consulting gig. ASP has a coupla features I'd really like to see in PHP, but in the end of the day, I'd still pick PHP's versatility.
But anyways, back to ASP vs JSP: JSP sucks! there's no other way to put it: JSP is a hack on top of a kludge: it's ugly syntactically, it tries to fit a square peg (Java) into a round hole (server-side scripting) and it has nowhere the built-in functionality and features of either PHP or ASP.
Now, OTOH Java *servlets* (not JSP) rock! Session management, persistent objects, the ability to leverage Java's impressive APIs; it's the answer to any serious enterprise-level solution.
So, what to do? PHP4 is supposed to be able to manipulate Java objects *directly* (it's in the spec but I haven't used it). If that works, you can now throw away JSP, put all your middleware logic in Java beans and servlets and use PHP to do the string-parsing and HTML pushing...
If I were you, I'd try to push for PHP as a 'peer to' ASP with the advantage of being able to use your old Java work...
engineers never lie; we just approximate the truth.
I posted this on Kuro5hin earlier today. As a general rule, I don't like regurgitating posts, but this new/. story is sorta making my point, so here it is [dons asbestos suit, sets procmail filters on]:
I am amused when people (especially those on that Other Site) claim that anonymity is: a) a God-given right, or b) what the Net was based on.
Bollocks! (b) especially is wrong... when the ArpaNet/Internet started there was no anonymity because everybody on the Net just plain knew everybody else --the couple of CS schools and National Labs online were and still are a pretty close community. Then for years, the internet was full of static-only IPs and real-people-only e-mail addresses: you were far from anonymous.
Anonymity on the Net was introduced either directly by real privacy advocates (who can forget anon.penet.fi that Scientology practically shut down) or, more commonly, by anonymity-through-obscurity: FreeNet (the original Freenet, not the Gnutella/Napster clone) and later AOL, HotMail, etc, etc.
Now, (a) is even less true: what human society ever was based on anonymity? NONE. Actually, anonymity in the Real World is either associated with cults and/or military organizations --where the person 'must' be erased to establish groupthink-- or to avoid prosecution.
But what really, truly bugs me (and you cannot say than on/. on penalty of burning at the stake) is this whole self-serving attitude that the new wave of 'privacy advocates' is assuming. Anon.penet.fi served a noble cause; it might have been used for practical jokes and/or ordering God-knows-what, but it also served as the only way for whistleblowers to speak (e.g. the Scientology case). The 'new' Slashdot-fed "privacy advocates" (and they deserve to be in quotes) are using the otherwise noble flag of privacy (a whole different issue than anonymity as another post explained very well) to *justify* their *stealing*.
If you are too cheap, or too lazy and are gonna break the law by going on Napster and downloading some MP3s (which I won't say I haven't done), you should have, really, the balls to stand up and say: "I know this is wrong, I know this is stealing someone else's work without their consent, but I am just too damn {lazy|cheap}!". The attitude of "Since I am doing this, it *has* to be right" (vis-a-vis "It is right, that's why I will do it") is Puratinistic and hypocritical.
Are record companies charging too much for a CD? yes they are (cry-babies; a CD back in Europe can go for $20 easily). Are we *justified* to go out and steal the companies' property? no we are not. "Justified" is a very heavy word. Don't overuse it...
engineers never lie; we just approximate the truth.
Err... I for one always assumed that AOL's move to Linux will really just be a move to *Mozilla*, not another OS platform. People have to realize that Mozilla as it stands now is an *application platform*, as able (if not more so) than Java.
People can write whole Web-based applications in Mozilla. Look at Active State and Zope. AOL is smart enough to use this to their own advantage. Suddenly they got a version of AOL that can be deployed *over the network* to any PC or OS, w/o those pesky CDs. Added bonus: the main OS for testing this platform is (coincidentally, because of Netscape's incompetence) a *free* one, that they can do anything to.
Welcome back to the future: Mozilla is the VT100 of tomorrow:-)...
engineers never lie; we just approximate the truth.
I think one of the biggest missing links in the Free Software/Open Source world is a competitor to Exchange/Lotus Notes. IMAP and LDAP are a start, but we need a way to sync files, calendars, contacts, etc. I think both the GNOME and KDE people are working on something like that, so maybe finishing up a module will be enough (say file syncing).
engineers never lie; we just approximate the truth.
This Beast, this NewOS we're talking about may end up being only a remote relative of Unix, but maybe it should feel like Unix, if not for any other reason, just to be able to use the huge codebase of C/C++ code out there that's already been written for Unix.
It may sound like I am contradicting myself here, but I am really not. Unix is full of crud right now. And there are two ways to get rid of crud: 1) What BeOS or AtheOS have done; clean slate, only use portions of Unix design that you like, 2) What Apple did: Use a Unix framework and layer-by-layer remove crud, replacing it with better code of your own.
Because of the dynamics of OSS one of the rules of CatB is, in order to get people to work on a big OSS project, that some of the functionality needs to be there: i.e. shit has to work, so people can play with it. The overhead to do method (1) this way becomes much, much greater than method (2).
So, let's look around: what components are out there in OSS Unix-land that one can re-use in this NewOS w/o compromising on technology or design?:
The Linux kernel and device drivers; or if you want to be more cutting edge, the GNU HURD kernel. I'd go with Linux, because of the widespread device support.
GCC
Qt and/or GTK on the GUI front. I dunno how dependant they are on X, or if they abstract X functionality well.
There are gaping holes in this, of course, but it's much easier to start with something that already works, modify it to work under your very best design, and as components and functionality come in, you can potentially remove the old, aging components and put in new ones (e.g. swap the Linux kernel with a Mach-derived microkernel). I actually think most of this can be done starting with a Linux distro.
BTW, I don't want a great *consumer* OS. I think that distinction (between consumer and business, server and workstation) is bogus. A good design should scale; it should handle being a server just as well as being a workstation for Mom.
engineers never lie; we just approximate the truth.
...BSD with the configuration cleaned up, with Application bundles a la NeXT and the awesome-sounding architecture of Frameworks. Finally, someone cleaned *some* of the 40 years of Unix crud.
BUT the real question is: Why isn't anyone in the Open Source World doing this? Why isn't anybody cleaning up the kruft that is Unix/Linux/BSD? I don't want any more/etc,/var; I don't want badly written, unsearchable man pages. I don't want applications that install themselves in twenty different places in my system without asking. I want maintainability and manageability.
Apple has shown the way; bundles kick ass. XML for system configuration kicks ass. OSS can copy that design and improve on it:
XHTML for documentation; searchable, anotated documentation that can be tied back to the Internet --damn it, that's where "errata" belongs: as a link to my local manual; not buried in some "knoweledge base" somewhere.
Protected userland documents that are *not* tied to some a priori defined hierarchical structure. I want a 'soup' of documents I created that can be searched, sorted, sliced and diced by a local search engine.
I want a system that can introspect for a change; I want PHP to be able to find Apache by itself, configure it and itself (after asking of course) and, while it's at it, do the same for MySQL. OOP is not just for GUIs: System-wide, componentized configuration a la OSX extended even further.
X needs to be thrown away; just dump the whole ugly freaking mess. It belongs somewhere to access remote Unix machines, but it does not belong as the main layer under my GUI. BeOS (and AtheOS) has shown the way. Can we get the AtheOS GUI code and plug it in under Qt or GTK?
This is OSS: enough coders and the problems *will* be solved...
[Rant Off]
engineers never lie; we just approximate the truth.
Re:Its proponents would of course be called...
on
AtheOS
·
· Score: 2
Actually, "Atheos" in Greek *does* mean "atheist"... I dunno if it's on purpose or not though...
engineers never lie; we just approximate the truth.
I don't believe that all software will be shipped abroad --and I would want it too, coz then I could move back home to Europe:-)... there are so, so many things that require face-to-face contact between engineers to develop a common design path, common goals and of course, team spirit. Of course all this depends on the company and the people running it.
Now, back to the original poster's question. Yes, there are a lot of H-1B slave-shops. But there are also a lot of true high-tech companies that look for real talent. And usually, in the US, foreigners hang out with foreigners: if a coupla get in a firm, they will bring more. Actually, the pattern I have seen is that either the company has a *lot* of non-Americans (like mine; us foreigners are about 70% of the developer team) or almost none.
So keep that in mind when you're looking for a job; if a company makes a big deal about the H-1, the paperwork and all that, it's probably a company that doesn't have that many non-US workers, and you'll probably be better off if you don't train the HR department yourself:-)... Now, to separate the H-1 slaveshops from the legitimate companies, look at their job offer: if their offer is say less than 80% of the industry average, it's probably a slaveshop --that 20% is actually justified, because if nothing else you'd have to get up to speed with the ways of business here, plus attorney fees and crap like that.
Good luck to ya...
engineers never lie; we just approximate the truth.
Slashdot Mirror
MP3, WMA, etc playback, probably from CD-R(W)s as well.
Built-in ethernet and ability to hook up to a Windows PC for multimedia access (see above). Don't forget Win2k (and WinME I think) have Connection Sharing built-in...
MS buys chunk of broadcast time from 2nd&3rd rate cable channels in the wee hours and starts piping encrypted exclusive programming for the X-Box. Who needs Time Warner when the Golf Channel will do?
With built-in ethernet, and broadband coming to the home, Video over DSL is around the corner. MS buys Covad/Northpoint and uses them for on-demand video to the XBox.
MS aint dumb folks. That machine has a lot of horsepower that would be wasted on just games. Plus, it's built on a well-known, widely-supported architecture. Getting multimedia capabilities, networking, etc. on this thing will be trivial compared to what Sega, Nintendo or Sony have to go through.
No, it should really read: IBM kills Dynix. IBM owns Sequent (them of the Dynix OS) and has been selling it pretty agressively as an enterprise platform (and it is a good one too). Well, Sequent has always used x86 processors --Monterey was supposed to unite AIX and Dynix, breathing life into the Sequent line post-IA64.
So basically they're killing Dynix (it was supposed to die) and substiting it with AIX RL (Linux). I.e. Linux is getting its first, official mainframe/micro line --yeah you can run it on RS/6000s, but you can also run plain AIX. In a coupla years you will *only* be able to run AIX RL on IA-64 Sequents...
engineers never lie; we just approximate the truth.
New Amiga-based PDA announced.
Microsoft Funds "KDE Foundation".
Linus Torvalds admits: "I run Win2K".
Pat Buchanan to nix Copyright Law: Geeks vote for Pat!
UI Expert proves: Vi is better than Emacs.
Finally free: the Linux Kernel is BSD'ed
etc, etc, et-freaking-cetera...
engineers never lie; we just approximate the truth.
I've seen plenty of comments mentioning Java's shortcomings but none mentioning C++'s: name-mangling and the lack of an industry standard.
Case in point: I work for a company that deploys a large C++ application on 5 platforms: Solaris, AIX, HPUX, NT and Dynix. For part of the functionality we rely on licensed, commercial, dynamically linked C++ binaries. The vendor of those binaries builds them with the native compiler on each platform. However, our code is heavily templated and performance-dependent and can only be compiled with 1-2 compilers, Kai's being the only one that works on all platforms... since Kai uses a different name-mangling structure than each native compiler, we've had to write C wrappers for everything.
Now, it seems that an ECMA-compliant, standardized language (if C# becomes that) will fix a lot of this nonsense...
engineers never lie; we just approximate the truth.
You can get an advanced unix shell for NT --I've seen a zsh port, and zsh can expand spaces in filenames (by quoting the path automatically) or you can shell out the money for 4NT (pardon the pun) which can also do that.
engineers never lie; we just approximate the truth.
Go to Ericsson's site and take a look at this baby. Approx. 2x4 in for a *global* (Europe/US) GSM phone. If that phone had some of the functionality of a PDA, I wouldn't want anything else.
:-) are terminals to the data stored there. Nokia, Ericsson and Motorola will take us there; they are all part of the Bluetooth SIG, and they are behind Symbian, which will probably end up kill the Palms --better OS, better industry support.
But hold on; what else do you need? a decent screen, an easy input system (e.g. Graffiti), a calendar, etc. Well, what if the screen and input part was a separate device, a "dumb PDA" if you will, about as big as that cell phone, but all the processing, memory, etc, were in the cellphone itself? So, e.g. you could have access to your calendar without the "dumb PDA" auxiliary device, but when you needed some better I/O, you could lug the tiny "dumb" terminal with you, or even use your laptop instead?
That's exactly the promise of Bluetooth: a "Personal Area Network", where naturally, the cellphone becomes the CPU (because more people are likely to lug around a cell than a Palm) and any other device (a PDA, a laptop, a printer, even your Casio wristwatch
engineers never lie; we just approximate the truth.
I have been administering a couple of high-end Linux clusters, and let me tell you: NFS for distributed writes is a disaster under Linux.
/. after all). I haven't had any experience with such a setup, I just assume Solaris can do NFS the best :-)...
After days and days of testing we eventually came to the conclusion that linux NFS cannot deal with multiple writes to the same server at the same time. We were using quad Xeon IIIs and gigabit networking (cLAN by Giganet) on a dedicated, hardware RAID 5 array, so hardware was definitely not an issue (unless the RAID 5 was tripping us up). I have heard the same complaints from other Beowulfers, so this seems to be indeed a major problem.
I would suggest you look to a Solaris server with maybe Linux clients, if you indeed want to deploy Linux (I am assuming, since this is
engineers never lie; we just approximate the truth.
Hear, hear, you're right on the money... this thing is totally unfeasible commercially at least. Plus, to add to your comments: I doubt they can get SONAR to work at all, at least from a supersonic version. It being faster than sound coming from the SONAR and all :-)...
engineers never lie; we just approximate the truth.
Well, I got me an AE degree too; such vehicles exist already --they are called wave-riders, riding the shockwave their blunt body is creating. Look at a buncha X Planes, and most significantly the X-33.
Also, all modern ICBM designs induce shock waves at their nose to push air aside and reduce overheating of the rocket body --at some point in the Cold War that was such a huge breakthrough that the US had fake, wooden, sharp noses attached over the blunt noses when moving the ICBMs around...
engineers never lie; we just approximate the truth.
Nit to pick: I doubt they're considering 340m/s as Mach 1. Underwater, that's probably gonna be closer to 1.5 Km/s --speed of sound changing with medium density and all that...
engineers never lie; we just approximate the truth.
I hope you're not completely de-charging LiIon batteries (most batteries in the last 2 yrs or so). Or maybe I didn't understand your post correctly...
engineers never lie; we just approximate the truth.
That's not a bad idea... but the big difference is that car companies don't hand out their 'mules' (pre-production prototypes) to regular drivers. I.e. if the car press gets a whiff of how fast the new Corvette goes, or how it feels to drive one, it has to be because GM leaked the info.
Another fundamental difference of 'hardware' and software engineering: in hardware your components have already been tested to death, you're usually only testing the integration and any new technologies (e.g. how many different engines did the Corvette have in its 50-yr history? 2? 4? 10? how many where also available in other GM cars? every single one).
OTOH, software has to exhaustively test every thing as much as they can, as so much crap is interdependent and people keep reinventing the wheel over and over again. More new code->bigger QA dept->need for 'public' or NDA-based beta programs.
engineers never lie; we just approximate the truth.
You mean something like this?
engineers never lie; we just approximate the truth.
I am pretty sure you can squeeze an MPEG-2 stream (DVD) over an 1MBps connection; there are a coupla companies already working on video over DSL, so it has to.
engineers never lie; we just approximate the truth.
My understanding (and this is second hand, so take it with a few grains of salt) is that PHP4 actually compiles to bytecode now, and that the compiler will just help you distribute that bytecode. Of course, if this is true, I don't understand why Zend doesn't have PHP Compiler ready now...
Now, about JSP: instead of me trying to convince you why it should be thrown away, read this great article from Servlets.com --hardly the place for Java critics...
engineers never lie; we just approximate the truth.
I said I havent tried the Java-thru-PHP approach. I am very curious to try it, so it might happen. But anyways:
You are correct, JSP is translated to Java and then compiled as a servlet. Now, what does that mean for the end-developer?
* Is JSP faster than PHP/ASP? No, not out of the box. Both PHP and ASP are compiled to bytecode, so loading time between a PHP/ASP page and a JSP servlet are not far off. The true advantage of JSP comes from the ability to communicate with Java beans that can do very nifty tricks (like DB connection pooling, persistent objects, etc). Other than that, a JSP servlet is not static (i.e. always loaded); it has to communicate with static beans/Java classes to get those advantages. Now, this was true a few months ago with JServ and GnuJSP, Tomcat might have changed things a bit. But what if PHP can do the same trick as JSP (i.e. communicate with persistent Java objects, so you don't have to replicate them with each page invocation)? I think some benchmarks would shed some light on this one.
* JSP is syntactically ugly. Both ASP and PHP have an advantage there.
* JSP is nowhere as widely used as either PHP or ASP. The 1.0 and 1.1 implementation where so very different, forcing people to port stuff over. I.e. there's a much larger libary of ready-made code for either ASP or PHP over JSP.
* Most importantly, (to me at least): JSP is Java; as much as I love Java for serious programming, it sucks for scripting: it's a rigid, serious programming language. PHP/ASP OTOH have all the nifty shortcuts to make code easier to develop and read (e.g: variables within strings, automatic conversion between strings and numbers, etc, etc).
As I said, I love servlets. For serious applications, servlets kick ass. For RAD, servlets are a pain and a half. PHP/ASP Java connectivity can bridge that gap and get rid of that half-baked technology Sun came up with to be in the server-side-scripting market (i.e JSP)...
engineers never lie; we just approximate the truth.
Yeah, but this test was comparing apples to oranges:
* C was given the choice of two not-very-good compilers: GCC and MSVC. From experience, I have seen the same code (especially math or array-intensive code) execute an order of magnitude faster when compiled with Kai CC or Portland Group CC. OTOH, Java was using the top of the line compilers and JVM (e.g. MS's JVM is well known to be much faster than even Sun's in Solaris...)
* Java had the advantage of run-time optimization. If you go to Ars Technica and read up on HP's Dynamo, you'll see how run-time optimization *alone* can give you a 15-20% improvement in speed with *compiled* binaries. Granted, run-time optimization is 'in the box' for the Java platform while, besides Dynamo, C/C++ are stuck without it.
Even if you dismiss the run-time optimization advantage as an integral part of the test, the choice of compilers *did* have a speed effect...
At any rate, I *am* a Java fan --I am just curious to see some true, fair benchmarks.
engineers never lie; we just approximate the truth.
ASP is nice; I am a PHP man myself, but I recently had to look into ASP for a consulting gig. ASP has a coupla features I'd really like to see in PHP, but in the end of the day, I'd still pick PHP's versatility.
But anyways, back to ASP vs JSP: JSP sucks! there's no other way to put it: JSP is a hack on top of a kludge: it's ugly syntactically, it tries to fit a square peg (Java) into a round hole (server-side scripting) and it has nowhere the built-in functionality and features of either PHP or ASP.
Now, OTOH Java *servlets* (not JSP) rock! Session management, persistent objects, the ability to leverage Java's impressive APIs; it's the answer to any serious enterprise-level solution.
So, what to do? PHP4 is supposed to be able to manipulate Java objects *directly* (it's in the spec but I haven't used it). If that works, you can now throw away JSP, put all your middleware logic in Java beans and servlets and use PHP to do the string-parsing and HTML pushing...
If I were you, I'd try to push for PHP as a 'peer to' ASP with the advantage of being able to use your old Java work...
engineers never lie; we just approximate the truth.
engineers never lie; we just approximate the truth.
Err... I for one always assumed that AOL's move to Linux will really just be a move to *Mozilla*, not another OS platform. People have to realize that Mozilla as it stands now is an *application platform*, as able (if not more so) than Java.
:-)...
People can write whole Web-based applications in Mozilla. Look at Active State and Zope. AOL is smart enough to use this to their own advantage. Suddenly they got a version of AOL that can be deployed *over the network* to any PC or OS, w/o those pesky CDs. Added bonus: the main OS for testing this platform is (coincidentally, because of Netscape's incompetence) a *free* one, that they can do anything to.
Welcome back to the future: Mozilla is the VT100 of tomorrow
engineers never lie; we just approximate the truth.
I think one of the biggest missing links in the Free Software/Open Source world is a competitor to Exchange/Lotus Notes. IMAP and LDAP are a start, but we need a way to sync files, calendars, contacts, etc. I think both the GNOME and KDE people are working on something like that, so maybe finishing up a module will be enough (say file syncing).
engineers never lie; we just approximate the truth.
This Beast, this NewOS we're talking about may end up being only a remote relative of Unix, but maybe it should feel like Unix, if not for any other reason, just to be able to use the huge codebase of C/C++ code out there that's already been written for Unix.
It may sound like I am contradicting myself here, but I am really not. Unix is full of crud right now. And there are two ways to get rid of crud: 1) What BeOS or AtheOS have done; clean slate, only use portions of Unix design that you like, 2) What Apple did: Use a Unix framework and layer-by-layer remove crud, replacing it with better code of your own.
Because of the dynamics of OSS one of the rules of CatB is, in order to get people to work on a big OSS project, that some of the functionality needs to be there: i.e. shit has to work, so people can play with it. The overhead to do method (1) this way becomes much, much greater than method (2).
So, let's look around: what components are out there in OSS Unix-land that one can re-use in this NewOS w/o compromising on technology or design?:
The Linux kernel and device drivers; or if you want to be more cutting edge, the GNU HURD kernel. I'd go with Linux, because of the widespread device support.
GCC
Qt and/or GTK on the GUI front. I dunno how dependant they are on X, or if they abstract X functionality well.
There are gaping holes in this, of course, but it's much easier to start with something that already works, modify it to work under your very best design, and as components and functionality come in, you can potentially remove the old, aging components and put in new ones (e.g. swap the Linux kernel with a Mach-derived microkernel). I actually think most of this can be done starting with a Linux distro.
BTW, I don't want a great *consumer* OS. I think that distinction (between consumer and business, server and workstation) is bogus. A good design should scale; it should handle being a server just as well as being a workstation for Mom.
engineers never lie; we just approximate the truth.
BUT the real question is: Why isn't anyone in the Open Source World doing this? Why isn't anybody cleaning up the kruft that is Unix/Linux/BSD? I don't want any more
Apple has shown the way; bundles kick ass. XML for system configuration kicks ass. OSS can copy that design and improve on it:
XHTML for documentation; searchable, anotated documentation that can be tied back to the Internet --damn it, that's where "errata" belongs: as a link to my local manual; not buried in some "knoweledge base" somewhere.
Protected userland documents that are *not* tied to some a priori defined hierarchical structure. I want a 'soup' of documents I created that can be searched, sorted, sliced and diced by a local search engine.
I want a system that can introspect for a change; I want PHP to be able to find Apache by itself, configure it and itself (after asking of course) and, while it's at it, do the same for MySQL. OOP is not just for GUIs: System-wide, componentized configuration a la OSX extended even further.
X needs to be thrown away; just dump the whole ugly freaking mess. It belongs somewhere to access remote Unix machines, but it does not belong as the main layer under my GUI. BeOS (and AtheOS) has shown the way. Can we get the AtheOS GUI code and plug it in under Qt or GTK?
This is OSS: enough coders and the problems *will* be solved...
[Rant Off]
engineers never lie; we just approximate the truth.
Actually, "Atheos" in Greek *does* mean "atheist"... I dunno if it's on purpose or not though...
engineers never lie; we just approximate the truth.
I don't believe that all software will be shipped abroad --and I would want it too, coz then I could move back home to Europe :-)... there are so, so many things that require face-to-face contact between engineers to develop a common design path, common goals and of course, team spirit. Of course all this depends on the company and the people running it.
:-)... Now, to separate the H-1 slaveshops from the legitimate companies, look at their job offer: if their offer is say less than 80% of the industry average, it's probably a slaveshop --that 20% is actually justified, because if nothing else you'd have to get up to speed with the ways of business here, plus attorney fees and crap like that.
Now, back to the original poster's question. Yes, there are a lot of H-1B slave-shops. But there are also a lot of true high-tech companies that look for real talent. And usually, in the US, foreigners hang out with foreigners: if a coupla get in a firm, they will bring more. Actually, the pattern I have seen is that either the company has a *lot* of non-Americans (like mine; us foreigners are about 70% of the developer team) or almost none.
So keep that in mind when you're looking for a job; if a company makes a big deal about the H-1, the paperwork and all that, it's probably a company that doesn't have that many non-US workers, and you'll probably be better off if you don't train the HR department yourself
Good luck to ya...
engineers never lie; we just approximate the truth.