I was learning CORBA in 1997. Alas, it was another Sun driven technlogy, like Java, before Sun understood how the technology landscape was changing.
Bullshit. Sun was pushing RMI by 1997. CORBA is a standard backed by the hundreds of companies that signed up to the OMG. (At my side are some NextSTEP developer manuals which I'm amused to see even have an OMG logo on them).
The only FOSS CORBA implementation that came anywhere close to implementing CORBA 2.0 was Orbit
Bullshit again. Go check out omniORB, MICO, JacORB and TAO for starters. All open source, all actively maintained and all supporting CORBA version 2.0 or higher.
In the end CORBA suffered the same fate as Java for the same reason(s).
So, CORBA is like Java. Ubiquitous in big companies and capable of producing highly scalable systems with excellent reliability, but not popular with PHP wheenies on Slashdot.
No: it's all about the APIs and who's making them available. Got CORBA bindings for Google? How 'bout the National Weather Service?
Slightly tangential to your comment, but the weather channel (http://www.weather.com/), use the CORBA Event Service as the heart of their system. According to a posting on the MICO mailing list by one of their developers, the system scales to millions of events an hour.
The best book I've read on CORBA is one dealing with CORBA from a Java programmers perspective. It's called "Java Programming with CORBA", ISBN 0471376817. It has the best description of the POA I've read - I struggled with this particular feature of CORBA before, as I couldn't understand why it was designed the way it is from reading the Henning and Vinoski book. The Henning and Vinoski book is the definitive one for C++ programmers, but it is a pretty dull read even by teh standards of computer textbooks. It isn't helped by the quality of the C++ language mapping, which predates the ANSI C++ standard. The mapping could have really benefited from use of things like the standard string class and containers.
Hennings opinions should be taken with a pinch of salt. While he co-authored the bet known book on CORBA, he has since left the CORBA arena following the merger of his own company with IONA, and is now pushing an alternative technology called ICE.
As for CORBA? The concepts like the POA are too over engineered, but the underlying ideas of language neutrality, a simple interface definition language and a common wire protocol have not been bettered. SOAP has an awful interface definition language based on XML, and is also dependent on XML in many other areas, despite the high overheads of parsing the format. SOAP also suffers from using HTTP as it's primary transport, as this is also highly inefficient. RMI is a great solution for end to end Java systems, but forces a reliance on a proprietary third party bridge if it needs to be intergrated into a non-Java environment.
One area where CORBA is particularily useful, is in event or notification based systems. These standard services offer the same features as Java JMS, but also benefit from a standard wire protocol and standard mappings for languages other than Java.
Overall, CORBA's demise seems to be greatly exagerated. It's not a fashionable technology, nor it is perfect. However, it provides the glue between many systems in organisations such as banks, warehouses and media organisations that no other technology could adequately fill. A number of companies are making a very tidy living from consulting both for their own CORBA implementations as well as open source ones such as MICO and omniORB.
I can speak from personal experience as someone who has successfully used CORBA both as an RPC mechanism and an events system at my last two employers. Most recently, I used CORBA to replace a SOAP based system. Compared to SOAP, the CORBA based system was faster, easier to work with and far more reliable.
Sun is new to open sourcing its proprietary products? That strange, because amongst other things, they open sourced their implementations of RPC and NFS years ago. Sun are by no means new to this open source thing and as well as their own stuff, they've acted as mentors to a number of outside projects. For instance, Sun provided John Ousterhout with an office to use while he worked on Tcl/Tk.
I too love NetBSD, but shipping with both vi and ed is stupid. Personally, I don't think an editor should be included at all, since pkgsrc makes adding one trivial.
Swing adopted a lot of ideas from the NeXTstep API. A thorough implementation of the MVC architecture, and extensive use of design patterns. This makes it counter intuitive to programmers used to the Win32 or Motif way of doing things, but once you understand how to create decent data models and separate them from the actual GUI code it leads to more modular, maintainable code.
At work, we have three programmers who routinely program GUI applications in Java. Two use Swing, write well structured code that encourages reuse. They love Swing (except the JTree widget). The other programmer prefers SWT, which he isn't allowed to use, creates spaghetti code monsters and whinges about the Swing API all the time. This isn't an isolated case either, it is amusingly similar to the situation at my previous employer - the substandard programmer there also preferred SWT over Swing.
You forgot to mention the fact that you *still* need the sun jdk to build the package.
About a year ago I did. However, I've boostrapped all subsequent builds from the native one.
... and you conveniently snipped the bit of the Description where it says that, really, "using it in a production environment is at your own risk".
I did snip it, but the post I was replying to emphasised a requirement for SuSE packages and Linux compatibility to run Java. I snipped the output from pkg_info because I was interested in showing the availability of a native JDK and the dependencies - X11 and openmotif. As for risk, I use the native JDK in a production environment with Tomcat 5.5.x, and I haven't encountered a single problem.
Half a SuSE installation? I haven't even got Linux compatibility compiled into my kernel.
$ uname -a
NetBSD tietokone.panews.press.net 3.0_STABLE NetBSD 3.0_STABLE (LAPTOP) #0: Fri Apr 14 22:31:45 BST 2006 root@tietokone.panews.press.net:/usr/src/sys/arch/ i386/compile/LAPTOP i386
$ pkg_info jdk14
Information for jdk14-1.4.2.8:
Comment:
Java Development Kit 1.4.2
Requires:
openmotif>=2.2.3
xorg-libs>=6.9.0
Description:
This is Sun's Java[tm] Development Kit, version 1.4.2, made buildable
and usable natively on *BSD/i386 by Greg Lewis and a host of others.
I countered an assertion that there's lots of Linux savvy programmers in the world, by noting that in my experience they can't be found in much of the UK. Now, there's a fairly large IT industry in the UK, so the lack of decent programmers either means that the UK is suffering an unusual lack of them, or the original posters statement is bollocks. Given that many continental companies recruit from the UK, that must mean that there's a lack of them all over Europe. Finally, having worked in the States (Silicon Valley), I can state with some certainty that most companies are staffed by bodgers and the clueless.
with the abundance of Linux savy programmers in the world it makes very good sense to use a widely deployed system.
Congratulations. I assume you're not in the UK, as I've found it hard work recruiting programmers who are "savvy" with anything, let alone embedded Linux development.
OK, so you used BSD instead. That's fine -- but from what I've seen the BSDs embedded abilities are far less than Linux. So you probably had to do a lot of work.
Speaking from experience (writing controller software for cranes and the like), NetBSD is much easier to work with than Linux. This is even on hardware like Arcom's Viper, which come with a prepackaged Linux development kit. Linux is too difficult to strip down, and the developer kits that come with many embedded systems are based on very old versions of the kernel and libraries. The age of these kernels and libraries, and the lack of any understanding of what a stable ABI or API is amongst Linux hackers, means that getting upto date third party code to work is a nightmare. With NetBSD (and FreeBSD, not so sure about OpenBSD) backwards compatability is excellent.
Thankfully I almost have. Glibc is written in what I describe as a "very baroque" way. For instance, large parts of the code don't actually look much like C thanks to the over use of the C preprocessor. If you compare it with the equivalent parts of the C libraries in any of the BSD's, you can see how the rigid application of things like the GNU coding standards and certain peoples distinct coding practices have resulted in glibc being overly elaborate and large.
We owe RMS a huge debt because he single-handedly kickstarted the free software movement.
No, he kickstarted the GNU movement - the free software movement already existed, most notably in the form of BSD Unix. Like GNU, the BSD Unix codebase has gradually evelolved into a largely complete implementation of Unix that is not encumbered by an AT&T license. What Stallman did differently to the people at Berkeley, was to use a license that placed restrictions on the way GNU software is distributed. Stallman sees this as a necessity to ensure GNU software stays "free", by preventing someone from distributing binaries without providing access to the source code.
Most bands end up owing the record company for the priviledge of having their cds distributed by them, which is why I think most bands would be better off without record contracts.
Rubbish, unless they signed to a major in the wake of a bit of bandwagon jumping that is (Coldplay have a hit, every major signs a couple of drippy indie bands on three album deals with an option on the last two). In my own experience of small record labels, they cover your recording costs at the very least. Having recently recorded an album to the same standard at personal expense, I can now see that the outlay for a record label is considerable. Then you consider that majors have to give bands an advance that they can actually live on as well as covering their recording costs.
Having toured as support to major label bands, I can see the difference that it makes when you can spend time doing music for a living rather than a hobby. It's incredibly rare to find a semi-pro or amateur band that can put on a good a show as a band that can afford to spend weeks rehearsing solidly before a tour. The alternative for many people is trying to do music on the dole, which is pretty fucking dismal, or playing up on the non-music side of your "celebrity" - step forward Pete Doherty and countless gun toting rap twats.
Right. So bands should try and cover their costs by selling T-shirts rather than "ephemeral" stuff like CD's of their music. They better add a screen printer to their usual shopping list of ephemeral stuff like amps, guitars and synths then.
When I used to have a 50 mile drive to and from work, the only good thing about it was that I could listen to the Today programme and the 6:00pm news on Radio 4. Quality journalism, which even the fallout from the disgraceful Hutton inquiry hasn't entirely destroyed. On TV the only thing that comes close is Channel 4 news, but even that's been watered down over the last couple of years - much less coverage of international news, and a willingness to cut the original hour long format in half on regular occasions.
I'd read that IBM have a port of the Sun JDK for PowerPC Linux. While GCJ and Classpath are progressing fast, I don't think they're quite ready for all the Swing and Tomcat/JSP based stuff that I'd like to do. The Swing classes at the very least are still a bit patchy. However, both projects have come on in leaps and bounds in the last couple of years, and along with the new Fortran 95 compiler I'm looking forward to the next few releases of GCC4.
NetBSD may be a better choice for older hardware, as it consumes less resources than recent versions of Linux. Note that this is not an anit-Linux troll, much of that extra resource hungriness seems to come from the added functionality rather than superfluous bloat - and despite my personal preference for NetBSD, I'm considering putting FC5 onto my PowerBook, as there appears to be support for Java on PowerPC Linux.
Slashdot Mirror
I was learning CORBA in 1997. Alas, it was another Sun driven technlogy, like Java, before Sun understood how the technology landscape was changing.
Bullshit. Sun was pushing RMI by 1997. CORBA is a standard backed by the hundreds of companies that signed up to the OMG. (At my side are some NextSTEP developer manuals which I'm amused to see even have an OMG logo on them).
The only FOSS CORBA implementation that came anywhere close to implementing CORBA 2.0 was Orbit
Bullshit again. Go check out omniORB, MICO, JacORB and TAO for starters. All open source, all actively maintained and all supporting CORBA version 2.0 or higher.
In the end CORBA suffered the same fate as Java for the same reason(s).
So, CORBA is like Java. Ubiquitous in big companies and capable of producing highly scalable systems with excellent reliability, but not popular with PHP wheenies on Slashdot.
No: it's all about the APIs and who's making them available. Got CORBA bindings for Google? How 'bout the National Weather Service?
Slightly tangential to your comment, but the weather channel (http://www.weather.com/), use the CORBA Event Service as the heart of their system. According to a posting on the MICO mailing list by one of their developers, the system scales to millions of events an hour.
The best book I've read on CORBA is one dealing with CORBA from a Java programmers perspective. It's called "Java Programming with CORBA", ISBN 0471376817. It has the best description of the POA I've read - I struggled with this particular feature of CORBA before, as I couldn't understand why it was designed the way it is from reading the Henning and Vinoski book. The Henning and Vinoski book is the definitive one for C++ programmers, but it is a pretty dull read even by teh standards of computer textbooks. It isn't helped by the quality of the C++ language mapping, which predates the ANSI C++ standard. The mapping could have really benefited from use of things like the standard string class and containers.
Hennings opinions should be taken with a pinch of salt. While he co-authored the bet known book on CORBA, he has since left the CORBA arena following the merger of his own company with IONA, and is now pushing an alternative technology called ICE.
As for CORBA? The concepts like the POA are too over engineered, but the underlying ideas of language neutrality, a simple interface definition language and a common wire protocol have not been bettered. SOAP has an awful interface definition language based on XML, and is also dependent on XML in many other areas, despite the high overheads of parsing the format. SOAP also suffers from using HTTP as it's primary transport, as this is also highly inefficient. RMI is a great solution for end to end Java systems, but forces a reliance on a proprietary third party bridge if it needs to be intergrated into a non-Java environment.
One area where CORBA is particularily useful, is in event or notification based systems. These standard services offer the same features as Java JMS, but also benefit from a standard wire protocol and standard mappings for languages other than Java.
Overall, CORBA's demise seems to be greatly exagerated. It's not a fashionable technology, nor it is perfect. However, it provides the glue between many systems in organisations such as banks, warehouses and media organisations that no other technology could adequately fill. A number of companies are making a very tidy living from consulting both for their own CORBA implementations as well as open source ones such as MICO and omniORB.
I can speak from personal experience as someone who has successfully used CORBA both as an RPC mechanism and an events system at my last two employers. Most recently, I used CORBA to replace a SOAP based system. Compared to SOAP, the CORBA based system was faster, easier to work with and far more reliable.
Sun is new to open sourcing its proprietary products? That strange, because amongst other things, they open sourced their implementations of RPC and NFS years ago. Sun are by no means new to this open source thing and as well as their own stuff, they've acted as mentors to a number of outside projects. For instance, Sun provided John Ousterhout with an office to use while he worked on Tcl/Tk.
Alien bacteria in India? Maybe that's why I get the chronic shits everytime I go there.
Hmm, looks like my sarcasm wasn't as obvious as I thought ...
I too love NetBSD, but shipping with both vi and ed is stupid. Personally, I don't think an editor should be included at all, since pkgsrc makes adding one trivial.
Matthias Ettrich created KDE. He did not found TrollTech. The summary is incorrect. That is all.
Swing adopted a lot of ideas from the NeXTstep API. A thorough implementation of the MVC architecture, and extensive use of design patterns. This makes it counter intuitive to programmers used to the Win32 or Motif way of doing things, but once you understand how to create decent data models and separate them from the actual GUI code it leads to more modular, maintainable code.
At work, we have three programmers who routinely program GUI applications in Java. Two use Swing, write well structured code that encourages reuse. They love Swing (except the JTree widget). The other programmer prefers SWT, which he isn't allowed to use, creates spaghetti code monsters and whinges about the Swing API all the time. This isn't an isolated case either, it is amusingly similar to the situation at my previous employer - the substandard programmer there also preferred SWT over Swing.
You forgot to mention the fact that you *still* need the sun jdk to build the package.
About a year ago I did. However, I've boostrapped all subsequent builds from the native one.
I did snip it, but the post I was replying to emphasised a requirement for SuSE packages and Linux compatibility to run Java. I snipped the output from pkg_info because I was interested in showing the availability of a native JDK and the dependencies - X11 and openmotif. As for risk, I use the native JDK in a production environment with Tomcat 5.5.x, and I haven't encountered a single problem.
There's a proper (binary blobless) driver for FreeBSD as well as the wrapper, but it isn't as capable or reliable as the OpenBSD one.
Half a SuSE installation? I haven't even got Linux compatibility compiled into my kernel.
$ uname -a/ i386/compile/LAPTOP i386
NetBSD tietokone.panews.press.net 3.0_STABLE NetBSD 3.0_STABLE (LAPTOP) #0: Fri Apr 14 22:31:45 BST 2006 root@tietokone.panews.press.net:/usr/src/sys/arch
$ pkg_info jdk14
Information for jdk14-1.4.2.8:
Comment:
Java Development Kit 1.4.2
Requires:
openmotif>=2.2.3
xorg-libs>=6.9.0
Description:
This is Sun's Java[tm] Development Kit, version 1.4.2, made buildable
and usable natively on *BSD/i386 by Greg Lewis and a host of others.
I countered an assertion that there's lots of Linux savvy programmers in the world, by noting that in my experience they can't be found in much of the UK. Now, there's a fairly large IT industry in the UK, so the lack of decent programmers either means that the UK is suffering an unusual lack of them, or the original posters statement is bollocks. Given that many continental companies recruit from the UK, that must mean that there's a lack of them all over Europe. Finally, having worked in the States (Silicon Valley), I can state with some certainty that most companies are staffed by bodgers and the clueless.
with the abundance of Linux savy programmers in the world it makes very good sense to use a widely deployed system.
Congratulations. I assume you're not in the UK, as I've found it hard work recruiting programmers who are "savvy" with anything, let alone embedded Linux development.
OK, so you used BSD instead. That's fine -- but from what I've seen the BSDs embedded abilities are far less than Linux. So you probably had to do a lot of work.
Speaking from experience (writing controller software for cranes and the like), NetBSD is much easier to work with than Linux. This is even on hardware like Arcom's Viper, which come with a prepackaged Linux development kit. Linux is too difficult to strip down, and the developer kits that come with many embedded systems are based on very old versions of the kernel and libraries. The age of these kernels and libraries, and the lack of any understanding of what a stable ABI or API is amongst Linux hackers, means that getting upto date third party code to work is a nightmare. With NetBSD (and FreeBSD, not so sure about OpenBSD) backwards compatability is excellent.
Don't forget the GNU C Library.
Thankfully I almost have. Glibc is written in what I describe as a "very baroque" way. For instance, large parts of the code don't actually look much like C thanks to the over use of the C preprocessor. If you compare it with the equivalent parts of the C libraries in any of the BSD's, you can see how the rigid application of things like the GNU coding standards and certain peoples distinct coding practices have resulted in glibc being overly elaborate and large.
We owe RMS a huge debt because he single-handedly kickstarted the free software movement.
No, he kickstarted the GNU movement - the free software movement already existed, most notably in the form of BSD Unix. Like GNU, the BSD Unix codebase has gradually evelolved into a largely complete implementation of Unix that is not encumbered by an AT&T license. What Stallman did differently to the people at Berkeley, was to use a license that placed restrictions on the way GNU software is distributed. Stallman sees this as a necessity to ensure GNU software stays "free", by preventing someone from distributing binaries without providing access to the source code.
Most bands end up owing the record company for the priviledge of having their cds distributed by them, which is why I think most bands would be better off without record contracts.
Rubbish, unless they signed to a major in the wake of a bit of bandwagon jumping that is (Coldplay have a hit, every major signs a couple of drippy indie bands on three album deals with an option on the last two). In my own experience of small record labels, they cover your recording costs at the very least. Having recently recorded an album to the same standard at personal expense, I can now see that the outlay for a record label is considerable. Then you consider that majors have to give bands an advance that they can actually live on as well as covering their recording costs.
Having toured as support to major label bands, I can see the difference that it makes when you can spend time doing music for a living rather than a hobby. It's incredibly rare to find a semi-pro or amateur band that can put on a good a show as a band that can afford to spend weeks rehearsing solidly before a tour. The alternative for many people is trying to do music on the dole, which is pretty fucking dismal, or playing up on the non-music side of your "celebrity" - step forward Pete Doherty and countless gun toting rap twats.
Right. So bands should try and cover their costs by selling T-shirts rather than "ephemeral" stuff like CD's of their music. They better add a screen printer to their usual shopping list of ephemeral stuff like amps, guitars and synths then.
Two dogs are sat in Pavlovs parlour. One turns to the other and says "Watch this, if I start salivating he starts ringing a little bell".
"VB, much like generic beer and America's Funniest Home Videos is an enabling technology for stupid people." - Anonymous
For some reason I thought VB stood for Victoria Bitter when I first read your sig. Then I realised you meant Visual Basic. Now I'm not sure again.
When I used to have a 50 mile drive to and from work, the only good thing about it was that I could listen to the Today programme and the 6:00pm news on Radio 4. Quality journalism, which even the fallout from the disgraceful Hutton inquiry hasn't entirely destroyed. On TV the only thing that comes close is Channel 4 news, but even that's been watered down over the last couple of years - much less coverage of international news, and a willingness to cut the original hour long format in half on regular occasions.
I'd read that IBM have a port of the Sun JDK for PowerPC Linux. While GCJ and Classpath are progressing fast, I don't think they're quite ready for all the Swing and Tomcat/JSP based stuff that I'd like to do. The Swing classes at the very least are still a bit patchy. However, both projects have come on in leaps and bounds in the last couple of years, and along with the new Fortran 95 compiler I'm looking forward to the next few releases of GCC4.
NetBSD may be a better choice for older hardware, as it consumes less resources than recent versions of Linux. Note that this is not an anit-Linux troll, much of that extra resource hungriness seems to come from the added functionality rather than superfluous bloat - and despite my personal preference for NetBSD, I'm considering putting FC5 onto my PowerBook, as there appears to be support for Java on PowerPC Linux.