Nope, the rockets (V2s) were travelling much too fast to hear an approach and at first people didn't know what caused the explosion. The Government blamed an early hit up the road from me in Chiswick, west London, on a gas main going up. It didn't take long for people to figure out what was going on and to humorously tag them 'flying gas mains'.
Doodlebugs or buzz-bombs (V1s), were actually much more frightening since their guidance system dependend on the engine cutting out and the missile diving down in silence. Hearing this was a trigger for people to run for cover. Fortunately my mother's family had time to hit the basement when one landed in their back-garden in Essex. (For extra points, WWII buffs can explain the story why these were landing in Essex and not London).
Whoop-de-doo! At last, I can write GNOME apps and make them available for:
1. My distro, to share with all those other SuSE 7.3 on AMD K6-III users
2. Other geeks who carry big enough machines around to compile with
Excuse me while I just schedule a rewrite of my business application for all those GNOME enthusiasts out there waiting to build it!
Of course, the idea of GNOME 2 including a wacky and untried bit of rocket science like a VIRTUAL MACHINE to provide for sensible code distribution would wayyyy too much to deal with. REAL CODERS have no time for toys like IBM's Java SWT for GTK, Miguel's.NET VM clone and 'Parrot' - REAL APPS will only ever be written in C. To all those non-geeks with their wimpy embedded Linux boxes listen: get a clue (preferably a shelf-full) and upgrade, fer chrissakes!
Interesting as these detailed implementation aspects are, I'd like to take a moment to step back and look at the potential of, er, Parrot-like developments.
Linux as a platform won't continue to grow relative to.NET and Java unless it includes a VM. As we know, cross-hardware platform distribution issues are already affecting PPC and ARM users and this will get worse as small non-x86 devices spread.
Linus should therefore make a New Year resolution be to 'anoint' a VM as the future target platform and encourage libraries to be built around it. Parrot could be this VM, but from this point of view Parrot doesn't offer any fundamental technical advantage over the Java or.NET VMs, which brings me on to my next point.
While code distribution requirements alone are sufficient to justify a 'Linux VM', there is another very interesting potential benefit which I haven't seen discussed yet. This applies specifically to Open Source and the principles behind it. This is that a VM could be implemented where the source code and compiled code were semantically equivalent. This means, of course, that all that ever needs to be distributed is the 'compiled' form and, by its very nature, this code is always open.
This is almost the case with Java bytecode, in that decompilers such as JAD can (usually) produce editable code from compiled.class files. However, comments (and layout) are not stored in the bytecode and this is starting to emerge as a significant problem for advanced development tools. For a brief illustration, consider a problem with IBM's new Eclipse IDE. Eclipse extracts Javadoc-style comments from source code that describe methods, parameters and other API aspects. It makes these available as tool-tip-like hints when you start to code a call to a particular method. However, commercial code is not distributed as source but as class files with accompanying Javadoc-generated HTML files. Although, between the two, all the information that Eclipse needs is there in principle, in practice it is hard for it to put the pieces back together and use them.
Perhaps a better example of sourcecode equivalence goes back to the old days of home computers with interpreted BASIC. Most people will remember that BASIC code was 'tokenized', including keywords, blank lines, punctuation and comments, in a way that could be both edited and executed. Unfortunately, the spread of this useful model was halted when PCs became big enough for 'real' compilers to run.
What's needed, therefore, is a more sophisticated version of the old tokenized program representation, and such systems have been developed for Scheme (and FORTH, I think), usually known as Abstract Syntax Tree interpreters, in contrast to bytecode interpreters like Parrot's current implementation. There may be some middle ground here - I have a few pointers to research in this area, such as Anton Ertl's work.
Now, having solved Linux's portability problem and given a big boost to open source, we could perhaps rest there with some satisfaction. However, I'd argue that there's one more important requirement to address and this time it will be one that is familiar to programmers already, if mostly only those working with languages like LISP and Scheme. This is the provision of dynamic reflective capabilities, meaning the ability to treat programs as data.
They say that any large C application includes an interpreter for a higher-level language. In finance, we see applications supporting the entry of complex formulas; in CAD/CAM people build parts which are themselves programmable such as a parameterizable aircraft undercarriage. These features turn users into programmers, but programmers working with a high-level, specialized language.
LISP is a much better language to build such applications (which is why a lot of AutoCAD is in LISP) since LISP programs are themselves data structures that can be added and played with at run-time. Java,.NET, Perl 5 and Python have some limited features in this area, mostly to do with loading code dynamically and supporting development tools, but they are awkward and ugly in the extreme. There is demonstrably huge potential for a mainstream language with this level of flexibility - any sophisticated business will be able to come up with examples like those above.
So, to wrap up, it looks like others have left the goalposts wide open for us and there is a significant opportunity to exploit. It will be interesting to see whether this Parrot can do more than mimic other platforms.
Most of the GOF patterns are specific to a class of languages, in fact they go further and embody implementation assumptions such as local vs. distributed, transient vs. persistent etc.
Take a look at how the estimable Peter Norvig achieves the aims of many of the patterns in LISP: Design Patterns in Dynamic Programming. Are these LISP implementations of the DP examples?
Merci pour le feedback (how's my French? I'm going to Calais tomorrow... think I might be struggling!)
Unfortunately, Slashdot had a spasm and I couldn't post my earlier response. Here's another attempt:
Persistence: Not sure about memory-mapped files - these lack transactions, queries and those other good 'enterprise' features. I think you're closer when you mention meta-programming further down.
Long-lived processes: I checked your link, but from what I could understand this seemed to be more about passing objects between threads. Anyway, the idea is that the process state is preserved without the kind of explicit programming involved in a workflow system - 'business process' and program process are congruent.
Dynamic / user programming: Yes, I agree LISP-like languages are attractive for this reason. Python and Java can do some dynamic and reflective stuff, but it's much more awkward than in some LISP-derived systems.
OCAML: I've heard really wonderful things about this, but people tend to emphasise the type-safety and functional programming aspects. I'd certainly like to check out the meta-programming-like features to see how far it could be turned into 'enterprise programming nirvana'!
cheers
alex
Next time include the requirements
on
Lightweight Languages
·
· Score: 2, Interesting
Yes, great idea, but one could argue that both academics and the worthy hackers are somewhat removed from typical user contexts. There are plenty of requirements in the 'enterprise applications' space (not to mention the scientific modelling space etc.) that are virtually never addressed in basic programming language, but only in endless 'frameworks', 'wizards' and other add-ons that are really band-aids for concepts not considered at language design-time. For example:
Commercial information tends to be persistent, not transitory. A good language should work directly with stored data.
Processes in organizations are long-lived and distributed, whereas typical programming languages just deal with transient threads etc. (outside workflow systems such as WebLogic Integration).
Programs represent rules, algorithms and other forms of knowledge that end-users will want to add to (e.g. a discount formula). Not only should the environment allow run-time modification and extension, it should also support representations and syntaxes accessible by non-programmers.
Every action has a principal actor associated with it, and typical commercial environments need to record who it was for auditing and access control purposes. If a programming language has no concept of Principal, one has to be stuck awkwardly on the side (e.g. Java EJB isCallerInRole).
Transactions are a very common programming model. At the very least, there should be support for creating and propagating transaction IDs, restarting procedures etc.
What else? Run-time metrics, versioning, SQL-style set predicates... well, you get the idea. People have to wake up to the fact that there is still a huge disconnect here.
(Amazing to think that Java gave Microsoft some ideas and a wide-open goal, and they came up with C#).
Just as I was reading this thread, Richard Noble (the guy who took the land speed record a while ago) was on a radio phone-in over here telling people about his plan for independent travel using his Farnborough F1 plane. (Hmmm, hope the plane goes faster than Richard's web site...)
He says:
"...enter the Farnborough F1 air taxi, which flies point to point faster than a congested airliner and the Farnborough integrated Ops system, which will enable you to book your on demand travel zipcode to zipcode off the web. This means you can be picked up by a ground cab from your office or home, meet up with the F1 at a local airfield and arrive at your postcode destination 1000 miles away in under 4 hours door to door. That's about half the airline time and the best bit is yet to come. The whole activity is low stress with costs comparible to a business class airline fare and you need never go to a major airport again for short-haul travel!
Key to all this is the importance of avoiding self deception. Constantly we check each other out - Have we got this right? Is there some fatal flaw in the project which means that it can never succeed? So far the only real problem is the very difficult finance - the rest is do-able. We can also take great confidence in the fact that NASA has come up with much the same ideas, though with different emphasis for their SATS (Small Aircraft Transportation System programme) which they believe will treble airspace travel capacity."
Whether well-founded or not in general, the point is that the post does lack any rationale. That's the problem with it. This should be obvious to anyone other than the person that modded my comment down.
Now to address the specific points you have (now) raised:
Support for generics has been in development for a while and you can get the prototype from the JDC site.
Assertions are available now in the 1.4 VM, and are intended for the DBC features you describe.
If you prefer a more native style GUI library, you now have an alternative in the form of the new IBM SWT, which uses Windows elements, like the old AWT, as I and others have pointed out no less than three times here in this thread.
I hope people find this information of use, but based on current trends I suspect they will have to browse at -1 to find it.
Is it too much to ask moderators to check for coherence and at least a semblance of a rationale for a given argument? This is just a baseless rant - let's differentiate/. from Usenet. Please.
A completely impractical assumption (and, incidentally, one that is spectacularly incompatible with Open Source, at least, open source written in C).
When was the last time you used a kernel that was really monolithic, one that had been built, supplied and tested as a unit, by a common engineering team? The fact is that all modern systems are supplied in untestably complex configurations, which if reliability is not to be compromised, must be able to protect themselves from problem components.
If the design choice was really between copying memory and passing pointers that allowed the receiver to stamp all over the sender's address space then life would be rather depressing. However, in the absence of hardware features like capabilities and Multics-style protection levels, there is a solution in the form of a safe, intermediate language such as Java bytecode. This way, you only have to trust your VM/JIT compiler for basic address-space integrity.
Slow? Well, device drivers probably shouldn't be the first part of Linux to be bullet-proofed in this way, but for serious components (think KDE applications currently using DCOP etc.) the VM can easily outperform native code, because it can optimize the execution path *across* separately loaded components, and eliminate null procedures such as unused access checks, RPCs for local objects etc.
Linux (and Linus, by the sound of it) need to wake up to the power of VMs. MS apps will soon no longer be tied to x86, Java is still growing, while efforts that could be used for Linux (Perl/Python and a few LISP engines) are niche environments, to say the least.
Anybody that believes Linux is still going to grow when it possesses zero inbuilt protection and requires apps to be manually recompiled for every platform variant is living in cloud-cuckoo land.
Did Baran and Davies' work matter to the ARPANET? It pretty much has to have.
I think so - I read that Davies and a colleague visited Baran and worked with him for a while. (I live near the NPL in Teddington, west London and so can point out to fellow geeks that that's where packets come from. But more importantly, it's the home of the Benny Hill show! (Teddington Studios)).
Yes, let's be fair - a mere 13 years or so after Windows first appeared to get mechanisms like a stable, cross-language API and distributed communications sorted out is not unreasonable. After all, the fact that other operating systems (Apollo Domain, Stratus VOS) had these features around 1987 can't have been expected to influence MS - how many MS programmers ever encountered real "enterprise applications"?:-)
Why should you want to generate libraries / executables? This can restrict the number of optimizations made significantly.
There's no fundamental reason to run each Java program in a separate VM either - see Echidna for a current mechanism to let them share. My understanding is that Sun were working on a more flexible task / VM / processor mapping capability to exploit MP servers better, but that was a while ago.
And it was such a nice little place when I was a kid... I can believe it, particularly the burglary, though to be fair the murder rate should be a fraction of an equivalent town in the US - a search of the local paper Get Reading yields just the one stiff found in recent months, the one I think you're referring to.
Thought about moving out of the town centre? You're probably in the worst place in the whole county...
cheers
alex
(Richmond, Surrey - CCTV capital of West London)
Mea culpa! Please note that the pattern I described *only* applies to messages of over 36 bytes, and frequencies exceeding a bushel per fortnight. Do not attempt to employ this paradigm outside this range, or you may invalidate the warranty.
I find message queues (MQ Series, JMS etc.) suck in the same way. At first it sounds great - decoupled, guaranteed delivery, publish/subscribe and so on - then you realize that these 'features' have horrendous drawbacks and a polling solution could have been written in 1/3 the time and eliminated swathes of management problems.
Of course, you can add an asynchronous 'notification' event on top of this structure if polling is expensive, but this is just a hint to go looking for a message. (Another lesson from Multics, I believe - at least in its cousin VOS the s$wait_event call works like this).
Try subscriptions@private-eye.co.uk, but be warned that the content is not only very UK-centric, but includes its own bizarre slang, partly to get around local libel laws, e.g. Ugandan Discussions = sex.
There used to be a clone in the US called SPY, right down to the 'separated at birth' pictures of unlikely twins, but I think it disappeared.
Another dodgy quote:
There was thought to be little hope today of any informed coverage emerging from the thousands of tons of newsprint which left readers buried in voyeuristic cliche and bellicose sentimentality last week.
Still, it's an institution, and one which upholds fine British traditions of
cont. p94
Hmmm. Married in haste, we can now repent at leisure.
There probably are some fundamental limitations of the JVM - I expect talking to LISPers, embedded database people or workflow app developers, for example, would quickly elucidate a good set of requirements for an improved VM - but the CLR is such a transparent knock-off that it doesn't add anything significant itself, essentially MS seem to be relying just on better integrated tools to compete.
What there might be in this struggle that could benefit Linux developers remains a mystery to me.
Yes, The Guardian can be seen as a bit left of the rest of the broadsheets over here, but that's because it provides a platform for contrarian views as well as mainstream Blairite opinion. If you go by its editorials, you'll find it backing an armed response - Observer editorial (sister edition) As to its record, I'd say it and The Observer probably has the best record of any paper here, both in recent years in exposing dodgy government and going way back to its attitude to Nazism, colonialism etc.
So it's easy caricature, like this bit in the satirical rag Private Eye, but overall it is not much different from the NY Times, from which it has been reprinting columns verbatim this week. (Alan Rusbridger is The Guardian's editor):
Extremists Seen Dancing And Cheering In Guardian Fanatical anti-American supporters of the sinister leader Al-Rubbisha were seen openly celebrating in the offices of the Guardian newspaper at the news that the WTC had collapsed. Rubbisha was quick to deny the charge, saying "Only a proportion of my columnists, i.e. 80%, were involved."
(For those offended by the Eye's attempts at humour about the disaster, the magazine helpfully included a pre-printed response slip for cancelling subscriptions).
RT is not a niche - audio and video is mainstream, therefore a proper RT OS *should* be mainstream too. The fact that there isn't one is a perfect illustration of a highly distorted IT marketplace.
Slashdot Mirror
Nope, the rockets (V2s) were travelling much too fast to hear an approach and at first people didn't know what caused the explosion. The Government blamed an early hit up the road from me in Chiswick, west London, on a gas main going up. It didn't take long for people to figure out what was going on and to humorously tag them 'flying gas mains'.
Doodlebugs or buzz-bombs (V1s), were actually much more frightening since their guidance system dependend on the engine cutting out and the missile diving down in silence. Hearing this was a trigger for people to run for cover. Fortunately my mother's family had time to hit the basement when one landed in their back-garden in Essex. (For extra points, WWII buffs can explain the story why these were landing in Essex and not London).
Whoop-de-doo! At last, I can write GNOME apps and make them available for:
.NET VM clone and 'Parrot' - REAL APPS will only ever be written in C. To all those non-geeks with their wimpy embedded Linux boxes listen: get a clue (preferably a shelf-full) and upgrade, fer chrissakes!
1. My distro, to share with all those other SuSE 7.3 on AMD K6-III users
2. Other geeks who carry big enough machines around to compile with
Excuse me while I just schedule a rewrite of my business application for all those GNOME enthusiasts out there waiting to build it!
Of course, the idea of GNOME 2 including a wacky and untried bit of rocket science like a VIRTUAL MACHINE to provide for sensible code distribution would wayyyy too much to deal with. REAL CODERS have no time for toys like IBM's Java SWT for GTK, Miguel's
Interesting as these detailed implementation aspects are, I'd like to take a moment to step back and look at the potential of, er, Parrot-like developments.
.NET and Java unless it includes a VM. As we know, cross-hardware platform distribution issues are already affecting PPC and ARM users and this will get worse as small non-x86 devices spread.
.NET VMs, which brings me on to my next point.
.class files. However, comments (and layout) are not stored in the bytecode and this is starting to emerge as a significant problem for advanced development tools. For a brief illustration, consider a problem with IBM's new Eclipse IDE. Eclipse extracts Javadoc-style comments from source code that describe methods, parameters and other API aspects. It makes these available as tool-tip-like hints when you start to code a call to a particular method. However, commercial code is not distributed as source but as class files with accompanying Javadoc-generated HTML files. Although, between the two, all the information that Eclipse needs is there in principle, in practice it is hard for it to put the pieces back together and use them.
.NET, Perl 5 and Python have some limited features in this area, mostly to do with loading code dynamically and supporting development tools, but they are awkward and ugly in the extreme. There is demonstrably huge potential for a mainstream language with this level of flexibility - any sophisticated business will be able to come up with examples like those above.
Linux as a platform won't continue to grow relative to
Linus should therefore make a New Year resolution be to 'anoint' a VM as the future target platform and encourage libraries to be built around it. Parrot could be this VM, but from this point of view Parrot doesn't offer any fundamental technical advantage over the Java or
While code distribution requirements alone are sufficient to justify a 'Linux VM', there is another very interesting potential benefit which I haven't seen discussed yet. This applies specifically to Open Source and the principles behind it. This is that a VM could be implemented where the source code and compiled code were semantically equivalent. This means, of course, that all that ever needs to be distributed is the 'compiled' form and, by its very nature, this code is always open.
This is almost the case with Java bytecode, in that decompilers such as JAD can (usually) produce editable code from compiled
Perhaps a better example of sourcecode equivalence goes back to the old days of home computers with interpreted BASIC. Most people will remember that BASIC code was 'tokenized', including keywords, blank lines, punctuation and comments, in a way that could be both edited and executed. Unfortunately, the spread of this useful model was halted when PCs became big enough for 'real' compilers to run.
What's needed, therefore, is a more sophisticated version of the old tokenized program representation, and such systems have been developed for Scheme (and FORTH, I think), usually known as Abstract Syntax Tree interpreters, in contrast to bytecode interpreters like Parrot's current implementation. There may be some middle ground here - I have a few pointers to research in this area, such as Anton Ertl's work.
Now, having solved Linux's portability problem and given a big boost to open source, we could perhaps rest there with some satisfaction. However, I'd argue that there's one more important requirement to address and this time it will be one that is familiar to programmers already, if mostly only those working with languages like LISP and Scheme. This is the provision of dynamic reflective capabilities, meaning the ability to treat programs as data.
They say that any large C application includes an interpreter for a higher-level language. In finance, we see applications supporting the entry of complex formulas; in CAD/CAM people build parts which are themselves programmable such as a parameterizable aircraft undercarriage. These features turn users into programmers, but programmers working with a high-level, specialized language.
LISP is a much better language to build such applications (which is why a lot of AutoCAD is in LISP) since LISP programs are themselves data structures that can be added and played with at run-time. Java,
So, to wrap up, it looks like others have left the goalposts wide open for us and there is a significant opportunity to exploit. It will be interesting to see whether this Parrot can do more than mimic other platforms.
Most of the GOF patterns are specific to a class of languages, in fact they go further and embody implementation assumptions such as local vs. distributed, transient vs. persistent etc.
Take a look at how the estimable Peter Norvig achieves the aims of many of the patterns in LISP: Design Patterns in Dynamic Programming. Are these LISP implementations of the DP examples?
Merci pour le feedback (how's my French? I'm going to Calais tomorrow... think I might be struggling!)
Unfortunately, Slashdot had a spasm and I couldn't post my earlier response. Here's another attempt:
Persistence: Not sure about memory-mapped files - these lack transactions, queries and those other good 'enterprise' features. I think you're closer when you mention meta-programming further down.
Long-lived processes: I checked your link, but from what I could understand this seemed to be more about passing objects between threads. Anyway, the idea is that the process state is preserved without the kind of explicit programming involved in a workflow system - 'business process' and program process are congruent.
Dynamic / user programming: Yes, I agree LISP-like languages are attractive for this reason. Python and Java can do some dynamic and reflective stuff, but it's much more awkward than in some LISP-derived systems.
OCAML: I've heard really wonderful things about this, but people tend to emphasise the type-safety and functional programming aspects. I'd certainly like to check out the meta-programming-like features to see how far it could be turned into 'enterprise programming nirvana'!
cheers
alex
Commercial information tends to be persistent, not transitory. A good language should work directly with stored data.
Processes in organizations are long-lived and distributed, whereas typical programming languages just deal with transient threads etc. (outside workflow systems such as WebLogic Integration).
Programs represent rules, algorithms and other forms of knowledge that end-users will want to add to (e.g. a discount formula). Not only should the environment allow run-time modification and extension, it should also support representations and syntaxes accessible by non-programmers.
Every action has a principal actor associated with it, and typical commercial environments need to record who it was for auditing and access control purposes. If a programming language has no concept of Principal, one has to be stuck awkwardly on the side (e.g. Java EJB isCallerInRole).
Transactions are a very common programming model. At the very least, there should be support for creating and propagating transaction IDs, restarting procedures etc.
What else? Run-time metrics, versioning, SQL-style set predicates... well, you get the idea. People have to wake up to the fact that there is still a huge disconnect here.
(Amazing to think that Java gave Microsoft some ideas and a wide-open goal, and they came up with C#).
Just as I was reading this thread, Richard Noble (the guy who took the land speed record a while ago) was on a radio phone-in over here telling people about his plan for independent travel using his Farnborough F1 plane. (Hmmm, hope the plane goes faster than Richard's web site...)
He says:
"...enter the Farnborough F1 air taxi, which flies point to point faster than a congested airliner and the Farnborough integrated Ops system, which will enable you to book your on demand travel zipcode to zipcode off the web. This means you can be picked up by a ground cab from your office or home, meet up with the F1 at a local airfield and arrive at your postcode destination 1000 miles away in under 4 hours door to door. That's about half the airline time and the best bit is yet to come. The whole activity is low stress with costs comparible to a business class airline fare and you need never go to a major airport again for short-haul travel!
Key to all this is the importance of avoiding self deception. Constantly we check each other out - Have we got this right? Is there some fatal flaw in the project which means that it can never succeed? So far the only real problem is the very difficult finance - the rest is do-able. We can also take great confidence in the fact that NASA has come up with much the same ideas, though with different emphasis for their SATS (Small Aircraft Transportation System programme) which they believe will treble airspace travel capacity."
Gotta love these plucky inventor types!
Whether well-founded or not in general, the point is that the post does lack any rationale. That's the problem with it. This should be obvious to anyone other than the person that modded my comment down.
Now to address the specific points you have (now) raised:
Support for generics has been in development for a while and you can get the prototype from the JDC site.
Assertions are available now in the 1.4 VM, and are intended for the DBC features you describe.
If you prefer a more native style GUI library, you now have an alternative in the form of the new IBM SWT, which uses Windows elements, like the old AWT, as I and others have pointed out no less than three times here in this thread.
I hope people find this information of use, but based on current trends I suspect they will have to browse at -1 to find it.
Is it too much to ask moderators to check for coherence and at least a semblance of a rationale for a given argument? This is just a baseless rant - let's differentiate /. from Usenet. Please.
Two more and I think we've got a record...
Where? I've only seen the C# language spec. being offered to ECMA, and the Mono people have explicitly ruled out implementing the libraries/services.
But, exasperated by constant confusion in the financial pages, the UK fell in with American practice, so now it's 10e9 in current usage.
There's a similar problem with DDMMYY dates, but in this case we're sticking with the continentals. I think.
A completely impractical assumption (and, incidentally, one that is spectacularly incompatible with Open Source, at least, open source written in C).
When was the last time you used a kernel that was really monolithic, one that had been built, supplied and tested as a unit, by a common engineering team? The fact is that all modern systems are supplied in untestably complex configurations, which if reliability is not to be compromised, must be able to protect themselves from problem components.
If the design choice was really between copying memory and passing pointers that allowed the receiver to stamp all over the sender's address space then life would be rather depressing. However, in the absence of hardware features like capabilities and Multics-style protection levels, there is a solution in the form of a safe, intermediate language such as Java bytecode. This way, you only have to trust your VM/JIT compiler for basic address-space integrity.
Slow? Well, device drivers probably shouldn't be the first part of Linux to be bullet-proofed in this way, but for serious components (think KDE applications currently using DCOP etc.) the VM can easily outperform native code, because it can optimize the execution path *across* separately loaded components, and eliminate null procedures such as unused access checks, RPCs for local objects etc.
Linux (and Linus, by the sound of it) need to wake up to the power of VMs. MS apps will soon no longer be tied to x86, Java is still growing, while efforts that could be used for Linux (Perl/Python and a few LISP engines) are niche environments, to say the least.
Anybody that believes Linux is still going to grow when it possesses zero inbuilt protection and requires apps to be manually recompiled for every platform variant is living in cloud-cuckoo land.
--
alex
Most reasonable...
Did Baran and Davies' work matter to the ARPANET? It pretty much has to have.
I think so - I read that Davies and a colleague visited Baran and worked with him for a while. (I live near the NPL in Teddington, west London and so can point out to fellow geeks that that's where packets come from. But more importantly, it's the home of the Benny Hill show! (Teddington Studios)).
He (or she) is correct - Bynari is positioned as an Exchange equivalent - please mod up.
cheers
alex
Yes, let's be fair - a mere 13 years or so after Windows first appeared to get mechanisms like a stable, cross-language API and distributed communications sorted out is not unreasonable. After all, the fact that other operating systems (Apollo Domain, Stratus VOS) had these features around 1987 can't have been expected to influence MS - how many MS programmers ever encountered real "enterprise applications"? :-)
Why should you want to generate libraries / executables? This can restrict the number of optimizations made significantly.
There's no fundamental reason to run each Java program in a separate VM either - see Echidna for a current mechanism to let them share. My understanding is that Sun were working on a more flexible task / VM / processor mapping capability to exploit MP servers better, but that was a while ago.
--
cheers
alex
And it was such a nice little place when I was a kid... I can believe it, particularly the burglary, though to be fair the murder rate should be a fraction of an equivalent town in the US - a search of the local paper Get Reading yields just the one stiff found in recent months, the one I think you're referring to.
Thought about moving out of the town centre? You're probably in the worst place in the whole county...
cheers
alex
(Richmond, Surrey - CCTV capital of West London)
Mea culpa! Please note that the pattern I described *only* applies to messages of over 36 bytes, and frequencies exceeding a bushel per fortnight. Do not attempt to employ this paradigm outside this range, or you may invalidate the warranty.
Thank you.
--
I find message queues (MQ Series, JMS etc.) suck in the same way. At first it sounds great - decoupled, guaranteed delivery, publish/subscribe and so on - then you realize that these 'features' have horrendous drawbacks and a polling solution could have been written in 1/3 the time and eliminated swathes of management problems.
Of course, you can add an asynchronous 'notification' event on top of this structure if polling is expensive, but this is just a hint to go looking for a message. (Another lesson from Multics, I believe - at least in its cousin VOS the s$wait_event call works like this).
--
There used to be a clone in the US called SPY, right down to the 'separated at birth' pictures of unlikely twins, but I think it disappeared.
Another dodgy quote:
There was thought to be little hope today of any informed coverage emerging from the thousands of tons of newsprint which left readers buried in voyeuristic cliche and bellicose sentimentality last week.
Still, it's an institution, and one which upholds fine British traditions of cont. p94
Hmmm. Married in haste, we can now repent at leisure.
There probably are some fundamental limitations of the JVM - I expect talking to LISPers, embedded database people or workflow app developers, for example, would quickly elucidate a good set of requirements for an improved VM - but the CLR is such a transparent knock-off that it doesn't add anything significant itself, essentially MS seem to be relying just on better integrated tools to compete.
What there might be in this struggle that could benefit Linux developers remains a mystery to me.
--
As to its record, I'd say it and The Observer probably has the best record of any paper here, both in recent years in exposing dodgy government and going way back to its attitude to Nazism, colonialism etc.
So it's easy caricature, like this bit in the satirical rag Private Eye, but overall it is not much different from the NY Times, from which it has been reprinting columns verbatim this week. (Alan Rusbridger is The Guardian's editor):
Extremists Seen Dancing And Cheering In Guardian
Fanatical anti-American supporters of the sinister leader Al-Rubbisha were seen openly celebrating in the offices of the Guardian newspaper at the news that the WTC had collapsed. Rubbisha was quick to deny the charge, saying "Only a proportion of my columnists, i.e. 80%, were involved."
(For those offended by the Eye's attempts at humour about the disaster, the magazine helpfully included a pre-printed response slip for cancelling subscriptions).
--
ROFL! Thanks Dwonis, I guess your support has been rewarded. Do you want to set one up for me like that? :)
RT is not a niche - audio and video is mainstream, therefore a proper RT OS *should* be mainstream too. The fact that there isn't one is a perfect illustration of a highly distorted IT marketplace.