...and by reading the description of the code, it was obvious that there were some issues in it ("As I alluded to, all three arguments are highly dynamic and constantly updated within the while() loop. There is a great deal of pointer arithmetic in this loop.")
As a general rule of the thumb, if some code does pointer arithmetic in loops with a lot of conditional tests and pointers assignment, it is probably buggy.
I know that this it seems presumptuous, but, over the years, I developed a good sense of looking at a block of C looping code, and saying if it is buggy based on the feel of the complexity in it, the presence of constants, etc, etc. If you can't "see" the invariants in the code by looking at it, it should raise a red flag.
There is that famous Kernighan saying:
"Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it."
Code should be dumb. In that case, they probably could have written slower but dumber code. Maybe by first testing if the path needs rewriting (most path don't), then use a slower/dumber rewriting loop, so the penalty would be paid only by the use of "." and "..". We are in 2008. Complex code is a liability. And vista is already slow, and that is not due to performance of path rewriting routines...
First, the wealthy do consume much more government spending than the poor. In a direct way, through the various pork techniques, indirect ways via use of the infrastructure or with specific laws (from IP protection, tariff, regulations, and a miriad of other ways -- like getting huge bonuses in financial companies, crashing them and letting the taxpayers with the bill).
Also, it can be argued that the wealthy do not even have to consume government resources to put a large burden on society as a whole. They use more land, consume more non-renewable resources, etc, etc.
Anyway, the grand-parent said that "Suppose that every day, ten men go out for beer and the bill for all ten comes to $100.". I point that this is definitely not the way to society is working, hence his whole analogy is flawed.
You call that "exactly on the money with verifiable data". Let me laugh.
In the second half of your post, you built such a nice strawman that I am not even going to argue with. Just read my post back, and tell me if the poor's beer is the same size as the rich's beer...
What a load of crap. If you want to do an analogy, do it right:
Suppose that every day, ten men go out for beer and get the following (numbers not accurate, but I don't pretend to have a PhD in economy):
The first man get no beer The second man get 1 beer The third man get 1 beer The fourth man get 1 beer The fifth man get 2 The sixth would get 3. and so one until: The tenth man (the greediest) get 50 beers.
> No software should cost more than $10k Exception: scientific software (and I mean the really advanced stuff, simulations, etc), math analysis, etc, etc AND EVEN THEN
Obviously, you work in the scientific sector, and also have a pretty narrow-minded view.
Having written other kind of software, I can tell you that you the complexity involved in "enterprisey software" can be mind boggling. For instance, think about reading, implementing and deploying some of the complex financial rules (we're talking about thousands upon thousands pages of regulations, with variations per countries) out there, in a market composed of a handful of customers.
In that case, the cost of the software is directly related to the cost of building it (ie: how much it would cost for the customer to build his own), and such customers are perfectly happy to pay millions (yes, millions) so a software vendors take care of all the issues related to said software.
For instance, as developers are also users, one can argue that the GPL gives freedom to developers that the BSD license doesn't. For instance, if you work on a BSD project, a company can come in, buy the main contributors of the code, close it, and prevent you, as one of the original contributors, of enjoying the contribution of that key members, effectively killing the project.
As a developer, I generally seek for BSD libraries and GPL tools. The first guarantees that I can mix-and-match them to whatever I want, the second that they will be forever available.
It is extremely easy to end up being paid roughly 0$ of royalties because the product ends up being sold as some sort of "add on" module to some other product, at a roughly zero price.
There are a 2 main clauses that must be present:
1) Be sure that the contract have a termination date. It saved me once. Take a date in the future, like 3 to 5 years, to say that the whole arrangement will disappear. It may seem a longtime, but you DO NOT WANT to have ANY liability following you for the rest of your life. Don't trust lawyers on that. You want, written black on white, in plain english, as the very first section of the contract, that all your contractual obligations vanishes at a certain date. Require that on every contract. The additional upside, is that, if they are happy with you, they'll have to renegociate in a few years (but it is not the main point. The main point, is that, if everything goes bad, you want to be able to say "this date is away: I have nothing to do with those guys anymore")
2) Ask money for non-competition. Each month, you have to be paid xxx$, or the non compete clause is waived. If they fire you and want you to continue to non-compete, that is the amount they have to pay you. That amount should be roughly equal to your salary. There should be no reference to the way they fire you (they can sell out, be bought, or whatever). Just say that, if they don't pay you that specific amount, the next month, you are definitely freed from the non-compete clause. This will also have the added value of them having to get back to you if they want to merge the company with another one.
When those two things are present, then you can start talking about the details. But, if they refuse any of those two clauses walk... no... run away.
Thanks for clearing the thing. I was wondering about what Sophie was. It sortof reminded me of Alice. Then I saw your post, and the familiarities are astounding.
* Named from a feminine first name from children books. Check * Authoring System. Check. * Written in Smalltalk. Check. * That are going to be rewritten in Java. Check.
Conclusion, Sophie will follow Alice on the path of irrelevance. Rewriting a Smalltalk app in Java swing is an exercice in futility. It will be a slow bugged mess, which is what Alice is now.
Of course, it will also suffer the second syndrom system, ie: they will try to cram as much as new ideas as possible in that version, and will fail (don't think that because the proposal says "the very same thing" that it will. No. This is not how projects work).
Frankly, I don't know how Rybka would rate against Anand in a real match, and I don't think we'll ever know, because FIDE doesn't want it to happend (a bit like what occured with checkers -- a fascinating story http://en.wikipedia.org/wiki/Marion_Tinsley ). Furthermore, computers and human have such different ways of playing that I am not sure that elo ratings between computers and humans is a correct statistical measure (because you can win against Kramnik, as he may blunder, like in the Deep Fritz game where he blundered a mate in one -- but you can't expect any blunder from a computer, short of a software bug.)
Personally, I would love to think that human have some chance against computers, but I don't think so anymore. Beating a computer at chess is an exercise in futility, like outrunning a car, or generating machine code faster than a compiler...
There are several commercial use for better engines:
1) Game analysis. When you have played a game against and lost, you try to understand why you lost. Sometimes it is because you made an obvious blunder, but when you get better at the game, you start loosing for strategical reasons (lost control of a certain square, etc, etc). Having a good engine helps you try new ideas, and play a lot of what-if scenarios
2) Game understanding. When you follow a live tournament between grandmasters, having a good engine can give you an explanation about what the underlying ideas are ("Why doesn't he plays Nb6? You try it, and get the answer instantly")
3) Correspondence & Centaur Chess. Correspondence chess are long running games where both players have access to whatever they want. It delivers very subtle games, where the strategy is a very important aspect, as all the tactical blunders are removed by the use of good chess engines. Centaur chess is the same with lower time control.
And, of course, bragging rights are important too: having a better engine than other people in the chess club is a bit like having the better graphic card among fps players...
As you may have seen, playing against the engine is not one of the uses. Rybka is supposedly at 3200 elo. By definition, 200 elo points higher means you have a 75% win probability. The current world champion is at 2800, which means that he have a 6% win probability against rybka. Good club chess play is around 2000 (it takes several years to reach that level -- at that level, you can generally play blind, or multiple opponents, etc, etc). Such players have a 1 against 4000 chance against a 3200 player. Which means zero chance...
1) You need opening preparation to play. Well, that's a given, and it was already the case 20 years ago. Now, with computers, opening preparation is easier, so players are better prepared.
2) Grandmasters have not memorized ending tablebases, first because it is impossible, and second, because grandmasters are still much much better than computers in endgames. Endgame is not about memorisation, but about technique. And yes, you have to work that too.
3) Average chess games is 40 moves long, not 60
4) Computers make opening books deeper but also wider: lines that were considered unplayable are now playable. This adds diversity to the games.
5) Middlegame last anywhere from 0 to 20 moves. Not "1 to 5".
6) Capablanca already said that chess was so analysed that it was boring and draw in 1930. Guess what: he was wrong.
So, what is exactly your complain ? That you have to work a lot to play competitively at your local chess club ? Well, that is not because of computers, it is because the players there are better than you are.
And that is not due to computer play. It is due to the internet. Because on the chess servers, you have people playing chess as if it was WoW. So, the guy you face at that local chess club, maybe he plays 40 "3+0" games a day...
> These days it seems all but the most classified tech makes it into market very, very quickly.
No, you didn't got it right. The article meant that, in 10 years, everybody will be enrolled in the military due to the global draft to fight in the war after the economic collapse...
You are not the typical consumer. Your are a technology enthusiast. People like you are happy with BR, but you are not the main consumer market.
I told that over and over again to my HD fanatics friends:
1/ DVD is good enough. 2/ The competition to DVD is DiVX, proving that DVD is indeed good enough. 3/ Movie fanatics already have all their movies in DVD format. Only fans that don't have much DVDs (ie: that are young) may want to go BR for their entire collection. 4/ BR and DVDs are not compatible 5/ Difference between VHS and DVD was huge (no tear, instant access to scenes, better quality, can be played on computers). Differences between DVD and BR are comparatively very small.
Getting BR to the mass of consumer will be a very very costly exercise in futility.
Killer feature for the next format is being download-friendly. Want to see a movie ? Click on the right web site, see all movies ever done, enter a small amount of money, get the movie permanently on your devices (yes, permanently and with a 's', because consumers will not give away what they already get with DVDs). Of course, there are already plenty of formats to can do that, but the one that'll get ubiquitous will win. I hope it'll not gonna be flash:-)
> The EULA for Firefox deals with the use of the trademark "Firefox," only. In order to keep the trademark, they have to enforce it. If the EULA didn't exist, Mozilla Corporation could lose the sole right to call a browser "Firefox."
That is bullshit. Software were trademarked before EULAs existed, and they never once lost trademark because of the lack of EULA. A simple short splashscreen with Firefox(tm) is more than enough for trademark.
> I can count on one hand the number of times I've had a problem with Firefox that would have been solved by it being in its own process.
I restart firefox roughly twice per hour when developing my javascript application. Having 10 concurrent tabs executing heavy javascript/ajax generally hangs the browser.
Of course, extenions (in particular FireBug) are probably responsible of that, and it is painful but not a showstopper. A process per tab model would probably be better for my usage...
> Whenever a game is coming out that I'm interested in, if it's laden with such anti-consumer machinations, I intend to e-mail the publisher explaining that I would rather pirate it than pay them money thanks to their greed.
Wtf ? Skip it. Don't pirate it. Use the money to buy a different game, with no DRM, and tell the publisher/developer of *that* game why you bought it (be specific).
The problem with your way is that you encourage their behavior. I don't understand why you accept to jump through hoops just so you can send them 50 bucks.
I skipped HL2, I skipped BioShock, I skipped a lot of others. I will skip Spore also. They won't get my 50$ until they come back to sane protections.
> Interestingly, the most important thing is missing from the summary
Not only that, but also the fabulous restriction:
"You may not [...] disassemble the drive"
This is ridiculous. A drive overwritten with zero data will, by definition, returns 0s through ATA commands. The reason why some people overwrite sensible data several time is to guard against a possible scanning transmission electron microscopy, which, of course would need the disk to be disassembled to be performed.
> Your scheme ignores the fact that self is a value that can be assigned.
What are you talking about ? The fact that the variable self (ie: the memory offset on the stack frame generated by the compiler) can be re-assigned in objective-C have absolutely nothing to do with the scheme I describe.
How are can it be for you to understand ?
In the remote site, look if the value returned by actual call of the implementation of the method is the same bit value we sent it to.
In that case set a bit value in the packet sent through the wire.
In the local site, if the bit is set, return the exact bit value of the local object we tried to send the message to.
> Your proposal introduces new semantics that require the language to distinguish between returning self and returning any other id.
You cannot seriously call some code to implement an optimization in NSConnection a "new semantic" in the language. (And beleive me, there were quite a few ad-hoc optimizations in NeXTstep code.)
To focus to what the thread is all about: there was no need to remove the "return self" idiom because of performance issues in DO, as a very simple optimization in DO code could have taken care of that (with no change in the semantic of the language).
Slashdot Mirror
...and by reading the description of the code, it was obvious that there were some issues in it ("As I alluded to, all three arguments are highly dynamic and constantly updated within the while() loop. There is a great deal of pointer arithmetic in this loop.")
As a general rule of the thumb, if some code does pointer arithmetic in loops with a lot of conditional tests and pointers assignment, it is probably buggy.
I know that this it seems presumptuous, but, over the years, I developed a good sense of looking at a block of C looping code, and saying if it is buggy based on the feel of the complexity in it, the presence of constants, etc, etc. If you can't "see" the invariants in the code by looking at it, it should raise a red flag.
There is that famous Kernighan saying:
"Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it."
Code should be dumb. In that case, they probably could have written slower but dumber code. Maybe by first testing if the path needs rewriting (most path don't), then use a slower/dumber rewriting loop, so the penalty would be paid only by the use of "." and "..". We are in 2008. Complex code is a liability. And vista is already slow, and that is not due to performance of path rewriting routines...
First, the wealthy do consume much more government spending than the poor. In a direct way, through the various pork techniques, indirect ways via use of the infrastructure or with specific laws (from IP protection, tariff, regulations, and a miriad of other ways -- like getting huge bonuses in financial companies, crashing them and letting the taxpayers with the bill).
Also, it can be argued that the wealthy do not even have to consume government resources to put a large burden on society as a whole. They use more land, consume more non-renewable resources, etc, etc.
Anyway, the grand-parent said that "Suppose that every day, ten men go out for beer and the bill for all ten comes to $100.". I point that this is definitely not the way to society is working, hence his whole analogy is flawed.
You call that "exactly on the money with verifiable data". Let me laugh.
In the second half of your post, you built such a nice strawman that I am not even going to argue with. Just read my post back, and tell me if the poor's beer is the same size as the rich's beer...
What a load of crap. If you want to do an analogy, do it right:
Suppose that every day, ten men go out for beer and get the following (numbers not accurate, but I don't pretend to have a PhD in economy):
The first man get no beer
The second man get 1 beer
The third man get 1 beer
The fourth man get 1 beer
The fifth man get 2
The sixth would get 3.
and so one until:
The tenth man (the greediest) get 50 beers.
Furthermore, the tenth man owns the bar.
Then the bill for all ten comes, bla bla bla.
> No software should cost more than $10k Exception: scientific software (and I mean the really advanced stuff, simulations, etc), math analysis, etc, etc AND EVEN THEN
Obviously, you work in the scientific sector, and also have a pretty narrow-minded view.
Having written other kind of software, I can tell you that you the complexity involved in "enterprisey software" can be mind boggling. For instance, think about reading, implementing and deploying some of the complex financial rules (we're talking about thousands upon thousands pages of regulations, with variations per countries) out there, in a market composed of a handful of customers.
In that case, the cost of the software is directly related to the cost of building it (ie: how much it would cost for the customer to build his own), and such customers are perfectly happy to pay millions (yes, millions) so a software vendors take care of all the issues related to said software.
Neither license is the one true license.
For instance, as developers are also users, one can argue that the GPL gives freedom to developers that the BSD license doesn't. For instance, if you work on a BSD project, a company can come in, buy the main contributors of the code, close it, and prevent you, as one of the original contributors, of enjoying the contribution of that key members, effectively killing the project.
As a developer, I generally seek for BSD libraries and GPL tools. The first guarantees that I can mix-and-match them to whatever I want, the second that they will be forever available.
Do. Not. Do. That.
Ever.
It is extremely easy to end up being paid roughly 0$ of royalties because the product ends up being sold as some sort of "add on" module to some other product, at a roughly zero price.
There are a 2 main clauses that must be present:
1) Be sure that the contract have a termination date. It saved me once. Take a date in the future, like 3 to 5 years, to say that the whole arrangement will disappear. It may seem a longtime, but you DO NOT WANT to have ANY liability following you for the rest of your life. Don't trust lawyers on that. You want, written black on white, in plain english, as the very first section of the contract, that all your contractual obligations vanishes at a certain date. Require that on every contract. The additional upside, is that, if they are happy with you, they'll have to renegociate in a few years (but it is not the main point. The main point, is that, if everything goes bad, you want to be able to say "this date is away: I have nothing to do with those guys anymore")
2) Ask money for non-competition. Each month, you have to be paid xxx$, or the non compete clause is waived. If they fire you and want you to continue to non-compete, that is the amount they have to pay you. That amount should be roughly equal to your salary. There should be no reference to the way they fire you (they can sell out, be bought, or whatever). Just say that, if they don't pay you that specific amount, the next month, you are definitely freed from the non-compete clause. This will also have the added value of them having to get back to you if they want to merge the company with another one.
When those two things are present, then you can start talking about the details. But, if they refuse any of those two clauses walk... no... run away.
Why do you says "stupidly restrictive" ?
The real wording should be:
"Unlike the restrictive GPL, the BSD license lets people make choices that you might not like, for the sake of developer freedom."
and
"Unlike the permissive BSD, the GPL license lets developer make choices that you might not like, for the sake of code freedom."
You see ? Two faces of the same coin. Neither license is "stupid".
Thanks for clearing the thing. I was wondering about what Sophie was. It sortof reminded me of Alice. Then I saw your post, and the familiarities are astounding.
* Named from a feminine first name from children books. Check
* Authoring System. Check.
* Written in Smalltalk. Check.
* That are going to be rewritten in Java. Check.
Conclusion, Sophie will follow Alice on the path of irrelevance. Rewriting a Smalltalk app in Java swing is an exercice in futility. It will be a slow bugged mess, which is what Alice is now.
Of course, it will also suffer the second syndrom system, ie: they will try to cram as much as new ideas as possible in that version, and will fail (don't think that because the proposal says "the very same thing" that it will. No. This is not how projects work).
Frankly, I don't know how Rybka would rate against Anand in a real match, and I don't think we'll ever know, because FIDE doesn't want it to happend (a bit like what occured with checkers -- a fascinating story http://en.wikipedia.org/wiki/Marion_Tinsley ). Furthermore, computers and human have such different ways of playing that I am not sure that elo ratings between computers and humans is a correct statistical measure (because you can win against Kramnik, as he may blunder, like in the Deep Fritz game where he blundered a mate in one -- but you can't expect any blunder from a computer, short of a software bug.)
Personally, I would love to think that human have some chance against computers, but I don't think so anymore. Beating a computer at chess is an exercise in futility, like outrunning a car, or generating machine code faster than a compiler...
There are several commercial use for better engines:
1) Game analysis. When you have played a game against and lost, you try to understand why you lost. Sometimes it is because you made an obvious blunder, but when you get better at the game, you start loosing for strategical reasons (lost control of a certain square, etc, etc). Having a good engine helps you try new ideas, and play a lot of what-if scenarios
2) Game understanding. When you follow a live tournament between grandmasters, having a good engine can give you an explanation about what the underlying ideas are ("Why doesn't he plays Nb6? You try it, and get the answer instantly")
3) Correspondence & Centaur Chess. Correspondence chess are long running games where both players have access to whatever they want. It delivers very subtle games, where the strategy is a very important aspect, as all the tactical blunders are removed by the use of good chess engines. Centaur chess is the same with lower time control.
And, of course, bragging rights are important too: having a better engine than other people in the chess club is a bit like having the better graphic card among fps players...
As you may have seen, playing against the engine is not one of the uses. Rybka is supposedly at 3200 elo. By definition, 200 elo points higher means you have a 75% win probability. The current world champion is at 2800, which means that he have a 6% win probability against rybka. Good club chess play is around 2000 (it takes several years to reach that level -- at that level, you can generally play blind, or multiple opponents, etc, etc). Such players have a 1 against 4000 chance against a 3200 player. Which means zero chance...
Whatever. Let's reply to that obvious troll.
1) You need opening preparation to play. Well, that's a given, and it was already the case 20 years ago. Now, with computers, opening preparation is easier, so players are better prepared.
2) Grandmasters have not memorized ending tablebases, first because it is impossible, and second, because grandmasters are still much much better than computers in endgames. Endgame is not about memorisation, but about technique. And yes, you have to work that too.
3) Average chess games is 40 moves long, not 60
4) Computers make opening books deeper but also wider: lines that were considered unplayable are now playable. This adds diversity to the games.
5) Middlegame last anywhere from 0 to 20 moves. Not "1 to 5".
6) Capablanca already said that chess was so analysed that it was boring and draw in 1930. Guess what: he was wrong.
So, what is exactly your complain ? That you have to work a lot to play competitively at your local chess club ? Well, that is not because of computers, it is because the players there are better than you are.
And that is not due to computer play. It is due to the internet. Because on the chess servers, you have people playing chess as if it was WoW. So, the guy you face at that local chess club, maybe he plays 40 "3+0" games a day...
I use parallels in full screen. My mac have BSODs all the time.
> These days it seems all but the most classified tech makes it into market very, very quickly.
No, you didn't got it right. The article meant that, in 10 years, everybody will be enrolled in the military due to the global draft to fight in the war after the economic collapse...
You are not the typical consumer. Your are a technology enthusiast. People like you are happy with BR, but you are not the main consumer market.
I told that over and over again to my HD fanatics friends:
1/ DVD is good enough.
2/ The competition to DVD is DiVX, proving that DVD is indeed good enough.
3/ Movie fanatics already have all their movies in DVD format. Only fans that don't have much DVDs (ie: that are young) may want to go BR for their entire collection.
4/ BR and DVDs are not compatible
5/ Difference between VHS and DVD was huge (no tear, instant access to scenes, better quality, can be played on computers). Differences between DVD and BR are comparatively very small.
Getting BR to the mass of consumer will be a very very costly exercise in futility.
Killer feature for the next format is being download-friendly. Want to see a movie ? Click on the right web site, see all movies ever done, enter a small amount of money, get the movie permanently on your devices (yes, permanently and with a 's', because consumers will not give away what they already get with DVDs). Of course, there are already plenty of formats to can do that, but the one that'll get ubiquitous will win. I hope it'll not gonna be flash :-)
My TRS-80 Model 100 is awesome (not that the model 200 have a much bigger screen).
But the poster will be pretty disapointed if he tries to read fat32 usb drive with that...
> The EULA for Firefox deals with the use of the trademark "Firefox," only. In order to keep the trademark, they have to enforce it. If the EULA didn't exist, Mozilla Corporation could lose the sole right to call a browser "Firefox."
That is bullshit. Software were trademarked before EULAs existed, and they never once lost trademark because of the lack of EULA. A simple short splashscreen with Firefox(tm) is more than enough for trademark.
> I can count on one hand the number of times I've had a problem with Firefox that would have been solved by it being in its own process.
I restart firefox roughly twice per hour when developing my javascript application. Having 10 concurrent tabs executing heavy javascript/ajax generally hangs the browser.
Of course, extenions (in particular FireBug) are probably responsible of that, and it is painful but not a showstopper. A process per tab model would probably be better for my usage...
The guy in http://www.youtube.com/watch?v=Lt1Yo610lG0 explains it all.
Yes, it is a STARGATE TO HELL ! Doom fans, rejoice ! Happy ! Happy !
Funnny, but... a 4 billion lines text file that is just over 1Gb ? I don't think so...
> Whenever a game is coming out that I'm interested in, if it's laden with such anti-consumer machinations, I intend to e-mail the publisher explaining that I would rather pirate it than pay them money thanks to their greed.
Wtf ? Skip it. Don't pirate it. Use the money to buy a different game, with no DRM, and tell the publisher/developer of *that* game why you bought it (be specific).
The problem with your way is that you encourage their behavior. I don't understand why you accept to jump through hoops just so you can send them 50 bucks.
I skipped HL2, I skipped BioShock, I skipped a lot of others. I will skip Spore also. They won't get my 50$ until they come back to sane protections.
> Interestingly, the most important thing is missing from the summary
Not only that, but also the fabulous restriction:
"You may not [...] disassemble the drive"
This is ridiculous. A drive overwritten with zero data will, by definition, returns 0s through ATA commands. The reason why some people overwrite sensible data several time is to guard against a possible scanning transmission electron microscopy, which, of course would need the disk to be disassembled to be performed.
How can this ends on slashdot ? Don't know...
> ... demonstrated your ignorance ...
Do you really believe that insulting me makes your position any more valid?
Anyway, thanks for the nice day, and have one too.
I replied to your (admittedly funny) "changing code in NSConnection would change objc semantic" argument.
http://slashdot.org/comments.pl?sid=612415&cid=24295167
(If you continue pretending that it would change objc semantic, put some code that would have its semantic changed by such an optimization)
> Your scheme ignores the fact that self is a value that can be assigned.
What are you talking about ? The fact that the variable self (ie: the memory offset on the stack frame generated by the compiler) can be re-assigned in objective-C have absolutely nothing to do with the scheme I describe.
How are can it be for you to understand ?
In the remote site, look if the value returned by actual call of the implementation of the method is the same bit value we sent it to.
In that case set a bit value in the packet sent through the wire.
In the local site, if the bit is set, return the exact bit value of the local object we tried to send the message to.
> Your proposal introduces new semantics that require the language to distinguish between returning self and returning any other id.
You cannot seriously call some code to implement an optimization in NSConnection a "new semantic" in the language. (And beleive me, there were quite a few ad-hoc optimizations in NeXTstep code.)
To focus to what the thread is all about: there was no need to remove the "return self" idiom because of performance issues in DO, as a very simple optimization in DO code could have taken care of that (with no change in the semantic of the language).
> You're wrong, you lose, get over yourself.
This have no impact on me. Give me hard facts.
I replied to the "self is a value on the stack" post.
Please read my answer carefully. Your move.
http://slashdot.org/comments.pl?sid=612415&cid=24289475