and the fact that with PHP you have code embedded in the HTML Just because you can does not mean you do, or do you embedd all Java code in the.jsps?
There's absolutely no reason why you could not write the same "simple intelligent classes" in PHP, just a different syntax mainly. And testing those with PHPUnit should also work quite well.
The issues I have with PHP are more the stupid scoping rules, its messy OO support and total lack of static checking. That said PHP can be powerful for lots of things.
You would probably use an accelerator on real projects anyway so the compile point doesn't count really.
I've used PHP without an accelerator in all "real projects" and never had any performance problems on the PHP-side. For web-applications the bottleneck is mostly the database so most "PHP is slow" vs "Java is slow" arguments are fairly pointless. The real issues are more often development speed and robustness. Of course if you're going to implement some nifty algorithms Java might be a better choice, but why not write those in any language and make it a PHP-extension or some kind of server?
Exceptions are required for enterprise quality applications. I'm not saying this is some ultimate solution, I've also missed exceptions badly, but if the application needs exception handling only in a few places it's quite simple to simulate them:
Wrap the function which throws the exception in a class
If the exceptional condition occurs have the function set a description in an instance field named for instance "exception" and then return
The client code calls the "getException"-method in the class of the function throwing exceptions
If exceptions occured, deal with them, otherwise proceed
Far from beautiful, but you know a lot of "enterprise applications" have been written in COBOL, Fortran or C which none of them has exception handling AFAIK (if gotos don't count).
Did you build the entire application in PHP, or only the "presentation layer"? I mean was there any Java in it, for instance in the database layer?
As a PHP-developer who maintains a fairly large site and intranet I've come to both love and hate PHP. Love it for it's development speed and ease. Hate it for its lack of static analysis, I mean yes the language is dynamically type-checked but that doesn't mean the interpreter couldn't check whether some code calls undefined functions or defined functions with the wrong amount of arguments at parse time so it would catch these errors without one having to test every code-path.
That and then the scoping absolutely terrible but most people know that already. That said I still believe PHP is a better solution than Java on the whole. Neither is really optimal but both are fairly good.
If they had been written in an object oriented language (such as C++) instead of Perl, the program would have been modular by default
Nonsense, OO design is difficult and people who code rubbish will produce OO rubbish as well. Of course Java rubbish is easier to understand than Perl rubbish.
Persistent cached data? So ocaml values can be instantiated once and then shared for the whole application like with a Java appserver? This is great stuff, I really have to try it out. How far along is it? Does it have any kind of session support or is that up to the application programmer?
Yes it's there but 4dos which I mentioned had auto-complete, command history, better directory listings etc. I just could not figure out how to configure it at that time so I was stuck with the idiotic CLI without even command history. What annoued me was the "It's there but we don't want you to use it" mentality.
It's not really "scripting" as such. OCaml programs are bytecode compiled and dynamically linked into the bytecode interpreter which runs inside Apache.
Does this mean that each request does not spawn an own process (as CGI usually does) but that the "script" is linked into apache? I mean is this not the way PHP for example works? Do you have an assessment on how much this speeds up the processing compared to CGI?
Well I didn't move to Linux just to use a command line, but one of my biggest annoyances with Windows is the lack of a good command line by default. I remember becoming Windows unfriendly when Win95 crippled my DOS prompt. Yes I know it was still possible to do everything you could before (run 4dos etc.) I just didn't know how at the time.
As you say, science is all about explaining what we see in the universe
Well then what is the difference between science, religion and philosophy?
The difference, in my opinion, is that in science you can measure what we can see in the universe and prove theories wrong. The only explaining power of science is in the theories which may be created by any means from intuition to coin tossing. The scientific part is predicting data from the theory and then measure some real data and see if the theory stands the test. Even if the theory explains some data-set it is in no way a guarantee that the theory is true, because the underlying real reason might just give similar output for the measured data-set.
Therefore science does not explain anything, it works through disproving, not proving. So whatever the motivation is, there is certainly nothing wrong with "poking holes" into theories from a scientific point of view. If there is, then the theory has become religion.
I wonder where exactly they draw the line for the copyright. Is it okay to call it an "unofficial translation". Writing a parody based on the original book is legal in most countries, could this not be a similar variant of the original?
A big part of his point was that even though the games are really complex, people learn them, even kids who can't concentrate five minutes on a book. That means we could find good learning principles in the games.
Since you seem to know something about these devices I might as well ask for some clarifications. Is it one screen that displays in both directions or two screens on top of each other? If it is the former then isn't the other screen "mirrored" rendering text pretty hard to read? Of course pictures usually work equally good no matter if they are mirrored or not.
In the camera example one could use software to invert the mirroring, if the phone is open the inner display shows text the right way, if the phone is closed ergo for the outer screen.
Not really replying to you, but your comment just clears one point very well.
I'm currently thinking of some drastic changes because I can't keep up with porting the new versions.
I think this is exactly the thing which most of the people saying "why didn't you use the code from project Y?" don't understand. If you use someone elses code you have to maintain a fork of that code. That means syncing with new versions to fix bugs etc. Now this will not be trivial since it is someone elses code, and probably you won't know it as well as if it were your own. So often the maintenance hassles from using someone else's code can be a very big factor against its use.
No website, no code just a problem and a few ideas...
I'm not sure it will work very well. In fact I doubt it will, information extraction works best for huge corpuses of text with some fairly simple regularity in their structure. It can still be natural language, but one might make the observation that A is usually before B, and bias the system for such a special case. There was one project where they mined medical journal articles for which genes were found to affect which properties. I don't have a link to the paper right now but if you're interested I can look it up.
I might put up a website at some point. The project is both school work and commercial work so I'll get around doing it at some point. The interesting part is school work and the useful part (the application) is commercial.
Ok, so can you find all emails and IMs from "Bonnie" without using the email-client or IM-client? Is it convenient? Could non-technical users use it?
Ye olde tree-based filesystem is great, but I still believe there are better ways to organize certain information. Relational databases are also pretty popular so they must solve something the filesystem does not.
Emails are usually not files, neither are contacts or calenders. They are in files in some specific format and not accessible to all operations that could benefit from them. If we are to build a better integrated system we either need to go back to all plain-text data in files in a hierarchical tree, or we need a use a more expressive yet standardized data model.
I'm not entirely surprised, usually when I get a good idea someone else has already implemented it before. The proprietary database is a bit of show-stopper though... I guess I have to check it out.
Yes this is certainly very cool, although the picture rotation example seems kind of lame from an extendability perspective =)
Continuations in the UI are really a significant idea and have the potential to make sessions much more useful. It would also solve the problem with UI-events "stealing" the focus from what you are trying to do. Instead of popping up som idiotic dialog when one is trying to type the event would wait nicely until the user can bothered to answer the question.
I wonder if it would be possible for the continuation not to contain only the state that it currently is in, but also the commands that got it there. If that would be possible then creating macros would become a lot more convenient. You would just perform the operations once, and then click the continuation and say "Make Macro", instead of repeating the steps with an operation recorder toggled.
In regards to plain text/personal information -- have you thought about looking at Bayesian filtering for a solution to that? I haven't (yet) but the idea is festering in my brain.
Bayesian filtering is another of those words that mean a lot of things;) Nowadays it is usually used in reference to the techniques used for spam-filtering, which is a very specific task. Classification: Spam / Not Spam. Basically everything that uses the bayesian view on statistics can be considered a bayesian method, without considering the underlying model. In other words for many statistical models it is possible to derive bayesian optimization schemes (or "learning rules"). A widely used set of language models are the Hidden Markov Models. I'm planning to use them on an information extraction problem (populating database tables from free-text descriptions), and that's about the closest to the problem we're discussing that I've been. You could probably use them as a partial solution here as well, but I can't think of any really clever scheme at the moment.
For personal information one would like to have something that clusters the data into different categories. There are lots of methods for this. One I'm familiar with is Self Organizing Maps (an example paper about them).
And finally, sorry to be boring, but I'm not currently working on anything that would create something like the system we've discussed =)
Actually, why don't you just ask the artist to decide which genre a song belongs to. The category (ies) that songs (data) belong to are often most evident to the originator.
Because different people have very different perceptions. Of course if the artist decides it's numetal-folk-country, then he certainly is "right" about that. But then bands playing similar music might call their music neoclassical funkcore just because they have a different background and the usefulness of the classification will be about as low as that of the current classifications. Therefore it's a lot more useful to know about similarity of bands (or songs) than of some "orthodox" classification.
When I said "personal information" I meant email, web-pages, text files and documents rather than code. Sure it would be cool with tools for dealing with code as well, but I suspect you would need to use different methods than for the natural language material to be succesful.
What I would like to have would be something that could integrate all information related to some "project" into a single, powerful interface. Something that would allow me to have information in different formats (txt, html, pdf, ps, doc, spreadsheets, images and so on) searchable from the same locations. Also it would need to have a convenient way of "commenting" material written by others and linking these commments to the right place in the right document, without me having to place e.g. HTML links in the right locations. Then discussion material (mailing-lists, forums) should also be integrated in the same interface.
I guess my main point is that different media should not be artificially classified by its media type but rather by its contents (or more specifically which project it belongs to). My secondary point would then be that everything should be searchable, rather than found using a URL. More like the web and a relational database and less like a filesystem tree.
Slashdot Mirror
and the fact that with PHP you have code embedded in the HTML .jsps?
Just because you can does not mean you do, or do you embedd all Java code in the
There's absolutely no reason why you could not write the same "simple intelligent classes" in PHP, just a different syntax mainly. And testing those with PHPUnit should also work quite well.
The issues I have with PHP are more the stupid scoping rules, its messy OO support and total lack of static checking. That said PHP can be powerful for lots of things.
You would probably use an accelerator on real projects anyway so the compile point doesn't count really.
I've used PHP without an accelerator in all "real projects" and never had any performance problems on the PHP-side. For web-applications the bottleneck is mostly the database so most "PHP is slow" vs "Java is slow" arguments are fairly pointless. The real issues are more often development speed and robustness. Of course if you're going to implement some nifty algorithms Java might be a better choice, but why not write those in any language and make it a PHP-extension or some kind of server?
I'm not saying this is some ultimate solution, I've also missed exceptions badly, but if the application needs exception handling only in a few places it's quite simple to simulate them:
Far from beautiful, but you know a lot of "enterprise applications" have been written in COBOL, Fortran or C which none of them has exception handling AFAIK (if gotos don't count).
Did you build the entire application in PHP, or only the "presentation layer"? I mean was there any Java in it, for instance in the database layer?
As a PHP-developer who maintains a fairly large site and intranet I've come to both love and hate PHP. Love it for it's development speed and ease. Hate it for its lack of static analysis, I mean yes the language is dynamically type-checked but that doesn't mean the interpreter couldn't check whether some code calls undefined functions or defined functions with the wrong amount of arguments at parse time so it would catch these errors without one having to test every code-path.
That and then the scoping absolutely terrible but most people know that already. That said I still believe PHP is a better solution than Java on the whole. Neither is really optimal but both are fairly good.
If they had been written in an object oriented language (such as C++) instead of Perl, the program would have been modular by default
Nonsense, OO design is difficult and people who code rubbish will produce OO rubbish as well. Of course Java rubbish is easier to understand than Perl rubbish.
Persistent cached data? So ocaml values can be instantiated once and then shared for the whole application like with a Java appserver? This is great stuff, I really have to try it out. How far along is it? Does it have any kind of session support or is that up to the application programmer?
Yes it's there but 4dos which I mentioned had auto-complete, command history, better directory listings etc. I just could not figure out how to configure it at that time so I was stuck with the idiotic CLI without even command history. What annoued me was the "It's there but we don't want you to use it" mentality.
Does this mean that each request does not spawn an own process (as CGI usually does) but that the "script" is linked into apache? I mean is this not the way PHP for example works? Do you have an assessment on how much this speeds up the processing compared to CGI?
Well I didn't move to Linux just to use a command line, but one of my biggest annoyances with Windows is the lack of a good command line by default. I remember becoming Windows unfriendly when Win95 crippled my DOS prompt. Yes I know it was still possible to do everything you could before (run 4dos etc.) I just didn't know how at the time.
As you say, science is all about explaining what we see in the universe
Well then what is the difference between science, religion and philosophy?
The difference, in my opinion, is that in science you can measure what we can see in the universe and prove theories wrong. The only explaining power of science is in the theories which may be created by any means from intuition to coin tossing. The scientific part is predicting data from the theory and then measure some real data and see if the theory stands the test. Even if the theory explains some data-set it is in no way a guarantee that the theory is true, because the underlying real reason might just give similar output for the measured data-set.
Therefore science does not explain anything, it works through disproving, not proving. So whatever the motivation is, there is certainly nothing wrong with "poking holes" into theories from a scientific point of view. If there is, then the theory has become religion.
I wonder where exactly they draw the line for the copyright. Is it okay to call it an "unofficial translation". Writing a parody based on the original book is legal in most countries, could this not be a similar variant of the original?
A big part of his point was that even though the games are really complex, people learn them, even kids who can't concentrate five minutes on a book. That means we could find good learning principles in the games.
Since you seem to know something about these devices I might as well ask for some clarifications. Is it one screen that displays in both directions or two screens on top of each other? If it is the former then isn't the other screen "mirrored" rendering text pretty hard to read? Of course pictures usually work equally good no matter if they are mirrored or not.
In the camera example one could use software to invert the mirroring, if the phone is open the inner display shows text the right way, if the phone is closed ergo for the outer screen.
Cool, with this device and a mirror I will be able to watch two screens showing the same image at the same time.
Seriously though, it will make it very hard to hide from your friends what a lousy nibbles player you are.
Professional? Do you think people who use software to avoid firewalls and proxies care about professional?
Not really replying to you, but your comment just clears one point very well.
I'm currently thinking of some drastic changes because I can't keep up with porting the new versions.
I think this is exactly the thing which most of the people saying "why didn't you use the code from project Y?" don't understand. If you use someone elses code you have to maintain a fork of that code. That means syncing with new versions to fix bugs etc. Now this will not be trivial since it is someone elses code, and probably you won't know it as well as if it were your own. So often the maintenance hassles from using someone else's code can be a very big factor against its use.
Well MAYBE. But my way is faster.
I'll check nerdfarm every now and then so I can join if I have something to contribute.
What kind of language model did you use for the IRC analyser? It would probably possible to create a model resilient to bad spelling.
No website, no code just a problem and a few ideas...
I'm not sure it will work very well. In fact I doubt it will, information extraction works best for huge corpuses of text with some fairly simple regularity in their structure. It can still be natural language, but one might make the observation that A is usually before B, and bias the system for such a special case.
There was one project where they mined medical journal articles for which genes were found to affect which properties. I don't have a link to the paper right now but if you're interested I can look it up.
I might put up a website at some point. The project is both school work and commercial work so I'll get around doing it at some point. The interesting part is school work and the useful part (the application) is commercial.
Ok, so can you find all emails and IMs from "Bonnie" without using the email-client or IM-client? Is it convenient? Could non-technical users use it?
Ye olde tree-based filesystem is great, but I still believe there are better ways to organize certain information. Relational databases are also pretty popular so they must solve something the filesystem does not.
Emails are usually not files, neither are contacts or calenders. They are in files in some specific format and not accessible to all operations that could benefit from them. If we are to build a better integrated system we either need to go back to all plain-text data in files in a hierarchical tree, or we need a use a more expressive yet standardized data model.
I'm not entirely surprised, usually when I get a good idea someone else has already implemented it before. The proprietary database is a bit of show-stopper though... I guess I have to check it out.
Yes this is certainly very cool, although the picture rotation example seems kind of lame from an extendability perspective =)
Continuations in the UI are really a significant idea and have the potential to make sessions much more useful. It would also solve the problem with UI-events "stealing" the focus from what you are trying to do. Instead of popping up som idiotic dialog when one is trying to type the event would wait nicely until the user can bothered to answer the question.
I wonder if it would be possible for the continuation not to contain only the state that it currently is in, but also the commands that got it there. If that would be possible then creating macros would become a lot more convenient. You would just perform the operations once, and then click the continuation and say "Make Macro", instead of repeating the steps with an operation recorder toggled.
In regards to plain text/personal information -- have you thought about looking at Bayesian filtering for a solution to that? I haven't (yet) but the idea is festering in my brain.
;)
Bayesian filtering is another of those words that mean a lot of things
Nowadays it is usually used in reference to the techniques used for spam-filtering, which is a very specific task. Classification: Spam / Not Spam. Basically everything that uses the bayesian view on statistics can be considered a bayesian method, without considering the underlying model. In other words for many statistical models it is possible to derive bayesian optimization schemes (or "learning rules").
A widely used set of language models are the Hidden Markov Models. I'm planning to use them on an information extraction problem (populating database tables from free-text descriptions), and that's about the closest to the problem we're discussing that I've been. You could probably use them as a partial solution here as well, but I can't think of any really clever scheme at the moment.
For personal information one would like to have something that clusters the data into different categories. There are lots of methods for this. One I'm familiar with is Self Organizing Maps (an example paper about them).
And finally, sorry to be boring, but I'm not currently working on anything that would create something like the system we've discussed =)
Actually, why don't you just ask the artist to decide which genre a song belongs to. The category (ies) that songs (data) belong to are often most evident to the originator.
Because different people have very different perceptions. Of course if the artist decides it's numetal-folk-country, then he certainly is "right" about that. But then bands playing similar music might call their music neoclassical funkcore just because they have a different background and the usefulness of the classification will be about as low as that of the current classifications. Therefore it's a lot more useful to know about similarity of bands (or songs) than of some "orthodox" classification.
When I said "personal information" I meant email, web-pages, text files and documents rather than code. Sure it would be cool with tools for dealing with code as well, but I suspect you would need to use different methods than for the natural language material to be succesful.
What I would like to have would be something that could integrate all information related to some "project" into a single, powerful interface. Something that would allow me to have information in different formats (txt, html, pdf, ps, doc, spreadsheets, images and so on) searchable from the same locations. Also it would need to have a convenient way of "commenting" material written by others and linking these commments to the right place in the right document, without me having to place e.g. HTML links in the right locations. Then discussion material (mailing-lists, forums) should also be integrated in the same interface.
I guess my main point is that different media should not be artificially classified by its media type but rather by its contents (or more specifically which project it belongs to). My secondary point would then be that everything should be searchable, rather than found using a URL. More like the web and a relational database and less like a filesystem tree.