You all are wrong. 40% of world's email traffic is of USA and 99% of it is spam.
My yahoo, netscape and hotmail accounts are 99% full of junk. But even friends from US corporations complain about the same - their work mailboxes 99% full of junk at periods when a sysadmin switch off a spam filter temporary.
You like the democracy? You've got it. At least you've got what you call a democracy.
How about OLAP-based OS?
on
Opencroquet
·
· Score: 1
First of all, I belive it is not impossible that there can be 3D GUI, which can work only one specific OS, which becomes 3D OS in that case.
Also, the history of Microsoft proves, that some OS can be designed so bad that it cannot function properly (or at all) without one or more GUI components. Theoretically, if you will take 3dWM, Berlin/Fresco (or other post-X11 3D GUI) and embed it to the OS kernel (don't ask me, I don't know why, ask Microsoft - they know) than you will get 3D OS.
Now, forget GUI. But I still think 3D OS is possible. OLAP is typically based on multi-dimensional data mining. So, if (and somehow) OS kernel is functioning based on 3D data mining, then it's 3D OS:)
Most of moderns OS kernels do not need dimensions. Dimensions are needed to classify or to measure something (like to measure the position in case of GUI-based presentation). Modern OSes do not classify anything and they measure nothing. What if some future OS will start to classify or to measure something? Let's say, to measure the overload of nodes in order to schedule tasks in the distributed computing grid, huh?
Wait a minute, how about grids? The grid can be 3-dimensional, right? Than the OS controlling the grid is 3D OS!
So, OS can be 3D by one (or more) of the following three (or more) criteria:
if OS has a UI, which is 3D and it's unique, and it can work only on this OS;
if 3D data mining is essential part of the kernel work and/or the kernel panic without doing 3D data mining;
That'news from America. How about Europe and Japan, the places where they are always ahead of America in terms of technological progress and quality of life? Besides, Europe and japan do not have their Texas and do not have "Bush'ish" control over Mid-East oil. They have a better developed road infrastructure and higher health safity requirements. One more: thye hate to produce a "greenhouse" gas.
Seems to me that electric cars make more sense for Europe and Japan. Can anyone from Europe or Japan confirm or confront my specualtions?
Who use this phone devices today? I almost never received calls from telemarketers. But my yahoo acounts are flooded by spam.
I leave my yahoo addresses in all suspecius forms. In fact - in all forms. Of course I keep my home email addresses as private as possible. And many people do the same. It is not convinient. And it's not what we, email users, expected originally from the email system.
So what about email antispam laws? Many people and especially corporations would be happy to register their email accounts in the "do-not-call" list, and they are ready to pay for it to protect their privacy.
About "denial of service atack"... i am not sure I understand you clearly. but I can predict the stuation when the "reporter" will send the same fingerprint again and again even when s/he's got only one message with signed by that key. In order to prevent such sorts of situations a very simple thing could be done: when the message is send originally, it's got assigned a message ID. So, when you report a bad fingerprint you attach also a message ID and thus CA cannot count twice the same fingerprint when you've got actually only one bad message.
But if you mean DoS as a real DoS for CA... With my proposals or without it CA servers must be rotected against DoS atack right?
About a human right to send the anonymous message. Anounoymous means "some part or all your persoanl identification is hidden". Hidden from whom? When you meet me on the street I have my right to refuse your request to know my name. Unless you are a cop. This is essential in any civilization: there are authorities for whom you must comply and identfy yourself if they ask for it. Of course only if they are asking. Of course they cannot share your information. And of course if you don't have your ID then immigration service (INS) may give you many troubles upto to suspending many of human rights and freedoms.
Same in email. If you wanna send anonymous message then take some sort of anonymous from CA (without actual name on it, but still unique and with backtracking to your actual ID). Until you are doing anything wrong (like spam) - nobody will guess who is behind the key. But once your key is revoked then some authorities may come (using backtracking info about your actual ID when you have originally ordered that annymous key) and ask you to pay for what you've done.
Speaking about money: make a good spam penalties and return all investments soon. Of course many spammers will shutdown the business. But the others will adapt. Good, dynamic and flexible, optimization of N (trigger theshold to fire the penalty) and penalty amount will keep money flow. From the other side, use some part of internet taxes. That's the way we pay cops, right? We do the business, we pay taxes, cops get some money from it and keep our business safe (at least they are trying to do so).
Look at the subject. Sounds crazy? It is crazy - the traffic from/to virtual IP addresses is not routed. Even inside your proprietary networks. The only way to route it is to translate address to some real one from the list of non-anonymous addresses. Thus, every packet in Internet is identified by its source. Just in case of NAT the source points to that local ISP.
Same thing must be done in email business. All messages must have correct From field, and all must be signed with key identified by From field. All keys must be signed in a public (well known and trusted) CA. The router should relay only messages with verified signature keys.
No more anonymous email!
It is not a business of the router to watch the content of the message. But the header (let's think that the sig is the part of the header) must be checked. Specifically, the key of the sig must be verified in CA for being non revoked. Of course the revoke list can be cached on the router side for performance reasons.
Then it will be very easy to create a report gathering center (it could be a DB at the same CA), where you can send the fingerprints of bad guys. Once the key collects more than N (specified) bad reports, the key is revoked, and all messages with key won't be relayed anymore.
Of course, inside the interprise you can have "old-style" non-signed email. But in order to relay the message into outside of theintranet, it must be signed. That can be done automatically by the enterprise relay on behalf of corporate users. Even using the corporate key - remember NAT for IP?
Have this infrastructure on the place and it will often unnecessary to presecute spammers: their keys will be revoked after N bad reports.
By the way, how to send a bad report? Simple. Remember Y!Mail UI? They have a link in the message header area: "Report as a spam". Same button should be in every MUA UI: "Report the key a spammer's one". As well as another button: "Bleck messages with this key" - you still can do blocking (or any other filtering) of messages locally based on the key (ID).
Why a key is better than a return address? You must have a valid key. That's the key:)
Can you generate new keys with the same speed as you can generate your return addresses? I don't think so. And if you still can generate enough of key to keep your business profitable, then let's change your profit calculation forumal. From now on, the spammer has to pay big penalties for revoked keys.
Why not? I like when the most and resently used choice in auto-completion URL list or in menu is going up. I like the idea of menu hiding rarely used items until I want a complete list of menu items. I like bayessian mail-filters letting me still to have access to not-in-white-list messages.
Of course it will take time when word auto-completetion in office major suits will be improved enough to satisfy 99% of users (though it's not that bad already, 80%?). I think OOo and other open-source has more chances to adapt with good quality and speed than MS Word and other proprietary companies. I'll explain.
I remember back in Rusiia how Microsoft and IBM came with first "russified" (UI translated into Russian) software. People has been laughing all the time and that didn't help their sales. A direct translation into Russian language didn't work: words were either too long to fit the place in UI, or there was no clear direct translation (word-to-word) at all. By the time many of local software companies produced own application, of course with Russian UI. Some has very lucky language, anothers not. What's happened is people begun to use in their daily conversations, user manuals and emails "lucky" russian words and sentencies and ignored "unlucky" ones. In few version-generations the local software market had own new dialect of Russian language, mixed of lucky traslated words, invented new words (or newly used old words) and translittterated English words (writen in Russian letters accordingly to the sound). The democraty won and Microsoft has to use what Russians has created themselves, abondoning the previous too academical choice of own translation.
How is it realted to Open source? Simple. Opensource software most likely doesn't use the choice of a single authority. Instead, it use "lucky" choices of the community of users and developers. Open source anti-spam filters has more chances than proprietary ones. UI developers of self-adapting components will share the most lucky choices and strategies and less user users will be annoyed by bad adaptation.
How is it trelated to cell phones? Same way. But there is one problem. Today the software on my cell is proprietary. I agree that protocol drivers should be proprietary. But I want to install UI in my cell like I do with UI on my Linux desktops. I want to change it, reprogram it and adapt it for my own need and habits. 99% of cell users do not know what I am talking about. Many of desktop users don't understand it either - but they still use the result of work of that 1% users who understand enough to change it and who is motivated enough to do it righ.
Basically, once cell phone UI will be open-sourced - it will adapt and you won't be annoyed.
BTW, why just UI? Perhaps some guy want to right own protocols. But we talked about UI adaptavity.
I don't belive the parent. He wa so stupid at time of trying Gentoo, then he jumped to FreeBSD and was successfull? There is a serious inconsistense in his words.
Last time I've tried FreeBSD (several 4.*) the system required to know BSD and ports just to install it from the first time. The documnetation was terrible (recursively written for people who already know FreeBSD), forums answers mostly "RTFM". The building of my customized BSD kernel was just a nightmare comparing to what I used to do with Linux kernel (despite the fact that it didn't support everything I had in my PC, which was supported fine by Linux kernel).
If some OS is not for newbies - that would be *BSD. After Linux/Solaris/AIX/NT multi-year experience I was screwed up with FreeBSD. The first impression was: why this OS is so unfriendly if loose many functionality and drivers jumping from Linux to BSD?
But what I liked in FreeBSD was ports (eventually I've been learning FreeBSD anyway, partially just learn ports). So, when I found a Linux distro with the package management system like ports (right - Portage), I immidiately fired all my RH, YDL and FBSD installations (one by one actually), installed Gentoo and now I don't see the chance that any better distro will come around any time soon.
Well, better than Gentoo? There is such OS already, it's new Gentoo! Despite the fact that Gentoo is already the best OS, it's constantly being improved, enhanced, advanced and further developed.
There are many examples of programming in XML. XSL is one of them, XQuery is another one. Also, I can mention XSP (XML server pages for either Cocoon or Axkit) and DPT (Zope analog of XSP). You can find also various ontology programming dialects of XML on Semantic Web
The good point of XML is its unified syntax - you don't need a new parser each time, you already have one and you concentrate on semantics rather than syntax.
That's why there are such projects as XML representation of FlatCurry. Similar implementations I've met also in mail-lists about Mercury (BTW, they have also discussed S-exps as a basis of new syntax). Even ASN.1 goes to XML.
Again, I think XML tags are worse readable (and less compact) than Lisp S-expressions. Thast's why I believe that eventualy the market attraction of Lisp-like languages will tise again.
The switch from Lisp wasnt the mistake, it was the move to C syntax that was wrong, replacing the unreadable parentheses with semi-colons, curly brackets, squiggles, dots and loads of other stupid ideas.
What make Lisp's parentheses to be unreadable while XML parentheses to be still readable? By my opinion, Lisp parentheses are much more readable and that's the thing I miss in any other languages.
Never trust a language with more brackets than code
How about XML?
Besides, every language has barackets. Just in Lisp all bracket are unified to be as (). Other languages use various forms of brackets: (), {}, [], ";", "new-line". Python use even space identation instead of brackets (the only thing I hate in Python!).
When the code is the data, unified brackets are much better for verification purposes. That's why XML also has unified brackets (just badly designed, IMHO).
Relation model has a good math behind. Object model has not a math model - it has an organizational model. For relational model there is a math helping to verify the consistency of the data you query. You cannot do such verification with object model. That's why I don't see the future without ORDBMS (Object-Relational).
The best ORDBMS I found today is PostgreSQL, which has one more very important feature implementatio: it's very programmable database - it has ability to write stored functions on scripting languages and you can extend the database with new interpreters. That helps programmers to store the bahavior (functional data).
CORBA is Web-services done right -- it simply doesn't do it over port 80
Wrong. What is the port number doesn't really matter. How the protocol is organized matters. CORBA originally has been designed with two problems:
Syncronous communication - asyncronous (fire and forget) messaging is still in dreams of real-life CORBA developers;
Tight coupling - both client and server has to really have the same knowledge about the shared (passed, called) object;
CORBA might be not a bad choice for prototyping at earlier days of GNOME, but today it must begone. I think it must be replaced by SOAP. So, let's wait for MONO becoming stable, then let's get rid off gnorba and then let's compare GNOME with then's KDE.
1. There is no "database of eyes". Your eye you keep with you. At the moment of initial registration an input device will take a "pattern" of your eye: the "pubic key" which can be producted from the image of your eye, and it can be compared with run-time "private" keys later. Each eye produce a unique key (actually the pair: public and private), but either private or public key cannot be used to reproduce the image of the eye. If someone will still your "public key" from the DB it can be used. The private key is compared with the "public key" in runtime and it is stored nowhere. Both "keys" are signed. The "private key" is signed with the unique certificate of the input device. The "public key" is signed with the unique certificate of DB (actually, of authorization system). The leaking of either key is useless in such a case.
2. An input device takes a key based not only on static image, but also on micro-motions of your eye, which are also uniue. So don't worry, your dead eye cannot be used either.
3. Whole software of the authorization system with DB is useless to steal as it won't work without the hardware system key, which value is used to generate runtime system certificates. In order to make useless the stealing of the whole building the system is usually geographically distributed.
4. The input device is useless to steal as it is connected in the LAN to similar hardware key. Besides, it can take a picture of the room around and won't work if it's changed too much.
5. The only problemis that is left is about loosing your eye. Then you have to go through all bank offices and re-register yourself to use the image of your pennis or any other unique part of your body. But that is offtopic of your question.
but what are kids doing with computers at that age? Mostly drawing pictures, writing essays, making posters, and maybe a little bit of programming.
The list above is cross-platform. Of course there are some differences between such programs for Windows, MacOS, but kids won't notice them and won't understand them either. It's too early for that - they just see computer almost first time in their life.
After using Macs in schools ("using": see the list above), kids go to business and they do the same with Windows. Again, there are few differences between MacOS and Windows, but they are compensated by differences in task lists as well.
So, what's the result? Apple helps to Microsoft. Seriously. Apple sponsor schools with Macs. Students educate how to use computers. They come to business and buy a cheaper commodities: wintel products.
Now, how to make sure that Linux in schools will not help to wintel sales, but instead it will help to Linux sales (or Linux downloads)? Simple. Don't teach in schools how to use same functions on computers as they are on windows. Instead teach how to use computers in a way that impossible or difficult on windows (and macos). The keywords here are CLI, Scripting and Security. Briefly, modern regular (non-professional CS) user must know something about what is inside PC/OS, how to program for personal needs, and how to protect own informational life.
Command Line Interface is not a relict of 70s, it's a basement of control. Do you want to control your OS yourself? - Use CLI. Other wise let GUI (with bugs) to help you to think that you control your system. CLI is a manual-stick car (no automatic transmission). If a student will know how and why to use CLI - there will be less problems with GUI usage as well. CLI is key to know the computer. Isn't it a goal for students to use computers in schools?
Modern scripting languages are not CLI macro agreements. Look at Python: it has a bytecode compiler, a garbage collector and it's OOP. How is it different from Java? Only in a good sense, mostly speed and size. Python lets a student to write rapidly and interactively small scripts accordingly to tasks from a text book. Results work right the way
and often faster (no monster-JVM).
Many computer science experts will support me when I will tell: don't teach imperative, teach functional. If a student know how to use FP, there will be no problem with imperative languages. Otherwise is not true: many students after learning imperative languages in fact do not know how to program. By the way, Python has FP primitives. Besides Python, I would recommend students to use Scheme, Lisp, OCAML, Haskell and Prolog. And of course SQL.
You may notice that future non-programmers do not have to know so many langs. You're right. But they have to know at least one of them. Please, don't let Java or C or C++ to be one of them - they will learn nothing but useless syntax. Non-programmers are not going to use Java at home. As for scripting langs - no problem. That's what's happen to Visual Basic. But we don't want VB, we want something Linux-friendly, MS-independent and math-better.
Security is another key. More and more schools today consider of teaching Karate or other self-defense disciplines. Especially in urban areas. Today we are all living in one big cyber-urban area - Internet. Are you going to trust ISP or you are going to protect yourself?
I propose that schools will teach students how to hack each other on special hacking labs. If you know how to hack - you have more chances how to protect. Otherwise is not true. Today most of internet users still do not know that blind executing of unknown email attachments is most likely a virus infection. They don't know how to use MUA filters against the spam. No need to mention they don't know that using PGP or GPG signing/encryption can protect their private information in email messages as well as to protect against the spam. Why? B/c none taught them. You can tell all of tat to them hundreds of times. But until they will learn on practice how to win the war - they will continue to loose it.
I primarily use Linux but I couldn't pass up the powerbook when I looked at laptops.
Same story here. Almost. Same beginig, but with a different end. I've got Gentoo/PPC installed on my Powerbook. Same Gentoo ason my other PCs and PPCs. As a result I've got consistent environment (Gentoo 1.4) across several PC and PPC computers, saving my time (very valuable) on admin efforts. Besides, programming in such environment is just the pleasure.
Gentoo/PPC runs on that Powerbook faster than OSX. By the way, running MOL for the rest of few Mac (not open sourced) apps is much more stable than attempts to run Linux apps on OSX.
Micro BSD is dead. Few to go.
on
MicroBSD Is No More
·
· Score: 0, Flamebait
BSDi was dead way ago. Micro BSD is dead now. Who's next? Free? Open? Net?
Anyway. They are already dead. Only zombi use them. The rest of the world just ignores them. So do I.
You need that part (which is responsible for boot) of BIOS to boot (to start boot loader from the sector 0). You do not need that part of BIOS to work after the OS booter started work.
You do need Linux kernel to work all time as it handles processes and it has drivers for hardware.
But as other noticed (honestly - I forgot about it), the kernel still can (and does) use those BIOS functions which unify the usage of floppies and hard-disks and video.
Today you don't need anything "between hardware and OS". The kernel (AFAIK about Linux kernel) works directly with hardware resources. Old DOS did not have any kernel (Bill Gates was not so smart to write one), instead it had so-called bios.sys - the module which just called BIOS for some IO functions from the BIOS chip. Those days it was done for performance. I guess, today no OS use it.
Another function of BIOS was (and still is) to give the chance to configure some hardware CMOS parameters. Again, DOS was not able to do that. With Linux kernel you have NVRAM driver, which can change most of such parameters right from OS. I don't know any parameters, which either today or in near future would be unable to be set from a NVRAM driver of the OS kernel.
The main reason of changing those parameters was inflexibility. But today nonody will set IDE disk parameters - instead they used AUTO. Same for many other things. Although, some of CMOS might not be automatically tuned up, I am not sure.
The last (from what I can recall right now) reason to still use BIOS is to point to the boot device. Perhaps that the function that will survive longer than other. At least that's the on;y function I still use in Apple's firmware:)
Slashdot Mirror
My yahoo, netscape and hotmail accounts are 99% full of junk. But even friends from US corporations complain about the same - their work mailboxes 99% full of junk at periods when a sysadmin switch off a spam filter temporary.
You like the democracy? You've got it. At least you've got what you call a democracy.
Also, the history of Microsoft proves, that some OS can be designed so bad that it cannot function properly (or at all) without one or more GUI components. Theoretically, if you will take 3dWM, Berlin/Fresco (or other post-X11 3D GUI) and embed it to the OS kernel (don't ask me, I don't know why, ask Microsoft - they know) than you will get 3D OS.
Now, forget GUI. But I still think 3D OS is possible. OLAP is typically based on multi-dimensional data mining. So, if (and somehow) OS kernel is functioning based on 3D data mining, then it's 3D OS :)
Most of moderns OS kernels do not need dimensions. Dimensions are needed to classify or to measure something (like to measure the position in case of GUI-based presentation). Modern OSes do not classify anything and they measure nothing. What if some future OS will start to classify or to measure something? Let's say, to measure the overload of nodes in order to schedule tasks in the distributed computing grid, huh?
Wait a minute, how about grids? The grid can be 3-dimensional, right? Than the OS controlling the grid is 3D OS!
So, OS can be 3D by one (or more) of the following three (or more) criteria:
Seems to me that electric cars make more sense for Europe and Japan. Can anyone from Europe or Japan confirm or confront my specualtions?
Who use this phone devices today? I almost never received calls from telemarketers. But my yahoo acounts are flooded by spam. I leave my yahoo addresses in all suspecius forms. In fact - in all forms. Of course I keep my home email addresses as private as possible. And many people do the same. It is not convinient. And it's not what we, email users, expected originally from the email system. So what about email antispam laws? Many people and especially corporations would be happy to register their email accounts in the "do-not-call" list, and they are ready to pay for it to protect their privacy.
But if you mean DoS as a real DoS for CA... With my proposals or without it CA servers must be rotected against DoS atack right?
About a human right to send the anonymous message. Anounoymous means "some part or all your persoanl identification is hidden". Hidden from whom? When you meet me on the street I have my right to refuse your request to know my name. Unless you are a cop. This is essential in any civilization: there are authorities for whom you must comply and identfy yourself if they ask for it. Of course only if they are asking. Of course they cannot share your information. And of course if you don't have your ID then immigration service (INS) may give you many troubles upto to suspending many of human rights and freedoms.
Same in email. If you wanna send anonymous message then take some sort of anonymous from CA (without actual name on it, but still unique and with backtracking to your actual ID). Until you are doing anything wrong (like spam) - nobody will guess who is behind the key. But once your key is revoked then some authorities may come (using backtracking info about your actual ID when you have originally ordered that annymous key) and ask you to pay for what you've done.
Speaking about money: make a good spam penalties and return all investments soon. Of course many spammers will shutdown the business. But the others will adapt. Good, dynamic and flexible, optimization of N (trigger theshold to fire the penalty) and penalty amount will keep money flow. From the other side, use some part of internet taxes. That's the way we pay cops, right? We do the business, we pay taxes, cops get some money from it and keep our business safe (at least they are trying to do so).
Same thing must be done in email business. All messages must have correct From field, and all must be signed with key identified by From field. All keys must be signed in a public (well known and trusted) CA. The router should relay only messages with verified signature keys.
No more anonymous email!
It is not a business of the router to watch the content of the message. But the header (let's think that the sig is the part of the header) must be checked. Specifically, the key of the sig must be verified in CA for being non revoked. Of course the revoke list can be cached on the router side for performance reasons.
Then it will be very easy to create a report gathering center (it could be a DB at the same CA), where you can send the fingerprints of bad guys. Once the key collects more than N (specified) bad reports, the key is revoked, and all messages with key won't be relayed anymore.
Of course, inside the interprise you can have "old-style" non-signed email. But in order to relay the message into outside of theintranet, it must be signed. That can be done automatically by the enterprise relay on behalf of corporate users. Even using the corporate key - remember NAT for IP?
Have this infrastructure on the place and it will often unnecessary to presecute spammers: their keys will be revoked after N bad reports.
By the way, how to send a bad report? Simple. Remember Y!Mail UI? They have a link in the message header area: "Report as a spam". Same button should be in every MUA UI: "Report the key a spammer's one". As well as another button: "Bleck messages with this key" - you still can do blocking (or any other filtering) of messages locally based on the key (ID).
Why a key is better than a return address? You must have a valid key. That's the key :)
Can you generate new keys with the same speed as you can generate your return addresses? I don't think so. And if you still can generate enough of key to keep your business profitable, then let's change your profit calculation forumal. From now on, the spammer has to pay big penalties for revoked keys.
Am I am asking too much?
Of course it will take time when word auto-completetion in office major suits will be improved enough to satisfy 99% of users (though it's not that bad already, 80%?). I think OOo and other open-source has more chances to adapt with good quality and speed than MS Word and other proprietary companies. I'll explain.
I remember back in Rusiia how Microsoft and IBM came with first "russified" (UI translated into Russian) software. People has been laughing all the time and that didn't help their sales. A direct translation into Russian language didn't work: words were either too long to fit the place in UI, or there was no clear direct translation (word-to-word) at all. By the time many of local software companies produced own application, of course with Russian UI. Some has very lucky language, anothers not. What's happened is people begun to use in their daily conversations, user manuals and emails "lucky" russian words and sentencies and ignored "unlucky" ones. In few version-generations the local software market had own new dialect of Russian language, mixed of lucky traslated words, invented new words (or newly used old words) and translittterated English words (writen in Russian letters accordingly to the sound). The democraty won and Microsoft has to use what Russians has created themselves, abondoning the previous too academical choice of own translation.
How is it realted to Open source? Simple. Opensource software most likely doesn't use the choice of a single authority. Instead, it use "lucky" choices of the community of users and developers. Open source anti-spam filters has more chances than proprietary ones. UI developers of self-adapting components will share the most lucky choices and strategies and less user users will be annoyed by bad adaptation.
How is it trelated to cell phones? Same way. But there is one problem. Today the software on my cell is proprietary. I agree that protocol drivers should be proprietary. But I want to install UI in my cell like I do with UI on my Linux desktops. I want to change it, reprogram it and adapt it for my own need and habits. 99% of cell users do not know what I am talking about. Many of desktop users don't understand it either - but they still use the result of work of that 1% users who understand enough to change it and who is motivated enough to do it righ.
Basically, once cell phone UI will be open-sourced - it will adapt and you won't be annoyed.
BTW, why just UI? Perhaps some guy want to right own protocols. But we talked about UI adaptavity.
BSD is already dead, but it's still around.
Last time I've tried FreeBSD (several 4.*) the system required to know BSD and ports just to install it from the first time. The documnetation was terrible (recursively written for people who already know FreeBSD), forums answers mostly "RTFM". The building of my customized BSD kernel was just a nightmare comparing to what I used to do with Linux kernel (despite the fact that it didn't support everything I had in my PC, which was supported fine by Linux kernel).
If some OS is not for newbies - that would be *BSD. After Linux/Solaris/AIX/NT multi-year experience I was screwed up with FreeBSD. The first impression was: why this OS is so unfriendly if loose many functionality and drivers jumping from Linux to BSD?
But what I liked in FreeBSD was ports (eventually I've been learning FreeBSD anyway, partially just learn ports). So, when I found a Linux distro with the package management system like ports (right - Portage), I immidiately fired all my RH, YDL and FBSD installations (one by one actually), installed Gentoo and now I don't see the chance that any better distro will come around any time soon.
Well, better than Gentoo? There is such OS already, it's new Gentoo! Despite the fact that Gentoo is already the best OS, it's constantly being improved, enhanced, advanced and further developed.
The good point of XML is its unified syntax - you don't need a new parser each time, you already have one and you concentrate on semantics rather than syntax.
That's why there are such projects as XML representation of FlatCurry. Similar implementations I've met also in mail-lists about Mercury (BTW, they have also discussed S-exps as a basis of new syntax). Even ASN.1 goes to XML.
Again, I think XML tags are worse readable (and less compact) than Lisp S-expressions. Thast's why I believe that eventualy the market attraction of Lisp-like languages will tise again.
What make Lisp's parentheses to be unreadable while XML parentheses to be still readable? By my opinion, Lisp parentheses are much more readable and that's the thing I miss in any other languages.
How about XML?
Besides, every language has barackets. Just in Lisp all bracket are unified to be as (). Other languages use various forms of brackets: (), {}, [], ";", "new-line". Python use even space identation instead of brackets (the only thing I hate in Python!).
When the code is the data, unified brackets are much better for verification purposes. That's why XML also has unified brackets (just badly designed, IMHO).
Dead system for dead people :)
Check 'em out!
Some one should check 'em in before another one can check 'em out. :)
I used ICQ and IRC in Emacs - it's been working nice for me. Very good environment to create "semi-human semi-bots" :)
The best ORDBMS I found today is PostgreSQL, which has one more very important feature implementatio: it's very programmable database - it has ability to write stored functions on scripting languages and you can extend the database with new interpreters. That helps programmers to store the bahavior (functional data).
Wrong. What is the port number doesn't really matter. How the protocol is organized matters. CORBA originally has been designed with two problems:
- Syncronous communication - asyncronous (fire and forget) messaging is still in dreams of real-life CORBA developers;
- Tight coupling - both client and server has to really have the same knowledge about the shared (passed, called) object;
CORBA might be not a bad choice for prototyping at earlier days of GNOME, but today it must begone. I think it must be replaced by SOAP. So, let's wait for MONO becoming stable, then let's get rid off gnorba and then let's compare GNOME with then's KDE.2. An input device takes a key based not only on static image, but also on micro-motions of your eye, which are also uniue. So don't worry, your dead eye cannot be used either.
3. Whole software of the authorization system with DB is useless to steal as it won't work without the hardware system key, which value is used to generate runtime system certificates. In order to make useless the stealing of the whole building the system is usually geographically distributed.
4. The input device is useless to steal as it is connected in the LAN to similar hardware key. Besides, it can take a picture of the room around and won't work if it's changed too much.
5. The only problemis that is left is about loosing your eye. Then you have to go through all bank offices and re-register yourself to use the image of your pennis or any other unique part of your body. But that is offtopic of your question.
The list above is cross-platform. Of course there are some differences between such programs for Windows, MacOS, but kids won't notice them and won't understand them either. It's too early for that - they just see computer almost first time in their life.
After using Macs in schools ("using": see the list above), kids go to business and they do the same with Windows. Again, there are few differences between MacOS and Windows, but they are compensated by differences in task lists as well.
So, what's the result? Apple helps to Microsoft. Seriously. Apple sponsor schools with Macs. Students educate how to use computers. They come to business and buy a cheaper commodities: wintel products.
Now, how to make sure that Linux in schools will not help to wintel sales, but instead it will help to Linux sales (or Linux downloads)? Simple. Don't teach in schools how to use same functions on computers as they are on windows. Instead teach how to use computers in a way that impossible or difficult on windows (and macos). The keywords here are CLI, Scripting and Security. Briefly, modern regular (non-professional CS) user must know something about what is inside PC/OS, how to program for personal needs, and how to protect own informational life.
Command Line Interface is not a relict of 70s, it's a basement of control. Do you want to control your OS yourself? - Use CLI. Other wise let GUI (with bugs) to help you to think that you control your system. CLI is a manual-stick car (no automatic transmission). If a student will know how and why to use CLI - there will be less problems with GUI usage as well. CLI is key to know the computer. Isn't it a goal for students to use computers in schools?
Modern scripting languages are not CLI macro agreements. Look at Python: it has a bytecode compiler, a garbage collector and it's OOP. How is it different from Java? Only in a good sense, mostly speed and size. Python lets a student to write rapidly and interactively small scripts accordingly to tasks from a text book. Results work right the way and often faster (no monster-JVM).
Many computer science experts will support me when I will tell: don't teach imperative, teach functional. If a student know how to use FP, there will be no problem with imperative languages. Otherwise is not true: many students after learning imperative languages in fact do not know how to program. By the way, Python has FP primitives. Besides Python, I would recommend students to use Scheme, Lisp, OCAML, Haskell and Prolog. And of course SQL.
You may notice that future non-programmers do not have to know so many langs. You're right. But they have to know at least one of them. Please, don't let Java or C or C++ to be one of them - they will learn nothing but useless syntax. Non-programmers are not going to use Java at home. As for scripting langs - no problem. That's what's happen to Visual Basic. But we don't want VB, we want something Linux-friendly, MS-independent and math-better.
Security is another key. More and more schools today consider of teaching Karate or other self-defense disciplines. Especially in urban areas. Today we are all living in one big cyber-urban area - Internet. Are you going to trust ISP or you are going to protect yourself?
I propose that schools will teach students how to hack each other on special hacking labs. If you know how to hack - you have more chances how to protect. Otherwise is not true. Today most of internet users still do not know that blind executing of unknown email attachments is most likely a virus infection. They don't know how to use MUA filters against the spam. No need to mention they don't know that using PGP or GPG signing/encryption can protect their private information in email messages as well as to protect against the spam. Why? B/c none taught them. You can tell all of tat to them hundreds of times. But until they will learn on practice how to win the war - they will continue to loose it.
Same story here. Almost. Same beginig, but with a different end. I've got Gentoo/PPC installed on my Powerbook. Same Gentoo ason my other PCs and PPCs. As a result I've got consistent environment (Gentoo 1.4) across several PC and PPC computers, saving my time (very valuable) on admin efforts. Besides, programming in such environment is just the pleasure.
Gentoo/PPC runs on that Powerbook faster than OSX. By the way, running MOL for the rest of few Mac (not open sourced) apps is much more stable than attempts to run Linux apps on OSX.
Anyway. They are already dead. Only zombi use them. The rest of the world just ignores them. So do I.
Palladium is a very heavy metal. It may drag you down :)
But it won't make [micro]BSD alive :)
Lindows has a problem, b/c the giant doesn't like the sound of it. Was it the reason here? Many people might be confused: MicroBSD = MICROsoft BSD.
You do need Linux kernel to work all time as it handles processes and it has drivers for hardware.
But as other noticed (honestly - I forgot about it), the kernel still can (and does) use those BIOS functions which unify the usage of floppies and hard-disks and video.
Another function of BIOS was (and still is) to give the chance to configure some hardware CMOS parameters. Again, DOS was not able to do that. With Linux kernel you have NVRAM driver, which can change most of such parameters right from OS. I don't know any parameters, which either today or in near future would be unable to be set from a NVRAM driver of the OS kernel.
The main reason of changing those parameters was inflexibility. But today nonody will set IDE disk parameters - instead they used AUTO. Same for many other things. Although, some of CMOS might not be automatically tuned up, I am not sure.
The last (from what I can recall right now) reason to still use BIOS is to point to the boot device. Perhaps that the function that will survive longer than other. At least that's the on;y function I still use in Apple's firmware :)