Why "troll"? This anonimous coward is absolutely right.
The B1 certification, other than requiring years to be issued, only certifies a given system with a given hardware and a given configuration.
This means that even the same distribution on the same hardware with only a slight different configuration is no more B1. Even worse for different distributions, which may offer the same functionalities but using sligthly different way to do so.
There is really little use for this kind of certfication in real world, other than for throwing marketing hype to clueless customers, and just raises a false sense of security.
Anyway, SGI's involvement in writing securty patches for Linux deserves gratitude. They are working a lot (and somewhat quietly) to really offer interesting solutions for security, debugging, efficiency. I hope some of their ideas will be incorporated in the main source tree (and not just XFS - when it will be ready for prime time).
I can't do nothing but hope this is the good time. Although commercial multi-toolkit, multi-platform GUI frameworks have been existing for a long time, either they die out after few years, or offer the common denominator of all the platforms/toolkits they support (which usually is bare bone).
WxWindows may be the right tool at the right time doing the right job, and the fact it's free (as in speak) means it'll probably last longer than any closed-source GUI framework.
Besides, X can be configured to believe that 72 dots is one inch.
Measure how many DPI your monitor does at the resolutions you usually use.
startx -- -dpi xxx
then, when you specify "a point", software is able to correctly approximate it (i.e. when you view a PostScript document with gs and set the zoom to 1x, you see the real dimensions of the document).
Pity you can't indipendently set the horizontal and vertical DPI. This would be useful with the resolutions where a pixel is not a square.
Yes. But Shoutcast is a commercial product, as you can read on the license, and I doubt there is any source available. There are versions for some unices, indeed.
Icecast seems to do a great job and it's under GPL, but it needs an external mp3 encoder. This is the bad point: free ones like BladeEnc and Lame are under pressure by the Fraunhofer patent (at least in USA and Germany).
It's really a pity, because Lame 3.70 has a really good psycoachustic model, encodes at a speed that is twice the real time on a 300MHz system, can also do variable bitrate encoding (slower). It can also perform the re-encoding on the fly to change the bitrate, which is a must when you have to broadcast mp3 ad a certain speed.
But patents hold down all this in the non-commercial use, even in countries where patents on software are not allowed (Europe), due to some legal tricks (I believe).
1. Draw a big circle. 2. Label it "BSD". 3. Draw a little circle inside the big circle. 4. Label it GPL. 5.Draw another little circle inside the big circle, but be very careful to make sure that it doesn't touch the other little circle. 6. Label the second little circle "closed source".
Since all BSD (version 2) softare can be relicensed without notice (even if it not usually the case), I'll draw instead 1 big circle labeled BSD, which intersecates many other circles labeled GPL, LGPL, Artistic, MIT, MPL, QPL, ZPL, etc. Then, a circle labelled "closed source software" that intersecates mainly the BSD circle, but other circles as well.
The result? A pretty complex picture.
So: if you want your software to be used by as many as possible, choose BSD (version 2). If you want your software to stay free, choose GPL. If you want something in the middle, choose something in the middle. But please, stop whining on what's the best license (best for what?) and ask yourself "what do I want to do with my code?" and "what if I want to merge it with some other people's code or someone else wants to?"
Just a little note: while there is a Linux port of lsof, the tool on Linux to investigate about open files, sockets, etc. is definitively fuser, which IIRC should have every functionality offered by lsof (and more), and it's installed by default (while lsof is not).
Actually you can create a foo.h file that includes everything you need, and parse it with gcc -E -dD foo.h > foo.ph. This outputs the preprocessed code with all the definitions, so when you include it in every file you need, the preprocessor has to do a lot less work.
Many of the incremental linking benefits may be obtained with shared libraries, as described in this article from Dr. Dobbs Journal.
Of course, I don't know how this would require changes in the Corel's codebase, and probably Wine had to be used anyway, so... but, as you see, these are not real issues with new projects.
Did you read it correctly the first time? Propably not.
Are you sure there isn't any sort of side effect? Ok, yours was only a common spell error, but it ironically illustrates the idea.
Ads everywhere? Think of it as some sort of "ad pollution". You can breathe polluted air for 10 years and survive, but in the end it will kill you. Pervasive Ad probably won't kill you, but surely doesn't make you feel better.
I wouldn't exactly like the idea of my subconscious being tickled all the time, even if this makes me save some money (perhaps). OTOH, if it becomes popular, contracts for cellular phones without ads are going to be more expensive in the long term (grrr...).
I can only praise the inclusion in XFree86 of the JMK fonts as a standard. They are by far more readable than every other fixed-width font, and comes in a well assorted number of encodings. Since I adopted them everywhere, my eye strain greatly reduced.
FYI, I use a -jmk-neep-medium-r-normal--20-*-*-*-c-*-iso8859-1 on 1280x1024 17" displays. The `&' may look bizarre at first, but it's definitively impossible to mistake it for an `8' anymore (at every font size).
I don't know, but why not? Even my Psion series 5 comes with a TCP/IP stack, even if not fully implemented (only one interface allowed).
> Does it use X Windows?
I'd think it uses something based on the MicroWindows - Nano X projects (see http://microwindows.censoft.com/). The latter offers an Xlib compatible API, so using standard toolkits with them shouldn't be a particular problem (well, it's a problem when it cames to greyshades...)
> I guess you did not have enough pain fixing other people's code
Strangely enough, it's exactly what have been doing in the last two years and a half: maintenance of other people's C++ code.
And still I'm grateful that I have every dirtiness of C++ at hand to avoid myself major rewritings in the night, and if things are broken, I could always throw in another upper layer until the bottom one is rewritten. Cyclic developement. Usually works.
Personally, I believe that Java allows one to avoid only simple mistakes (syntax, style) that one sorts out in 6 months or so, not major ones (design), which takes years of experience. With the features of C++ and the existing tools (mostly Purify) I'm able to circumnvent quite easily even serious mistakes, like a sort of swiss knife that you may cut yourself with, but usually helps a lot to get working things delivered within schedule. I don't seriously believe that Java would be of great help here (unless I'd run every source through m4, but that's another story).
> You would be absolutely right if the world would only consist of code masters
I'm for an impact therapy: put newbie developers in front of their responsibilities, expecially when they work in a team of 10+ people with all of them needing to be able to read each other's code. Don't ever put them in the condition of feeling inadequate for the job: if the task is too complex for them, choose a simplier one that would fit.
Submit them to peer reviews, the sooner the better.
Suddenly, one quickly realizes that fiddling with new paradigms/styles/etc. which he is not absolutely sure and self-confident about has necessarily to be done at home in spare time if one really wants to... not at work. Slow? Yes, but gives better results in the long term.
But perhaps in the USA being on salary is not convenient anymore... and as a freelance programmer one has to put out something working - no matter how he did it - since the very start. You see what's wrong with that philosophy? One learns to code as fast as possiblle without regarding design (and sane grocking over things before writing code) and maintenance issues.
But hearing that from an hardware manifacturer & vendor (Sun) makes me a little suspicious. I know that when UNIX has been rewrote in C from assembly, a 30% of performance was lost in the way, but the advantages in portability and maintenance were huge compared to that. Does Java offer equal advantages? As today, I wouldn't be so sure.
>Is [] just an array or is an overloaded operator that makes some non-obvious assumptions?
"obj.grokThisThatWay()" does really do what it says to do? Does it says anything useful about its invariants? What keeps one using misleading names for symbols keeps also a C++ programmer to abuse operator overloading. If the object on which you overload the [] operator is not assimilable to a collection of some sort and what's in between the square brackets is not an index or a key of some sort used to retry a value from that collection, you interface is flawed no more and no less than if you called all your methods "foo()", "bar()" and "baz()", perhaps with misleading comments. And you won't notice it until it happens to you to use them.
> With MI if I hadd something to one of the base class, what happens especially if there are conflicting names?
MI is mainly used exactly as in Java and exactly as Bjarne said: as a way to implement multiple interfaces (abstract classes made only of pure virtual methods). Only you don't need two additional keywords to do that. If you want to use it in any other way, it's your problem, not mine.
And if someone points out that it's inherently leads away from code reuse, I'd reply that throwing code out of the window from time to time is healthy, because there are few designs that lasts immutate for 10+ years (unless you spend in absurde quantities of money and it happens to you to have a monopoly of some sort, or you are really clever). Until today, Java made that period even shorter for reasons that are not related with software design, but with the design of Java itself.
> It is almost difficult to make a Java program that is difficult to understand
It's equally difficult to make a C++ program that is difficult to understand, unless the design is really, really bad. But, for now, it's easier to get stuck with Java than with C++.
+++
Sometimes, it seems to me that some Java enthusiasts are seeking (with pain) for the One True Coding Style That Fits For All Useful Purpouses(TM), which is obviously hopeless, and incessantly compare Java with other languages not by its merits, but for What They Assume It Is Flawed With Language Foo Because They Instinctively Would Use It In A Flawed Way.
Relying on features missing (or imposed) don't seems to me a good strategy in the long time. Strangely enough no one ever mentions a real flaw of ANSI C and C++: guess what? Trigraphs, of course. They are there. Does anybody actually uses them? No, of course. You see, you don't need to use every feature a language offers if you don't need it. A so-called programmer should be able to distinguish by himself the situations in which a certain feature is useful, otherwise it wouldn't be a programmer, but a brainless code droid.
Really, I used to have a predilection for Java in the past. Now I regard it as nothing more than Another Respectable Language Which Has Its Grade Of Suckness Like Every Other Language Out There. 'nuff said.
A resizable ramdisk doesn't really make sense on OS with good disk-cache management.
> and an ENV:
Persistent environmnent variables were a poor kludge compared to what you get with any Unix-like system.
> I want straight-forward no-fuss super-efficient pixel-editors for low-color-(web!)graphics
Admiteddly, the Gimp is a big step forward WRT Dpaint 4. And you still are able to do editing of low-color web graphics...
> want free program downloads of a few hundred K with 25 languages included.
Never considered using GNU gettext for that? Or the X/Open catalogs (built-in the the GNU libc?)
> I want to script my applications from any other application!
Well, scheme is becoming the de facto standard for every application for which scripting makes sense.
>I want plug and play that work
Amiga was a well defined architecture. You could count on the fact that certain hardware was there. A today's intel box is more like the Frankenstein's monster, so doing PnP is somewhat more difficult.
> I want assigns
Just use symlinks. Perhaps use devfs. What's the problem?
> I want SEPARATE chips to handle all computer IO
You really believe that your video card and your sound card and your chipset are absolutely dumb? Uh?
> I want sensible file dialogs
Yes. The GTK guys should just copy that. Definitively. But the current GTK file window is not really far from that (for mere functionalities).
> I want hardcore hand-optimized assembler demos with breathtaking effects on an 7MHz 68K.
Yes. I'm missing them too.:-)
Perhaps you forget to mention that in addition to that you want protected memory and virtual memory, because that are something that only embedded systems could do without today. And yes, I remember GigaMem.
FYI, there is already a nice clone of Directory Opus, called Worker.
WRT porting Directory Opus to *nix, Amiga environment is so different from the typical *nix environment that probably it's easier to rewrite software like this from scratch.
But then, the release of the sources of Directory Opus will surely make Amiga users (and UAE ones) really happy, anyway.
> This brings up a good point: do you internationalize the source code?
Just my 0.02 Euro here.
I'm Italian, and I live in Italy. I'm just a C/C++ programmer like many others. The source code I write has all the comments and symbols names in English. This mainly for 2 reasons:
1) Technical language: programming needs a whole new class of technical terms. Simple words like `formatting', `font', etc. didn't have equivalents in Italian when they were introduced to the mass in the former '80. Simply, we adopted (and distorted) the usual English words. Attempts to create a brand new tech gerg just failed. After 20 years the mass starts using the Italian form for "directory", but still it's a minorty. Just figure yourself when it comes to source code...
2) Technical documentation: good technical documentation is written in English. Programmers like me avoid translations, just because they make life more complex (the effort is doubled: first you read it, then you try to figure out how the terms used are related to the well known English terms). I'm translating the GNU Emacs manual in Italian in my spare time, and it's a damn difficult work (and it's not at all about programming - figure it yourself when it comes to programming books).
3) I18n: English is good for symbol names, just because you don't have to do with accents. Italian words just look ugly without accents, and using ISO-8859-1 is still not a viable option for sources...:-)
4) Last but not least: English is the de facto `lingua franca' for programmers, just like Latin was for the europeans in the passed ages. If you start writing your sources using one of the 18 languages of India for comments and symbol names, probabilities are that only indians will ever peek into them, just taking out the rest of the world. And since most people giving help and advice on programming out there speak English well enough to make one understand them (perhaps except Alan Cox, sometimes;-), probabilities are that one's best choice is (for now) learn English.
Of course, I just use Italian in every other context, just like everyone else...
It does not work with some versions of Communicator, because in order to update the cookies info it first deletes the cookie.txt file and then rewrites it (recreating it as a normal text file), instead of truncating it to length zero and then appending.
The result is a freshly created cookies.txt.
Al least, this was the behaviour I experienced with some versions. So check if it's your case.
> Is it possible for virus code loaded at boot time to survive the boot up process and then continue to run as root?
The linux kernel takes over everything, so basically there shouldn't be any possibility for extraneous code to survive the boot process.
Of course, you could "infect" in some way a kernel image, but one has to be already root in order to do that, so basically it's pointless.
The problems arise when the user acts as a dumb monkey *as root*. There's little that you could do when some of your fundamental binaries are replaced by trojans (think of/bin/ls being replaced by trojan of some sort...).
So:
1) Don't take it too easy when you download something precompiled. Those md5sums are there for a reason, so use them!
2) Don't run as root when you don't need to. Use utilities like "sudo" only on trusted binaries.
3) Don't install something as root if you don't need to (in fact, there's little that actually NEEDS to be installed as root, and pratically nothing that actually NEEDS to stay necessarily in/usr/{bin/lib/sbin}).
The first beta of WordPerfect 8 for Linux was known for producing a possible security hole when installed as root, and the usual advice (until it was corrected) was to create a particular user just for WordPerfect binaries. Not a virus, not a trojan, but just a mistake. You can always do something like this for binaries that you don't trust 100%.
The remaining advice is the usual: make backups of data, make backups of data and still make backups of data. And possibly make a backup of your configuration files (not binaries, because reinstalling a Linux distribution from scratch generally doesn't take more that 1 hour, which shouldn't be a problem in a home environment). A CD recorder is something cheap enough that could do well today for home backups, if you can't afford a more expensive tape streamer.
Keep in mind these basic principles, and trojans will stay away from you for a loooooong time.
I'm not sure I understand what is so terribly difficult for someone in that company to simply recompile the modules for every kernel rev that comes out. It's not like that we have a new kernel rev every day. Gee, how terrible: someone actually has to type 'make' every other month, or so
Actually, things depend also on the kernel configuration.
For example, modules compiled for SMP kernels won't work on non-SMP kernels and vice-versa, and this is not the only issue. Someone on the kernel mailing list estimated that one should offer something in the order of tens of different binaries for a single module for various configurations.
It shouldn't be too problematic to put up a page with a form to choose different configurations for some kernel versions (and perhaps compile it on the fly before downloading), but it's still a pain for the inexperienced user who don't know even what a module is loaded in first instance and expects its installation process to just recognize as much hardware as possible, otherwise he gets stuck.
> When they came out they had a unique GUI that was stable, supported multitasking, used millions of colors
As an ex Amiga user and developer, I'd like to do just a correction, for the sake of what's true: whey they come out (A1000) they had 256Kb ram, 4096 colours but not at the same time, no protected memory, no virtual memory (68000 had no MMU), not even a battery backed-up internal clock.
Any pointer gone wild or every memory leak brought down the system in seconds. Strangely enough, this resulted in really bug-free applications, which is somewhat unusual for a personal computer.
But then, aside from that, Amiga's main advantages were in its hardware (which was cheap, for the time, and powerful) and its community (Fred Fish comes to mind), fairly balanced by the badness of Commodore's marketing department (which insisted on selling it as some sort of evoluted gaming console).
So, let's watch if something interesting comes out (at last, QNX now has another appealing set of widgets). But I definitively don't hold my breath.
Slashdot Mirror
The B1 certification, other than requiring years to be issued, only certifies a given system with a given hardware and a given configuration.
This means that even the same distribution on the same hardware with only a slight different configuration is no more B1. Even worse for different distributions, which may offer the same functionalities but using sligthly different way to do so.
There is really little use for this kind of certfication in real world, other than for throwing marketing hype to clueless customers, and just raises a false sense of security.
Anyway, SGI's involvement in writing securty patches for Linux deserves gratitude. They are working a lot (and somewhat quietly) to really offer interesting solutions for security, debugging, efficiency. I hope some of their ideas will be incorporated in the main source tree (and not just XFS - when it will be ready for prime time).
My 0.02 Euro.
I can't do nothing but hope this is the good time. Although commercial multi-toolkit, multi-platform GUI frameworks have been existing for a long time, either they die out after few years, or offer the common denominator of all the platforms/toolkits they support (which usually is bare bone).
WxWindows may be the right tool at the right time doing the right job, and the fact it's free (as in speak) means it'll probably last longer than any closed-source GUI framework.
Pity you can't indipendently set the horizontal and vertical DPI. This would be useful with the resolutions where a pixel is not a square.
Yes. But Shoutcast is a commercial product, as you can read on the license, and I doubt there is any source available. There are versions for some unices, indeed.
Icecast seems to do a great job and it's under GPL, but it needs an external mp3 encoder. This is the bad point: free ones like BladeEnc and Lame are under pressure by the Fraunhofer patent (at least in USA and Germany).
It's really a pity, because Lame 3.70 has a really good psycoachustic model, encodes at a speed that is twice the real time on a 300MHz system, can also do variable bitrate encoding (slower). It can also perform the re-encoding on the fly to change the bitrate, which is a must when you have to broadcast mp3 ad a certain speed.
But patents hold down all this in the non-commercial use, even in countries where patents on software are not allowed (Europe), due to some legal tricks (I believe).
My 0.02 Euro.
Since all BSD (version 2) softare can be relicensed without notice (even if it not usually the case), I'll draw instead 1 big circle labeled BSD, which intersecates many other circles labeled GPL, LGPL, Artistic, MIT, MPL, QPL, ZPL, etc. Then, a circle labelled "closed source software" that intersecates mainly the BSD circle, but other circles as well.
The result? A pretty complex picture.
So: if you want your software to be used by as many as possible, choose BSD (version 2). If you want your software to stay free, choose GPL. If you want something in the middle, choose something in the middle. But please, stop whining on what's the best license (best for what?) and ask yourself "what do I want to do with my code?" and "what if I want to merge it with some other people's code or someone else wants to?"
My 0.03 Euro. Damned inflation.
Hmmm. No. After all you are reading, and shaking you display in order to read further sounds really awkward at least.
> Lsof is a Unix-specific diagnostic tool.
Just a little note: while there is a Linux port of lsof, the tool on Linux to investigate about open files, sockets, etc. is definitively fuser, which IIRC should have every functionality offered by lsof (and more), and it's installed by default (while lsof is not).
Actually you can create a foo.h file that includes everything you need, and parse it with gcc -E -dD foo.h > foo.ph. This outputs the preprocessed code with all the definitions, so when you include it in every file you need, the preprocessor has to do a lot less work.
Many of the incremental linking benefits may be obtained with shared libraries, as described in this article from Dr. Dobbs Journal.
:-)
Of course, I don't know how this would require changes in the Corel's codebase, and probably Wine had to be used anyway, so... but, as you see, these are not real issues with new projects.
Just my 0.03 Euro (damned inflation
Perhaps something like Wotsit, but without the offsite links?
the springtime.
Did you read it correctly the first time? Propably not.
Are you sure there isn't any sort of side effect? Ok, yours was only a common spell error, but it ironically illustrates the idea.
Ads everywhere? Think of it as some sort of "ad pollution". You can breathe polluted air for 10 years and survive, but in the end it will kill you. Pervasive Ad probably won't kill you, but surely doesn't make you feel better.
I wouldn't exactly like the idea of my subconscious being tickled all the time, even if this makes me save some money (perhaps). OTOH, if it becomes popular, contracts for cellular phones without ads are going to be more expensive in the long term (grrr...).
FYI, I use a -jmk-neep-medium-r-normal--20-*-*-*-c-*-iso8859-1 on 1280x1024 17" displays. The `&' may look bizarre at first, but it's definitively impossible to mistake it for an `8' anymore (at every font size).
> Does it come with a TCP/IP stack?
I don't know, but why not? Even my Psion series 5 comes with a TCP/IP stack, even if not fully implemented (only one interface allowed).
> Does it use X Windows?
I'd think it uses something based on the MicroWindows - Nano X projects (see http://microwindows.censoft.com/). The latter offers an Xlib compatible API, so using standard toolkits with them shouldn't be a particular problem (well, it's a problem when it cames to greyshades...)
> So lets not standardize the GUI, but lets have Linus (or CmdrTaco) set some standards for the UI
Why not you? Put up an initiative.
Defining standards is always someone's else job...
> I guess you did not have enough pain fixing other people's code
Strangely enough, it's exactly what have been doing in the last two years and a half: maintenance of other people's C++ code.
And still I'm grateful that I have every dirtiness of C++ at hand to avoid myself major rewritings in the night, and if things are broken, I could always throw in another upper layer until the bottom one is rewritten. Cyclic developement. Usually works.
Personally, I believe that Java allows one to avoid only simple mistakes (syntax, style) that one sorts out in 6 months or so, not major ones (design), which takes years of experience. With the features of C++ and the existing tools (mostly Purify) I'm able to circumnvent quite easily even serious mistakes, like a sort of swiss knife that you may cut yourself with, but usually helps a lot to get working things delivered within schedule. I don't seriously believe that Java would be of great help here (unless I'd run every source through m4, but that's another story).
> You would be absolutely right if the world would only consist of code masters
I'm for an impact therapy: put newbie developers in front of their responsibilities, expecially when they work in a team of 10+ people with all of them needing to be able to read each other's code. Don't ever put them in the condition of feeling inadequate for the job: if the task is too complex for them, choose a simplier one that would fit.
Submit them to peer reviews, the sooner the better.
Suddenly, one quickly realizes that fiddling with new paradigms/styles/etc. which he is not absolutely sure and self-confident about has necessarily to be done at home in spare time if one really wants to... not at work. Slow? Yes, but gives better results in the long term.
But perhaps in the USA being on salary is not convenient anymore... and as a freelance programmer one has to put out something working - no matter how he did it - since the very start. You see what's wrong with that philosophy? One learns to code as fast as possiblle without regarding design (and sane grocking over things before writing code) and maintenance issues.
> But remember that for the first 8 - 10 years of C++
Java: a language that grows as its user base becomes more and more demanding...
C++: the same, only 8-10 years older.
Sorry, I can't afford to wait 8-10 years to find myself more or less where I am now, only with a different language.
> Speed of execution isn't always important
Of course, fortunately.
But hearing that from an hardware manifacturer & vendor (Sun) makes me a little suspicious. I know that when UNIX has been rewrote in C from assembly, a 30% of performance was lost in the way, but the advantages in portability and maintenance were huge compared to that. Does Java offer equal advantages? As today, I wouldn't be so sure.
>Is [] just an array or is an overloaded operator that makes some non-obvious assumptions?
"obj.grokThisThatWay()" does really do what it says to do? Does it says anything useful about its invariants? What keeps one using misleading names for symbols keeps also a C++ programmer to abuse operator overloading. If the object on which you overload the [] operator is not assimilable to a collection of some sort and what's in between the square brackets is not an index or a key of some sort used to retry a value from that collection, you interface is flawed no more and no less than if you called all your methods "foo()", "bar()" and "baz()", perhaps with misleading comments. And you won't notice it until it happens to you to use them.
> With MI if I hadd something to one of the base class, what happens especially if there are conflicting names?
MI is mainly used exactly as in Java and exactly as Bjarne said: as a way to implement multiple interfaces (abstract classes made only of pure virtual methods). Only you don't need two additional keywords to do that. If you want to use it in any other way, it's your problem, not mine.
And if someone points out that it's inherently leads away from code reuse, I'd reply that throwing code out of the window from time to time is healthy, because there are few designs that lasts immutate for 10+ years (unless you spend in absurde quantities of money and it happens to you to have a monopoly of some sort, or you are really clever). Until today, Java made that period even shorter for reasons that are not related with software design, but with the design of Java itself.
> It is almost difficult to make a Java program that is difficult to understand
It's equally difficult to make a C++ program that is difficult to understand, unless the design is really, really bad. But, for now, it's easier to get stuck with Java than with C++.
+++
Sometimes, it seems to me that some Java enthusiasts are seeking (with pain) for the One True Coding Style That Fits For All Useful Purpouses(TM), which is obviously hopeless, and incessantly compare Java with other languages not by its merits, but for What They Assume It Is Flawed With Language Foo Because They Instinctively Would Use It In A Flawed Way.
Relying on features missing (or imposed) don't seems to me a good strategy in the long time. Strangely enough no one ever mentions a real flaw of ANSI C and C++: guess what? Trigraphs, of course. They are there. Does anybody actually uses them? No, of course. You see, you don't need to use every feature a language offers if you don't need it. A so-called programmer should be able to distinguish by himself the situations in which a certain feature is useful, otherwise it wouldn't be a programmer, but a brainless code droid.
Really, I used to have a predilection for Java in the past. Now I regard it as nothing more than Another Respectable Language Which Has Its Grade Of Suckness Like Every Other Language Out There. 'nuff said.
My 0.02 Euro.
> Even more unix like: you can now mount a Hard Drive in to the directory tree of a different Hard Drive
IIRC it has been possible since DOS 3.30, using the assign.com in the standard distribution, but Windows {2,3,95}.x never liked it.
> I want a RamDisk
:-)
A resizable ramdisk doesn't really make sense on OS with good disk-cache management.
> and an ENV:
Persistent environmnent variables were a poor kludge compared to what you get with any Unix-like system.
> I want straight-forward no-fuss super-efficient pixel-editors for low-color-(web!)graphics
Admiteddly, the Gimp is a big step forward WRT Dpaint 4. And you still are able to do editing of low-color web graphics...
> want free program downloads of a few hundred K with 25 languages included.
Never considered using GNU gettext for that? Or the X/Open catalogs (built-in the the GNU libc?)
> I want to script my applications from any other application!
Well, scheme is becoming the de facto standard for every application for which scripting makes sense.
>I want plug and play that work
Amiga was a well defined architecture. You could count on the fact that certain hardware was there. A today's intel box is more like the Frankenstein's monster, so doing PnP is somewhat more difficult.
> I want assigns
Just use symlinks. Perhaps use devfs. What's the problem?
> I want SEPARATE chips to handle all computer IO
You really believe that your video card and your sound card and your chipset are absolutely dumb? Uh?
> I want sensible file dialogs
Yes. The GTK guys should just copy that. Definitively. But the current GTK file window is not really far from that (for mere functionalities).
> I want hardcore hand-optimized assembler demos with breathtaking effects on an 7MHz 68K.
Yes. I'm missing them too.
Perhaps you forget to mention that in addition to that you want protected memory and virtual memory, because that are something that only embedded systems could do without today. And yes, I remember GigaMem.
Don't be too nostalgic. Have a nice day.
WRT porting Directory Opus to *nix, Amiga environment is so different from the typical *nix environment that probably it's easier to rewrite software like this from scratch.
But then, the release of the sources of Directory Opus will surely make Amiga users (and UAE ones) really happy, anyway.
My 0.02 Euro, as usual.
> This brings up a good point: do you internationalize the source code?
:-)
;-), probabilities are that one's best choice is (for now) learn English.
Just my 0.02 Euro here.
I'm Italian, and I live in Italy. I'm just a C/C++ programmer like many others. The source code I write has all the comments and symbols names in English. This mainly for 2 reasons:
1) Technical language: programming needs a whole new class of technical terms. Simple words like `formatting', `font', etc. didn't have equivalents in Italian when they were introduced to the mass in the former '80. Simply, we adopted (and distorted) the usual English words. Attempts to create a brand new tech gerg just failed. After 20 years the mass starts using the Italian form for "directory", but still it's a minorty. Just figure yourself when it comes to source code...
2) Technical documentation: good technical documentation is written in English. Programmers like me avoid translations, just because they make life more complex (the effort is doubled: first you read it, then you try to figure out how the terms used are related to the well known English terms). I'm translating the GNU Emacs manual in Italian in my spare time, and it's a damn difficult work (and it's not at all about programming - figure it yourself when it comes to programming books).
3) I18n: English is good for symbol names, just because you don't have to do with accents. Italian words just look ugly without accents, and using ISO-8859-1 is still not a viable option for sources...
4) Last but not least: English is the de facto `lingua franca' for programmers, just like Latin was for the europeans in the passed ages. If you start writing your sources using one of the 18 languages of India for comments and symbol names, probabilities are that only indians will ever peek into them, just taking out the rest of the world. And since most people giving help and advice on programming out there speak English well enough to make one understand them (perhaps except Alan Cox, sometimes
Of course, I just use Italian in every other context, just like everyone else...
I hope this is a bit clarifying.
> ln -s /dev/null cookies.txt
It does not work with some versions of Communicator, because in order to update the cookies info it first deletes the cookie.txt file and then rewrites it (recreating it as a normal text file), instead of truncating it to length zero and then appending.
The result is a freshly created cookies.txt.
Al least, this was the behaviour I experienced with some versions. So check if it's your case.
My 0.02 euro.
Mmm... Now I see it! A conversation, 50 years from now:
- 1.tar.gz"
;-)
-- "Hi Scott, any news today today?"
-- "Oh, yes. Just a file, anyway".
-- "Anything interesting?"
-- "Look. I'm extracting it just now... well, here's the README... ok, it's GPL"
...time passes...
-- "Uhm, maybe it's useful. What was the URL?"
-- "iftp://iftp.BrainsRus.org/pub/apps/gpl/rms_20.5a
That is, RMS is finally able to release its whole brain under GPL
Just a joke, couldn't resist.
---
[1] IFTP: Insanely Fast Transfer Protocol.
> Is it possible for virus code loaded at boot time to survive the boot up process and then continue to run as root?
/bin/ls being replaced by trojan of some sort...).
/usr/{bin/lib/sbin}).
The linux kernel takes over everything, so basically there shouldn't be any possibility for extraneous code to survive the boot process.
Of course, you could "infect" in some way a kernel image, but one has to be already root in order to do that, so basically it's pointless.
The problems arise when the user acts as a dumb monkey *as root*. There's little that you could do when some of your fundamental binaries are replaced by trojans (think of
So:
1) Don't take it too easy when you download something precompiled. Those md5sums are there for a reason, so use them!
2) Don't run as root when you don't need to. Use utilities like "sudo" only on trusted binaries.
3) Don't install something as root if you don't need to (in fact, there's little that actually NEEDS to be installed as root, and pratically nothing that actually NEEDS to stay necessarily in
The first beta of WordPerfect 8 for Linux was known for producing a possible security hole when installed as root, and the usual advice (until it was corrected) was to create a particular user just for WordPerfect binaries. Not a virus, not a trojan, but just a mistake. You can always do something like this for binaries that you don't trust 100%.
The remaining advice is the usual: make backups of data, make backups of data and still make backups of data. And possibly make a backup of your configuration files (not binaries, because reinstalling a Linux distribution from scratch generally doesn't take more that 1 hour, which shouldn't be a problem in a home environment). A CD recorder is something cheap enough that could do well today for home backups, if you can't afford a more expensive tape streamer.
Keep in mind these basic principles, and trojans will stay away from you for a loooooong time.
My 0.00001 Euro
I'm not sure I understand what is so terribly difficult for someone in that company to simply recompile the modules for every kernel rev that comes out. It's not like that we have a new kernel rev every day. Gee, how terrible: someone actually has to type 'make' every other month, or so
Actually, things depend also on the kernel configuration.
For example, modules compiled for SMP kernels won't work on non-SMP kernels and vice-versa, and this is not the only issue. Someone on the kernel mailing list estimated that one should offer something in the order of tens of different binaries for a single module for various configurations.
It shouldn't be too problematic to put up a page with a form to choose different configurations for some kernel versions (and perhaps compile it on the fly before downloading), but it's still a pain for the inexperienced user who don't know even what a module is loaded in first instance and expects its installation process to just recognize as much hardware as possible, otherwise he gets stuck.
My 0.02 Euro.
> When they came out they had a unique GUI that was stable, supported multitasking, used millions of colors
As an ex Amiga user and developer, I'd like to do just a correction, for the sake of what's true: whey they come out (A1000) they had 256Kb ram, 4096 colours but not at the same time, no protected memory, no virtual memory (68000 had no MMU), not even a battery backed-up internal clock.
Any pointer gone wild or every memory leak brought down the system in seconds. Strangely enough, this resulted in really bug-free applications, which is somewhat unusual for a personal computer.
But then, aside from that, Amiga's main advantages were in its hardware (which was cheap, for the time, and powerful) and its community (Fred Fish comes to mind), fairly balanced by the badness of Commodore's marketing department (which insisted on selling it as some sort of evoluted gaming console).
So, let's watch if something interesting comes out (at last, QNX now has another appealing set of widgets). But I definitively don't hold my breath.
My 0.02 Euro