C is widely used because it allows the programmer to do exactly what he or she wants with (except in a few cases) the ability to be able to predict how the resultant binary will 'look'.
For example, the in-memory layout of a 'struct' is exactly how the programmer decided it should be - with the exception of padding, which has a well-defined behaviour anyway.
Similarly, the same applies to calling conventions, and to a certain extent, the raw machine code that gets generated.
C++, on the other hand, I hate, becuase it doesn't give you this fine-grained control (for example, the in-memory layout of a class containing virtual methods is largely implementation-defined, I believe).
The majority of the 'other' languages (with the exception of those such as Pascal, FORTRAN and COBOL) generally execute within a VM, which as well as letting you do lots of neat stuff (most of which you can do in C with a little bit of effort and a decent dynamic linker API), it also adds a layer of abstraction which means it's difficult to see how corresponds to assembler output. You're constrained by the VM, meaning that if you want to optimise for a particular CPU or architecture, you need to rebuild the compiler/interpreter/whatever and optimise the VM itself.
Internally, we use autoconf, automake and libtool exclusively. We found the autotools ideal for large projects, although we have made some slightly nonstandard modifications to some of the m4 and Makefile fragments.
Our largest project which uses the autotools is an operating system, based around Linux, glibc and X. All of the components of the OS, with the exception of glibc, build with automake and libtool.
This, at times, wasn't at all easy, but definitely worth the effort. Building a complete OS has never been easier:)
Hasn't the Mitsubishi Trium Mondo (a CE-based PDA/Phone) been available for a while?
As a side note - I recently picked up a Sony CMD-J5 to find it also does SMTP/POP3 in addition to the normal WAP services - and the WAP browser (Microsoft Mobile Explorer) can cope with 'simple HTML' - it's good enough to browse Slashdot in 'light' mode, at least:)
I've watched this brew for some time, speaking as someone who's involved with FSF and has seen arguments fly between the {RMS, Brad Kuhn} collective and Ulrich Drepper.
Quite frankly, I agree wholeheartedly with Ulrich. While I wouldn't personally have put it quite how he did (read his mailing list posts in the past: he doesn't mince his words:-), the sentiment I feel is very valid.
From what I, personally, have seen, Richard Stallman does behave in exactly the way that UIlrich describes: he believes that developers are *privledged* to be writing *their* software, which ultimately is *his* software, because it has a 'GNU' tag at the beginning. In reality, these developers are doing the FSF a favour, and a service, by allowing their hard work to be included in the GNU Project. Stallman rarely sees it this way, and treats inclusion as the GNU Project as his personal license to dictate how the software should be developed, and managed.
I don't like this one little bit, and I wholeheartedly support Ulrich's statements and sentiments.
It's refreshing to see an Ad guy who doesn't seem hell-bent on getting as many impressions as possible through whatever means he can.
I can envisage banner ads changing a little as time goes by.. say, three buttons on each side to "control" the adverts that are delivered to you. Each 'consumer' can have a profile which basically says which ads they like, don't like, and so on. For the consumer, it's ideal - only the 'right' ads will get shown, and being able to turn them off would be wonderful. For the advertisers, it's a dream come true - people build up their *own* demographic profile for you - a simple database query, and you can figure out how many people on Freshmeat like ThinkGeek compared to the number on Slashdot that are interested in GNU/Linux-based hosting.
This, of course, would allow each site to have an 'average profile', based on what its visitors think are the better/more preferred ads - what's nice about this, from a consumer point of view, is that you're not bombarded with the same adverts time after time when you surf between the different OSDN sites.
IMHO, this could be very cool stuff. And I hate banner ads. Normally.;)
I think it depends on whether you make the distinction between 'microprocessor' and 'microcontroller'. Sun obviously do, and I agree with them for the most part.
After all, your Palm V has a hell of a lot more overall resources to it than your average microcontrolled coffee machine.
I don't know.. I mean, credit where credit is due, Sun created Java, but they want it to be an open standard *and* have control at the same time. Call me a cynic, but I don't buy that.
I think that far too much Sun-bashing goes on, granted, but if they didn't act like they are then there probably wouldn't be nearly so much..
Qt adjusts just perfectly to the platform's native look and feel.
From the screenshots, it didn't look like it to me. Admittedly I've done a lot of multi-platform GUI work, and I'm a bit of a perfectionist, but it just didn't "fit in" quite right. Qt seemed to suffer from the common problem where everything is just *slightly* the wrong size. You don't notice it until you see the application alongside others on the same platform, but once you do, it's one of those niggles that becomes an annoyance every time you use the application.
IMHO the attacks against GNOME were pretty out of line too, to say the least..
Well, this is because ``Linux'' is a kernel, and that's what Linus is in charge of. Linux as an operating system (call it GNU/Linux, if you like, I'm not bothered either way, personally), is a whole different kettle of fish. Linus wants to see good things happen for the OS as a whole (I recall a recent post to the linux-kernel list where he talked about UI/end-user stuff), but he's not going to get involved in that arena, I'd imagine for a number of reasons, one primary one being time - there's only so many hours in a day, and so many things you can take part in.
Unfortunately, the media in general (minus a few exceptions) don't make the distinction between the OS, the kernel, or one particular distribution - let alone different kernel versions. Bloomberg was alive today with "A new version of the Linux operating system...", when new versions of the operating system are being released all the time.
It's unfortunate in some ways, because a problem with one aspect of one kernel version in one distribution that gets mis-publicised brings makes the whole arena look bad (which certainly isn't going to help get it onto the desktop).
Aside from mass-education, I can't figure out a good solution for this, so I'll shut up;-)
Okay, after a few replies, there seems to be a fairly obvious trend - Joe Public doesn't really care who gets involved, so long a *somebody* provides the applications that they want. Most Linux distributions include both KDE and GNOME, and for the most part, it boils down to which window manager you prefer - people will run whichever application does the job best, regardless of whether it uses GNOME or KDE.
I think it's about time both parties grew up and realised that a lot of the point of non-proprietry desktop environments is that you can mix and match you applications - and the majority of users do just that. Personally, I don't use either desktop, but run applications from both; I know a lot of other people do the same.
Perhaps now would be a good time for the GNOME and KDE groups to start concentrating on actually developing what they have instead of this continual one-upmanship, because if that's all it boils down to, neither will get anywhere.
Is there *anybody* out there who isn't biased either way and has an opinion on this? If it's not a KDE developer, it's a GNOME developer - how about some END-USER responses for a change?
Well, it could work, if stories were tagged properly. I mean, the majority of the stories are global. How about adding a separate "Regional News" section, as one of the Slashboxes? Maybe make it a little easier to get a story posted in one of those than on the main page, too *grin*
As well as these three, it's interesting to note that RedHat's stock took a turn for the better yesterday when it and IBM annouced that Big Blue would be shipping Linux pre-installed on a range of PCs (obviously shipping it on it's NetFinity range wasn't enough:>). I'd post a URL for the story, but it's easier to tell you how to find it than give the URL (it's horribly long); just take a look at the "Technology News" section on Bloomberg's site.
This document walks you through the process of cross-installing from a Linux box. You can also find the debs in/dists/sid/main/binary-hurd-i386/ on your favourite ftp.debian.org mirror. As far as building stuff from source on HURD goes, it's pretty similar to building for BeOS - it provides what it has to, and doesn't provide a lot of what it doesn't (e.g., the MAX_PATH definition). According to the standards, it's perfectly within it's rights to do this and still be POSIX compliant - so you have to do a bit of filling-in-the-blanks when you're building stuff that expects these things to be there. Most packages that're autoconf'd tend to build fine, and there's a fair selection of binary packages in the Debian distribution.
Slashdot Mirror
C is widely used because it allows the programmer to do exactly what he or she wants with (except in a few cases) the ability to be able to predict how the resultant binary will 'look'.
For example, the in-memory layout of a 'struct' is exactly how the programmer decided it should be - with the exception of padding, which has a well-defined behaviour anyway.
Similarly, the same applies to calling conventions, and to a certain extent, the raw machine code that gets generated.
C++, on the other hand, I hate, becuase it doesn't give you this fine-grained control (for example, the in-memory layout of a class containing virtual methods is largely implementation-defined, I believe).
The majority of the 'other' languages (with the exception of those such as Pascal, FORTRAN and COBOL) generally execute within a VM, which as well as letting you do lots of neat stuff (most of which you can do in C with a little bit of effort and a decent dynamic linker API), it also adds a layer of abstraction which means it's difficult to see how corresponds to assembler output. You're constrained by the VM, meaning that if you want to optimise for a particular CPU or architecture, you need to rebuild the compiler/interpreter/whatever and optimise the VM itself.
My two cents.
Internally, we use autoconf, automake and libtool exclusively. We found the autotools ideal for large projects, although we have made some slightly nonstandard modifications to some of the m4 and Makefile fragments.
:)
Our largest project which uses the autotools is an operating system, based around Linux, glibc and X. All of the components of the OS, with the exception of glibc, build with automake and libtool.
This, at times, wasn't at all easy, but definitely worth the effort. Building a complete OS has never been easier
Hasn't the Mitsubishi Trium Mondo (a CE-based PDA/Phone) been available for a while?
:)
As a side note - I recently picked up a Sony CMD-J5 to find it also does SMTP/POP3 in addition to the normal WAP services - and the WAP browser (Microsoft Mobile Explorer) can cope with 'simple HTML' - it's good enough to browse Slashdot in 'light' mode, at least
I've watched this brew for some time, speaking as someone who's involved with FSF and has seen arguments fly between the {RMS, Brad Kuhn} collective and Ulrich Drepper.
:-), the sentiment I feel is very valid.
Quite frankly, I agree wholeheartedly with Ulrich. While I wouldn't personally have put it quite how he did (read his mailing list posts in the past: he doesn't mince his words
From what I, personally, have seen, Richard Stallman does behave in exactly the way that UIlrich describes: he believes that developers are *privledged* to be writing *their* software, which ultimately is *his* software, because it has a 'GNU' tag at the beginning. In reality, these developers are doing the FSF a favour, and a service, by allowing their hard work to be included in the GNU Project. Stallman rarely sees it this way, and treats inclusion as the GNU Project as his personal license to dictate how the software should be developed, and managed.
I don't like this one little bit, and I wholeheartedly support Ulrich's statements and sentiments.
My two cents, FWIW.
I can envisage banner ads changing a little as time goes by.. say, three buttons on each side to "control" the adverts that are delivered to you. Each 'consumer' can have a profile which basically says which ads they like, don't like, and so on. For the consumer, it's ideal - only the 'right' ads will get shown, and being able to turn them off would be wonderful. For the advertisers, it's a dream come true - people build up their *own* demographic profile for you - a simple database query, and you can figure out how many people on Freshmeat like ThinkGeek compared to the number on Slashdot that are interested in GNU/Linux-based hosting.
This, of course, would allow each site to have an 'average profile', based on what its visitors think are the better/more preferred ads - what's nice about this, from a consumer point of view, is that you're not bombarded with the same adverts time after time when you surf between the different OSDN sites.
IMHO, this could be very cool stuff. And I hate banner ads. Normally. ;)
After all, your Palm V has a hell of a lot more overall resources to it than your average microcontrolled coffee machine.
I think that far too much Sun-bashing goes on, granted, but if they didn't act like they are then there probably wouldn't be nearly so much..
From the screenshots, it didn't look like it to me. Admittedly I've done a lot of multi-platform GUI work, and I'm a bit of a perfectionist, but it just didn't "fit in" quite right. Qt seemed to suffer from the common problem where everything is just *slightly* the wrong size. You don't notice it until you see the application alongside others on the same platform, but once you do, it's one of those niggles that becomes an annoyance every time you use the application.
IMHO the attacks against GNOME were pretty out of line too, to say the least..
Unfortunately, the media in general (minus a few exceptions) don't make the distinction between the OS, the kernel, or one particular distribution - let alone different kernel versions. Bloomberg was alive today with "A new version of the Linux operating system...", when new versions of the operating system are being released all the time. It's unfortunate in some ways, because a problem with one aspect of one kernel version in one distribution that gets mis-publicised brings makes the whole arena look bad (which certainly isn't going to help get it onto the desktop).
Aside from mass-education, I can't figure out a good solution for this, so I'll shut up ;-)
Okay, after a few replies, there seems to be a fairly obvious trend - Joe Public doesn't really care who gets involved, so long a *somebody* provides the applications that they want. Most Linux distributions include both KDE and GNOME, and for the most part, it boils down to which window manager you prefer - people will run whichever application does the job best, regardless of whether it uses GNOME or KDE.
I think it's about time both parties grew up and realised that a lot of the point of non-proprietry desktop environments is that you can mix and match you applications - and the majority of users do just that. Personally, I don't use either desktop, but run applications from both; I know a lot of other people do the same.
Perhaps now would be a good time for the GNOME and KDE groups to start concentrating on actually developing what they have instead of this continual one-upmanship, because if that's all it boils down to, neither will get anywhere.
{/RANT}
Is there *anybody* out there who isn't biased either way and has an opinion on this? If it's not a KDE developer, it's a GNOME developer - how about some END-USER responses for a change?
Well, it could work, if stories were tagged properly. I mean, the majority of the stories are global. How about adding a separate "Regional News" section, as one of the Slashboxes? Maybe make it a little easier to get a story posted in one of those than on the main page, too *grin*
As well as these three, it's interesting to note that RedHat's stock took a turn for the better yesterday when it and IBM annouced that Big Blue would be shipping Linux pre-installed on a range of PCs (obviously shipping it on it's NetFinity range wasn't enough :>). I'd post a URL for the story, but it's easier to tell you how to find it than give the URL (it's horribly long); just take a look at the "Technology News" section on Bloomberg's site.
This document walks you through the process of cross-installing from a Linux box. You can also find the debs in /dists/sid/main/binary-hurd-i386/ on your favourite ftp.debian.org mirror. As far as building stuff from source on HURD goes, it's pretty similar to building for BeOS - it provides what it has to, and doesn't provide a lot of what it doesn't (e.g., the MAX_PATH definition). According to the standards, it's perfectly within it's rights to do this and still be POSIX compliant - so you have to do a bit of filling-in-the-blanks when you're building stuff that expects these things to be there. Most packages that're autoconf'd tend to build fine, and there's a fair selection of binary packages in the Debian distribution.