Call graphs are nice, but call graphs of large applications done via static code analysis are so huge and dense that become useless. Call graphs taken from run-time profile data, with all those irrelevant nodes pruned out, are IMO much more useful, as they naturally direct you to the most interesting parts, where the action is.
I think so too. But I've found that command-line debuggers like gdb invariably show a very limited view of the code, which presents an obstacle for code comprehension.
If you're going to debug with the aim of understanding the code, then using a debugger with a graphical interface is a must. On that regard. I personally find Eclipse CDT very useful, since not only it has a graphical interface to gdb, but also has code navigation abilities which allows you to quickly jump back and forth the function/type definitions and so on.
I'm also using an home grown Python script based on that principle for my homepage. I'm actually in the process of rewriting it to use the more robust HTML and XHMTL parsing of the libxml2 python bindings, and to have even cooler stuff such as Xpath and XSLT under the control of your embedded python scripts.
You can also do the scripting on the client side. The Win32 version ActiveState Python has support for doing that - if you're on that sort of pseudo OS...;-)
Somehow, if they simply reported to MS without asking anything back, and made the explot public after a month, it would be OK? Even if it would bring less benefits for both parties than to make an arrangement?
there are four other kernels, btw, did you check into one of those? Did you have trouble with the default kernel? The process priorities problem was exceedingly rare.
I've been trying several kernels as the development focuses them (gentoo-sources, lolo-sources, I don't know the details about the process priorities bug, even if is unrelated the fact is that doing things with nice -19 (such as emerging a packaged), almost took the machine to a halt - now I can compile a package without that hapening, so it must be a problem in gentoo somewhere. Anyway, this isn't bugs.gentoo.org to write a bug report...
I've also never broken my userland to the extent that I require a reinstall; did you read the part about Gentoo being for power users and being a learning platform? Yeah, if you're coming from RedHat, obviously you're going to screw it up. That's no problem. But be patient until you learn how Linux works, that's what you're there for isn't it?
Ah! Ah! Ah!.... Oooohh... People are so funny (not to mention ridiculous) when they put labels on someone just because he use(s/d) distro A or B.
I always hear people saying "I use Debian because I just want to get work done" but something about it just doesn't ring true. Apt is a fantastic package manager I grant you, but whenever I install Debian I do without debconf. I know what I'm doing, thanks.
Don't see the point of that...
So from my perspective the main thing I'm concerned with is packages. Are they up to date? Are userland packages like KDE or X released promptly when there are new versions? The answer is an unfortunate no. The stable branch of Debian refuses to merge new versions in a dynamic way, instead insisting that stable is stable and you must use unstable to get the latest.
They are released prompt enough for me. And the packages I need the most for my work like Numeric Python, ATLAS, subversion have much more recent versions and better integration and packaging in debian than in portage. It is a fact that many people in my line of activity (engineering research) use debian. If I spent all my day using KDE/Gentoo applications as you seem to do, I'd probably agree with you.
So here I am having to either move to an unstable branch of Debian, which I've seen break before, or add all kinds of third party repositories (Can you say UNTRUSTED SOFTWARE) to my sources.list. What happened to just wanting to get work done?
Didn't you ever downloaded an unofficial ebuild from foruns.gentoo.org? I actually think that portage should allow having seperate repositories such as sources.list, because the current gentoo developers can't cope (or are not interested in coping) with all the potential packages (I had/have bug reports on bugs.gentoo.org of many months). BTW, all unofficial repositories I use on debian happen to be provided by official debian developers.
Things do get broken on unstable, but no more than they do on gentoo, and with the bonus that everytime they try to get xfree86/gcc/binutils package right, you don't have to recompile it over and over again.
Again, I don't defend that Debian is the holy grail of linux distributions, not only because such holy grail doesn't exists (every distro is suited for a certain purpose), but because I know that tomorrow I'll change distro again. After all, one great thing with linux/unix is that usually it oesn't take more than 1-2days to swith a distro and have everything as before since the most important part - your home dir - doesn't even need to be touched.
Actually I just switched from Gentoo to Debian on my main PC (a laptop with PIII Celeron 700MHz with 198MB ram). I was using Gentoo for a year now and just installed Debian on other less frequently used machines but now switched completly to Debian.
Basically in Gentoo I was sick of:
having to wait 10 min to rsync portage plus other 10min to rebuild its huge cache
using a broken kernel (the last drop was when I read that during a whole month the kernel wasn't obeying to the process priorities! See for your self in its CVS)
having to wait 3 minutes just to get the list of the installed packages!!
most times a new version of a library was out the packages that depend on it got broken and have to be manually recompiled
many packages I need couldn't be found on portage (I did contribute ebuilds many times, but it usually takes too long to get accepted)
closed developer community (the development mailing list isn't open - no, gentoo-dev is for experienced users, and not even that nowadays, since almost everybody uses the web forum or IRC and I have no time for that)
over the time alot of things fail to work properly and you basically feel that you should emerge world to get a fresh restart but that would take ages and sounds like reinstalling Windows...
The most surprising thing was that with Debian:
the computer got snapier - lower memory footprint
I could find almost all packages I need, even the lastest: if not on debian unstable, on unofficial apt repositories
apt and the debian mirror system is a really well thought and evolved system - download/query everythin is fast and efficient
kernels and kernel modules (especially ALSA) are so beautifuly handled with make-kpkg
I still can easily compile the performance critical applications from source to get the best of the processor capabilities
I know my computer isn't the fastest out there, but while the timings may vary, the scalability problems with portage are still there, so it's just a matter of time until faster computers start experience the same delays.
Anyway, this is also reflection of my interests and my experience. When I started with Linux I wanted RedHat because it was the most familiar to everybody else. I switched to Gentoo because I wanted more and latest stuff which I couldn't easily find on RedHat (RPM hell!), and I wouldn't mind if things got broken as I could usually help sorting everything out. But now my interests are narrowing down, and I still want the latest stuff, but I just don't want that to get in the way of my daily work.
I know that it may be difficult to KDE enthusiasts to dissociate the KDE project from this endeavor called 'KDE League'. And spite of all second intentions that many of the slashdot posters argue Dennis E. Powel and LinuxandMain have, the fact is that they are doing a tremendous service to the community, especialy the supportive KDE fans which seem to have been the most damaged party in all this story.
Since the begining that many people admired and feared the community spirit surrounding Linux and other popular open-source projects as KDE. But we can't forget that many tried and will try to make use of this community spirit for their own gain. That's why transparency should be praised, instead of being considered as an attack to the good name. , because transparency is the _only_ way to have a respectable image.
In contrary to your opinion, what I'll remember of this will be danger that these fund-raising Leagues/Foundations can be if their business model and intents aren't properly open.
I've read carefully all the thread on http://kerneltrap.org/node.php?id=444 but I don't share your opinion.
Larry first presented BitKeeper designed to both aid the kernel development and to be comercially viable. So far these two goals haven't collided: people could use freely BitKeeper for kernel development and BitKeeper has been growing as a comercial product.
But this changes does affect everything: it prevents people that (even if just remotely) contribute somehow to different SCM products and which were using BitKeeper freely has they were incentivated to revert their habits. The problem is not that they can't find a way around, but that Larry is taking away a present which had been given away freely.
The worst is that this change does not seem to be made to protect BitKeeper business model but seems instead a act of bad faith against a particural set of people.
This radically change my opinion about the usage of BitKeeper and the trust on the people behind it, and from I've read Debian maintainers feel the same.
Both alocate and the FS table index the content of a disk but with different purposes _and_ structures.
FS has a tree alike structure (which maps directly to the directories and files) to easily get from a file name to its position on the disk.
[s]locate has progressive encoding (from their website http://www.geekreview.org/slocate/) which maps from name fragments to their complete name.
Another difference is that slocate database structure is difficult to update, and you wouldn't want to do that everytime you rename/add/delete a file...
Note that I'm not an expert, and this is just my (incomplete and perhaps even wrong) understanding .
Although the Pentium-or-better machine requirement is quite true, there is no need of a bootable CD-ROM.
The first time I installed Gentoo, I did it without ever leaving the comfort of my existing RedHat system. It took several days to build everything I needed in Gentoo (such as GNOME, mozilla, and all other common apps), but all this time I kept working as before. In the end it was just necessary to add a new entry to grub and boot Gentoo! My personal home dir remained the same so everything was configured as I had it before! (You have to love linux for this!) I also did this a 2nd time between two Gentoo versions, to recompile Gentoo for using gcc-3.1.
Gentoo really saves you alot of time when installing upgrading packages, but for old machines that is no longer true. I attempted to install Gentoo on a AMD K6 350 and it took a day to build X and in the end, because I was also trying to use gcc-3.1 it failed! Since I just wanted any Linux running on that machine for testing purposes I installed RH. But I would never leave Gentoo on my personal system! It was the best medicine for my update fever!
Although I share most of RMS concerns he fails to acknowledge the different relations that the community has with Free software.
While some are hackers with higher ideals as he is, many are hobbists that simple enjoy programming. Others are just users. For some is the work that feeds their families. For others is a enormous market for their products or services. Many simply enjoy the fact of being part of something much bigger than them.
Spite of all this variety, Free software offers something to all this people, and all this people offer back something to the community.
The system that you have and enjoy so much is result of many people's work. GNU, Linux, BSD, XFree86, Gnome or KDE are just bricks which have little value when considered alone (hence the absurd of GNU/Linux) - they are nothing alone and, above all, they are nothing without the community which made/uses them. Although so many people hate it (and rightfuly too) even Microsoft had an important role by making the Personal Computer an acessible reality, and by being a monopoly which made the Open Source alternative florish so much.
We must accept this variety, instead of forcing our point of views on everybody else.
I think that one of the effects of RMS polemics is that he always makes one think, reminding the past, and planing the future as we want, instead of blindfoldly taking what the present gives. That is his major achievement.
It's clear to me that you just see what is right in front of your nose.
Every single sucessful open-source project was beaten by closed source solutions during a large initial period of its life. Now if the users didn't support it during that stage - bug reporting, bug fixing, etc. - then they would never had reached to a point that they beat the closed source solution.
There is nothing about ideology here. You just don't have the ability of looking forward in time, but perhaps this isn't needed in your line of jobs!
You seem to forget the advantages of open-source. Why do you think Linux got popular in the beginning? Surely wasn't because it had more features than the other closed source OSs. It was because people could improve it and suit it to their own needs.
When the closed source drivers don't work for you or the vendor stops supporting them don't come crying to the DRI developers. I hope that this never happens, but realize that you're aiming to your own foot when you make these broad statements dispising their work.
Making an OpenGL driver is an herculean endeavor since involves dealing with complex things such as low-level programming and 3D rendering, and to complicate even further usually there is lack of information due to intelectual property issues. Spite of that, open-source has shown to be a viable (if not better) alternative than closed-source in so many fields so why not 3D graphics as well? Is just a matter of believing in it and work in that sense.
Slashdot Mirror
Call graphs are nice, but call graphs of large applications done via static code analysis are so huge and dense that become useless. Call graphs taken from run-time profile data, with all those irrelevant nodes pruned out, are IMO much more useful, as they naturally direct you to the most interesting parts, where the action is.
I think so too. But I've found that command-line debuggers like gdb invariably show a very limited view of the code, which presents an obstacle for code comprehension.
If you're going to debug with the aim of understanding the code, then using a debugger with a graphical interface is a must. On that regard. I personally find Eclipse CDT very useful, since not only it has a graphical interface to gdb, but also has code navigation abilities which allows you to quickly jump back and forth the function/type definitions and so on.
See an article on DeveloperWorksabout that at h for doing that on the server side.
I'm also using an home grown Python script based on that principle for my homepage. I'm actually in the process of rewriting it to use the more robust HTML and XHMTL parsing of the libxml2 python bindings, and to have even cooler stuff such as Xpath and XSLT under the control of your embedded python scripts.
You can also do the scripting on the client side. The Win32 version ActiveState Python has support for doing that - if you're on that sort of pseudo OS... ;-)
Somehow, if they simply reported to MS without asking anything back, and made the explot public after a month, it would be OK? Even if it would bring less benefits for both parties than to make an arrangement?
I've been trying several kernels as the development focuses them (gentoo-sources, lolo-sources, I don't know the details about the process priorities bug, even if is unrelated the fact is that doing things with nice -19 (such as emerging a packaged), almost took the machine to a halt - now I can compile a package without that hapening, so it must be a problem in gentoo somewhere. Anyway, this isn't bugs.gentoo.org to write a bug report...
I've also never broken my userland to the extent that I require a reinstall; did you read the part about Gentoo being for power users and being a learning platform? Yeah, if you're coming from RedHat, obviously you're going to screw it up. That's no problem. But be patient until you learn how Linux works, that's what you're there for isn't it?Ah! Ah! Ah!.... Oooohh... People are so funny (not to mention ridiculous) when they put labels on someone just because he use(s/d) distro A or B.
I always hear people saying "I use Debian because I just want to get work done" but something about it just doesn't ring true. Apt is a fantastic package manager I grant you, but whenever I install Debian I do without debconf. I know what I'm doing, thanks.Don't see the point of that...
So from my perspective the main thing I'm concerned with is packages. Are they up to date? Are userland packages like KDE or X released promptly when there are new versions? The answer is an unfortunate no. The stable branch of Debian refuses to merge new versions in a dynamic way, instead insisting that stable is stable and you must use unstable to get the latest.They are released prompt enough for me. And the packages I need the most for my work like Numeric Python, ATLAS, subversion have much more recent versions and better integration and packaging in debian than in portage. It is a fact that many people in my line of activity (engineering research) use debian. If I spent all my day using KDE/Gentoo applications as you seem to do, I'd probably agree with you.
So here I am having to either move to an unstable branch of Debian, which I've seen break before, or add all kinds of third party repositories (Can you say UNTRUSTED SOFTWARE) to my sources.list. What happened to just wanting to get work done?Didn't you ever downloaded an unofficial ebuild from foruns.gentoo.org? I actually think that portage should allow having seperate repositories such as sources.list, because the current gentoo developers can't cope (or are not interested in coping) with all the potential packages (I had/have bug reports on bugs.gentoo.org of many months). BTW, all unofficial repositories I use on debian happen to be provided by official debian developers.
Things do get broken on unstable, but no more than they do on gentoo, and with the bonus that everytime they try to get xfree86/gcc/binutils package right, you don't have to recompile it over and over again.
Again, I don't defend that Debian is the holy grail of linux distributions, not only because such holy grail doesn't exists (every distro is suited for a certain purpose), but because I know that tomorrow I'll change distro again. After all, one great thing with linux/unix is that usually it oesn't take more than 1-2days to swith a distro and have everything as before since the most important part - your home dir - doesn't even need to be touched.
Actually I just switched from Gentoo to Debian on my main PC (a laptop with PIII Celeron 700MHz with 198MB ram). I was using Gentoo for a year now and just installed Debian on other less frequently used machines but now switched completly to Debian.
Basically in Gentoo I was sick of:
The most surprising thing was that with Debian:
I know my computer isn't the fastest out there, but while the timings may vary, the scalability problems with portage are still there, so it's just a matter of time until faster computers start experience the same delays.
Anyway, this is also reflection of my interests and my experience. When I started with Linux I wanted RedHat because it was the most familiar to everybody else. I switched to Gentoo because I wanted more and latest stuff which I couldn't easily find on RedHat (RPM hell!), and I wouldn't mind if things got broken as I could usually help sorting everything out. But now my interests are narrowing down, and I still want the latest stuff, but I just don't want that to get in the way of my daily work.
I know that it may be difficult to KDE enthusiasts to dissociate the KDE project from this endeavor called 'KDE League'. And spite of all second intentions that many of the slashdot posters argue Dennis E. Powel and LinuxandMain have, the fact is that they are doing a tremendous service to the community, especialy the supportive KDE fans which seem to have been the most damaged party in all this story.
Since the begining that many people admired and feared the community spirit surrounding Linux and other popular open-source projects as KDE. But we can't forget that many tried and will try to make use of this community spirit for their own gain. That's why transparency should be praised, instead of being considered as an attack to the good name. , because transparency is the _only_ way to have a respectable image.
In contrary to your opinion, what I'll remember of this will be danger that these fund-raising Leagues/Foundations can be if their business model and intents aren't properly open.
2D and 3D support is already on XFree86 and DRI CVS respectively. So the next XFree86 should most probably include it too.
I've read carefully all the thread on http://kerneltrap.org/node.php?id=444 but I don't share your opinion.
Larry first presented BitKeeper designed to both aid the kernel development and to be comercially viable. So far these two goals haven't collided: people could use freely BitKeeper for kernel development and BitKeeper has been growing as a comercial product.
But this changes does affect everything: it prevents people that (even if just remotely) contribute somehow to different SCM products and which were using BitKeeper freely has they were incentivated to revert their habits. The problem is not that they can't find a way around, but that Larry is taking away a present which had been given away freely.
The worst is that this change does not seem to be made to protect BitKeeper business model but seems instead a act of bad faith against a particural set of people.
This radically change my opinion about the usage of BitKeeper and the trust on the people behind it, and from I've read Debian maintainers feel the same.
Both alocate and the FS table index the content of a disk but with different purposes _and_ structures.
FS has a tree alike structure (which maps directly to the directories and files) to easily get from a file name to its position on the disk.
[s]locate has progressive encoding (from their website http://www.geekreview.org/slocate/) which maps from name fragments to their complete name.
Another difference is that slocate database structure is difficult to update, and you wouldn't want to do that everytime you rename/add/delete a file...
Note that I'm not an expert, and this is just my (incomplete and perhaps even wrong) understanding .
Although the Pentium-or-better machine requirement is quite true, there is no need of a bootable CD-ROM.
The first time I installed Gentoo, I did it without ever leaving the comfort of my existing RedHat system. It took several days to build everything I needed in Gentoo (such as GNOME, mozilla, and all other common apps), but all this time I kept working as before. In the end it was just necessary to add a new entry to grub and boot Gentoo! My personal home dir remained the same so everything was configured as I had it before! (You have to love linux for this!) I also did this a 2nd time between two Gentoo versions, to recompile Gentoo for using gcc-3.1.
Gentoo really saves you alot of time when installing upgrading packages, but for old machines that is no longer true. I attempted to install Gentoo on a AMD K6 350 and it took a day to build X and in the end, because I was also trying to use gcc-3.1 it failed! Since I just wanted any Linux running on that machine for testing purposes I installed RH. But I would never leave Gentoo on my personal system! It was the best medicine for my update fever!
Although I share most of RMS concerns he fails to acknowledge the different relations that the community has with Free software. While some are hackers with higher ideals as he is, many are hobbists that simple enjoy programming. Others are just users. For some is the work that feeds their families. For others is a enormous market for their products or services. Many simply enjoy the fact of being part of something much bigger than them. Spite of all this variety, Free software offers something to all this people, and all this people offer back something to the community. The system that you have and enjoy so much is result of many people's work. GNU, Linux, BSD, XFree86, Gnome or KDE are just bricks which have little value when considered alone (hence the absurd of GNU/Linux) - they are nothing alone and, above all, they are nothing without the community which made/uses them. Although so many people hate it (and rightfuly too) even Microsoft had an important role by making the Personal Computer an acessible reality, and by being a monopoly which made the Open Source alternative florish so much. We must accept this variety, instead of forcing our point of views on everybody else. I think that one of the effects of RMS polemics is that he always makes one think, reminding the past, and planing the future as we want, instead of blindfoldly taking what the present gives. That is his major achievement.
It's clear to me that you just see what is right in front of your nose. Every single sucessful open-source project was beaten by closed source solutions during a large initial period of its life. Now if the users didn't support it during that stage - bug reporting, bug fixing, etc. - then they would never had reached to a point that they beat the closed source solution. There is nothing about ideology here. You just don't have the ability of looking forward in time, but perhaps this isn't needed in your line of jobs!
You seem to forget the advantages of open-source. Why do you think Linux got popular in the beginning? Surely wasn't because it had more features than the other closed source OSs. It was because people could improve it and suit it to their own needs.
When the closed source drivers don't work for you or the vendor stops supporting them don't come crying to the DRI developers. I hope that this never happens, but realize that you're aiming to your own foot when you make these broad statements dispising their work.
Making an OpenGL driver is an herculean endeavor since involves dealing with complex things such as low-level programming and 3D rendering, and to complicate even further usually there is lack of information due to intelectual property issues. Spite of that, open-source has shown to be a viable (if not better) alternative than closed-source in so many fields so why not 3D graphics as well? Is just a matter of believing in it and work in that sense.