As they have stated many times, the reason for closed development lists is the fact that you currently have to sign NDA's with the major graphics chip companies in order to work on XFree. The NDA's are designed to keep the programming specs secret, but they allow for source code release of the drivers. This keeps the suits at ATi, Matrox, etc.. happy, and allows XFree to run on the various cards. Up till now, this has been an OK relationship.
With 4.0, however, the driver layer has been abstracted into loadable modules, so we should see the core server functionality opening up a lot (I hope!). I for one would like to see division along the lines of KDE/Gnome, with a group of people working on the NDA-requiring drivers, and the rest of the tree open on an anonymous CVS server. This would allow more "casual" eyes on the code, while keeping the "developers" who want to devote more time as the maintainers of the various modules and subsystems.
This would also let you get daily snaps of _most_ of the stuff, with driver snaps released frequently, too, since the driver people wouldn't have to constantly wait on the core people, and vice-versa.
The new Finder is really the NeXT WorkPlace.app, NeXT's verion of a file manager/app launcher. It just has the MacOS X "look-n-feel" added to it. I am personally pleasantly surprised to see a few of the NeXT UI features integrated in the Mac UI, and I wouldn't mind using one of these machines, except for two (related) things. I wish they would a least have the option to put menubars into the individual windows instead of a single one across the top, and allow us to use mouse focus (sloppy focus...even better!). Apple has always made nice machines, but this is the first time I actually thought to myself "Hey, I might like to own one of those!". I was too young/poor for the NeXT machines, but a fast G4 with OS X sitting next to my linux boxes would make me a happy man!
4.0 still follows the X protocols, as defined by the Open Group (or X Consortium, or X-men, or whatever they call themselves this week), it is simply a more modern rewrite of the current implementation and it adds some of the features present on some commercial unices. In order to stil be called an implementation of "The X Window System", XFree can not change the actual design very much, and therefore can not do much of what QNX is trying to do.
4.0 is almost completely modular now, instead of having core functionality compiled into all of the drivers, there is now a core binary that loads almost everything in from modules (similar to the kernel). That means drivers can be compiled separately (and even distributed binary-only, for the suits at the video card companies). Also, other Unices have had multihead and Xinerama for a while, and of course SGI has always smoked on 3D stuff.
The QNX stuff looks pretty impressive, given their attention to size constraints and "real-time"-ness, but I haven't worked with it, so can not speak to the quality.
Even though this is a snapshot, you would have thought that someone would have built this on a few of the common platforms before uploading the tarballs. On a Redhat 6.1 intel system, I had to hand fix a few errors in the build process. The most annoying (besides the doc dir) was the actual XFree86 binary not being built at all due to a bad link in the mesa includes, but "make World" and "make install" chugged right on past it instead of dying with an error.
You shouldn't need the location on the PCI bus unless you want multihead.
If you have two Matrox cards (say, one on the AGP bus and one on the PCI bus), you would have two Device sections, one with: Bus_ID "PCI:0:10:0" and one with: Bus_ID "PCI:1:0:0"
Of course, they have to have different names, but you get the picture...
Do you even know what the words "modular programming" mean?!?!
The snapshots are already broken into the "main" portion of the X sever and the driver modules, i.e. there is a binary that starts up, loads the config file, loads in the appropriate module for the particular graphics card family, handles the input devices (which are also modules that can be dropped in place without recompiling the X server), and then does all the back-end work like networking, communication, image setup, etc. It then just passes the actual images to the video driver to display. The drivers are all separate from the server itself, and can be programmed independantly, compiled separately, and then dropped into place. Sound familiar? That's just how the kernel works. All the hardware drivers are modular and can be loaded/unloaded at will. There are people all over the world who are not full-time "kernel developers" who simple have an itch, pull up the source to the driver for their particular peice of hardware, maybe come up with a patch or two, and email it to the appropriate maintainer. If it works, and is good enough, it gets into the next release; if not, it gets canned or modified. This is exactly the model that Xfree should go toward (and I think/hope/expect they will shortly after the 4.0 release)
The various modules for video, keyboard, mouse, touchpad, etc. can now all be developed separately, so that one doesn't (shouldn't is more accurate ATM) have to sign an NDA to be in on the development of the core functionality and design. Those who have the time, expertise, and interest in working on the FooX128Ultra(TM) chipset support can sign that NDA and start coding, while us regular mortals can just browse the up-to-the-minute anonymous CVS tree and see where we might want to chip in, or monitor the mailing list (read-only maybe, allow the "full developers" posting priviledges or something) to see if maybe there are areas we might be interested in, or a call for some "grunt work" like documentation or something else that even non-programmers could do.
Anyway, I just want to put aside the debate for a moment to say that, without a doubt, multihead and Xinerama ROCK ASS (with petrified hot grits, no less!) that was just to be humorous, don't confuse me with the lame-ass firstpost/grits/petrified-and-naked people)
Most people buy a computer that already has everything installed on it, or they buy an upgrade and pay $50 for Joe CompuGuy at CompUSA to install it. Most small businesses do the same thing, and when they are large enough hire a part-time or full-time tech to do it for them.
And for an exception to the rule... One of my co-workers has some friends at a small company here in town. They bought a copy of Corel Linux a month ago, but never got around to getting him to install it and set up samba (they wanted a file/print server, but didn't want to shell out the $$$ for NT). He called them the other day, and they had the secretary (secretary, mind you) installing it "just to see how easy it was"! She had made it all the way through the installation and was using Corel's "Windows Filesharing" configuration tool. Now I haven't used Corel, but the secretary didn't appear to have had any trouble installing everything (using the defaults, I assume). He is going to go over there after a few days to see how secure and "correct" the installation and setup are.
Nuts to the "Installing Windows is easy...Installing Linux is hard!" crowd!
That's why the product is called J++, the name _can't_ have the word "java" in it at all because it doesn't follow the Sun standard. The IDE produces code that looks suspiciously like java (and can run in MS's JVM). If you _carefully_ read their ad-speak, the only place it mentions "real java" (as opposed to their bastard version) is where it says it can produce "Pure Java" code that will run cross-platform. A _very_ fine point of distinction, and some lawyers and ad-men got paid a lot of money for figuring out how to sell the product, implying that it is a true Java IDE, without getting sued (they got sued anyway, but that was for the JVM, not the IDE).
Why do parents across the US dress their daughters up like little prostitutes and parade them in little miss beauty pageants, like JonBenet Ramsey?
It is the same thing.
Re:It's diluting the dedication of the artists...
on
Review:Toy Story 2
·
· Score: 1
No, you are reminiscing about the times when Walt himself pushed for the _stories_ and animation to be the best that they could be. These days Disney is just another group of money men keeping their stockholders happy pushing inferior movies to kids (along with all of the associated toy/product licensing that follows). Pixar has actually been able to work around that, though, and make some great movies under the Disney thumb.
If any of the posts from people who didn't like the movie had any content as to why they didn't like it, then they wouldn't get moderated down. All of the ones I've seen have just been "Toy Story/Pixar/Disney sucks!" type of posts (and the stupid NAKED AND PETRIFIED type post). If you don't like it, then post a reasonable explanation as to why, don't use profanity, and try to spell all of the words correctly.
I happen to have really liked the movie for its own sake. The story was good, the animation (in places, like Al's face when he was sleeping) are astounding. My 8 year old son and I both laughed through the whole thing (and we usually laughed in the same places, showing he got most of the jokes and references aimed at the adults), and the whole audience cheered at the end. The "buddy picture" aspect of the first carried over nicely, and I was glad to see that the rest of the main characters got to go along for the adventure this time.
As for Rob moderating this site with an iron fist (as you seem to think), Rob doesn't even moderate the stories, the readers do! Since Toy Story is not some controversial subject, a serious discussion about the film is just not possible. You either liked it or you didn't. If you liked it, then fine. If you liked all but one or two parts, then fine. If you thought is was the worst movie ever, then say so in a manner than doesn't draw flames and you will not get marked "Flamebait"!
It is a single bloated binary that crashes on a regular basis, and (more importantly) it doesn't follow a large chunk of the CSS, DOM, and DHTML standards. They keep saying "5.0 will fix all that!!" but it isn't even in beta.
Mozilla, while starting to look nice, is still not anywhere near the quality I would expect to use on a daily basis. The latest Milestone release is still very flawed, and what's worse, seems to be heading even more to being Windows-centric than Navigator 4.xx!! I mean come on, it starts a "Profile Manager" to set up a "Profile" for you when you first run it. Um, helllooo!!!! This is Unix, thank you, I already have a Profile, it is called my user account. Just create a ~/.mozilla directory for your settings and cache, and pop up a nice little window telling me so. Also, I tried both the Windows and Linux versions of M11, and the Windows version seemed more "complete", in that portions of the program that would crash the Linux version would work just fine on the Windows version.
Even KFM is a better browser than Netscape, and the Konq rewrite, with the added fully compliant Java/Javascript and CSS/DOM/DHTML standards will blow Mozilla out of the water.
I'm sorry to say this, being a long-time Netscape user (on both Windows and Linux) but Konq has working now what Mozilla will have six months from now, and given Netscape's track record of slowly but surely eroding the functionality and stability of the Unix versions of Navigator/Communicator, I for one will be happy to see Konq come out so that we can really have a decent browser.
From the info I've seen, Konqueror actually seems more complete in the support for the DHTML standards than Mozilla right now. I don't like to admit it, but IE is light years ahead of Communicator in most respects (except for the swiss-cheese that MS calls security). Konq actually seems like it will catch up to the DOM and CSS standards and maybe surpass both Mozilla and IE.
KDE as a "desktop environment" I could care less about. I really can't wait for a nice, standards compliant web browser on Unix (instead of Netscape's buggy, incomplete, just-plain-sucks offerings). I wonder if (since it based on QT) Konq can run on Windows? I have a project that I am working on that would be much easier if I could take advantage of some of the CSS and DOM features that are lacking in either IE or NN (or both). It would be nice to be able to just install Konq on the user's desktop, make a nice little icon for them to click on, and target the full specs instead of just the subset that NN or IE follows (or worse, the "extensions" to the standard that MS so gratefully put in to tempt developers). This is the first of several "browser-based" apps we are going to do in the near future, and I was shocked at the curent state of affairs in the standards compliance of the various browers.
The quad Xeon _will_ kick a single dual celeron's ass. But will the Xeon kick the ass of the ten dual celeron's in a server farm that costs the same as the quad Xeon? I think not.
HTTP, SMTP, and POP protocols all lend themselves to parallelization, since they are inherently stateless connections. A properly designed farm scales nicely and handles fault tolerance better than a single machine ever can (even with RAID arrays and hot-swap power and hard drives).
Save the quad Xeon for an application, database, or a TP server and split the web stuff off into a farm.
In order for my mom and my son to use the programs that they want to use, have already puchased, and learned to use (Quicken, a lot of educational games, etc...), then the computers that they must use must have intel-based hardware and a Win32-based OS.
These programs could run just fine on an AMD/Cyrix machine, but where else am I going to find another Win32-based OS for them? I would like the same freedom to choose which Win32 OS I use that I have in the Intel-based Unix realm. Hrm, I have Unix programs, and I can choose between the free/nonfree BSD's, Solaris x86, SCO (ugh!), Linux, etc... A simple recompile (if in source form) or using iBCS if in binary-only form and I can run the program with a minimum of fuss.
How about putting the Win32 API into the hands of an independant not-for-profit body and allowing "Windows Clones" to be created that must be certified. Gee, this sounds rather like POSIX or X or something.....
Re:3D? Who cares. I just want a 10 head display!
on
3D Window Manager
·
· Score: 1
Yes, it is called x2x, and can be set up so that as the mouse hits the edge of a screen, it "jumps" to the next monitor (on a different machine or on the same machine). If you go to the VNC website and look under contributions, there is a program called x2vnc that does that same thing with a windows machine, one mouse/keyboard on the unix machine controlling both the X display and a windows display via VNC. So you can use a combination of x2x and x2vnc to control any group of X-based or VNC-based displays.
This has got to be one of the most ridiculous things that I have ever heard about. I guess the fact that their technique almost perfectly matches with the 32-bit unix "Y38" problem is mere coincidence. Many, many people (including myself) have used 1938/2038 as the "window" in which old data gets updated to new dates for this very reason, and as "y2k" compliant date mechanisms and 64-bit dates become the norm, quite a few "pre2K" fixes are probably based on this technique.
Software patents are evil, and must be stopped before it is too late!!
(conspiracy theory) The _real_ reason that the.cf file is so complicated is so that Sendmail, Inc. can sell Sendmail Pro to everyone, with their nice GUI configuration software. No more editing.cf when you're in a hurry, no more running m4 to update the.cf, just click that there Sendmail logo on the desktop and go clicky-clacky till it works! (/conspiracy theory)
(old man voice) Back in my day, we didn't have none o' tham gooeys these kids are all crazy about...we used a hex editor on a VT100 to modify the sendmail binary directly...AND WE LIKED IT!"
The problem with the sendmail.cf file is that the.cf file was designed from the start to be parsed quickly by the program, not to be human readable. The only concession to human-readibility that the.cf makes is the comments. For a long-lived server, startup time is neglible compared to the runing time of the process, so for the long-lived instance of sendmail (the one that receives mail and processes the queue at regular intervals) why not make it a human readable file anyway? Plus the fact that servers are much faster now means that the overhead of the parsing time is very small.
Why doesn't sendmail use Berkeley DB for the main config info anyway? It already uses it for the aliases and other config files in/etc/mail. If speed in loading config info is so important, slurping the data in directly from a DB file would be faster than parsing text.
Slashdot Mirror
As they have stated many times, the reason for closed development lists is the fact that you currently have to sign NDA's with the major graphics chip companies in order to work on XFree. The NDA's are designed to keep the programming specs secret, but they allow for source code release of the drivers. This keeps the suits at ATi, Matrox, etc.. happy, and allows XFree to run on the various cards. Up till now, this has been an OK relationship.
With 4.0, however, the driver layer has been abstracted into loadable modules, so we should see the core server functionality opening up a lot (I hope!). I for one would like to see division along the lines of KDE/Gnome, with a group of people working on the NDA-requiring drivers, and the rest of the tree open on an anonymous CVS server. This would allow more "casual" eyes on the code, while keeping the "developers" who want to devote more time as the maintainers of the various modules and subsystems.
This would also let you get daily snaps of _most_ of the stuff, with driver snaps released frequently, too, since the driver people wouldn't have to constantly wait on the core people, and vice-versa.
The new Finder is really the NeXT WorkPlace.app, NeXT's verion of a file manager/app launcher. It just has the MacOS X "look-n-feel" added to it. I am personally pleasantly surprised to see a few of the NeXT UI features integrated in the Mac UI, and I wouldn't mind using one of these machines, except for two (related) things. I wish they would a least have the option to put menubars into the individual windows instead of a single one across the top, and allow us to use mouse focus (sloppy focus...even better!). Apple has always made nice machines, but this is the first time I actually thought to myself "Hey, I might like to own one of those!". I was too young/poor for the NeXT machines, but a fast G4 with OS X sitting next to my linux boxes would make me a happy man!
IANAGE, but here goes....
(I am not a GUI expert)
4.0 still follows the X protocols, as defined by the Open Group (or X Consortium, or X-men, or whatever they call themselves this week), it is simply a more modern rewrite of the current implementation and it adds some of the features present on some commercial unices. In order to stil be called an implementation of "The X Window System", XFree can not change the actual design very much, and therefore can not do much of what QNX is trying to do.
4.0 is almost completely modular now, instead of having core functionality compiled into all of the drivers, there is now a core binary that loads almost everything in from modules (similar to the kernel). That means drivers can be compiled separately (and even distributed binary-only, for the suits at the video card companies). Also, other Unices have had multihead and Xinerama for a while, and of course SGI has always smoked on 3D stuff.
The QNX stuff looks pretty impressive, given their attention to size constraints and "real-time"-ness, but I haven't worked with it, so can not speak to the quality.
Even though this is a snapshot, you would have thought that someone would have built this on a few of the common platforms before uploading the tarballs. On a Redhat 6.1 intel system, I had to hand fix a few errors in the build process. The most annoying (besides the doc dir) was the actual XFree86 binary not being built at all due to a bad link in the mesa includes, but "make World" and "make install" chugged right on past it instead of dying with an error.
wierd....
You shouldn't need the location on the PCI bus unless you want multihead.
If you have two Matrox cards (say, one on the AGP bus and one on the PCI bus), you would have two Device sections, one with:
Bus_ID "PCI:0:10:0"
and one with:
Bus_ID "PCI:1:0:0"
Of course, they have to have different names, but you get the picture...
Incidentally, multihead ROCKS ASS!
particulary when this posted here many moons ago (can't remember the article, too lazy to search for it...)
The snapshots are already broken into the "main" portion of the X sever and the driver modules, i.e. there is a binary that starts up, loads the config file, loads in the appropriate module for the particular graphics card family, handles the input devices (which are also modules that can be dropped in place without recompiling the X server), and then does all the back-end work like networking, communication, image setup, etc. It then just passes the actual images to the video driver to display. The drivers are all separate from the server itself, and can be programmed independantly, compiled separately, and then dropped into place. Sound familiar? That's just how the kernel works. All the hardware drivers are modular and can be loaded/unloaded at will. There are people all over the world who are not full-time "kernel developers" who simple have an itch, pull up the source to the driver for their particular peice of hardware, maybe come up with a patch or two, and email it to the appropriate maintainer. If it works, and is good enough, it gets into the next release; if not, it gets canned or modified. This is exactly the model that Xfree should go toward (and I think/hope/expect they will shortly after the 4.0 release)
The various modules for video, keyboard, mouse, touchpad, etc. can now all be developed separately, so that one doesn't (shouldn't is more accurate ATM) have to sign an NDA to be in on the development of the core functionality and design. Those who have the time, expertise, and interest in working on the FooX128Ultra(TM) chipset support can sign that NDA and start coding, while us regular mortals can just browse the up-to-the-minute anonymous CVS tree and see where we might want to chip in, or monitor the mailing list (read-only maybe, allow the "full developers" posting priviledges or something) to see if maybe there are areas we might be interested in, or a call for some "grunt work" like documentation or something else that even non-programmers could do.
Anyway, I just want to put aside the debate for a moment to say that, without a doubt, multihead and Xinerama ROCK ASS (with petrified hot grits, no less!) that was just to be humorous, don't confuse me with the lame-ass firstpost/grits/petrified-and-naked people)
Can I get an amen, brothers and sisters??
How many "mainstream" people install their OS?
Most people buy a computer that already has everything installed on it, or they buy an upgrade and pay $50 for Joe CompuGuy at CompUSA to install it. Most small businesses do the same thing, and when they are large enough hire a part-time or full-time tech to do it for them.
And for an exception to the rule...
One of my co-workers has some friends at a small company here in town. They bought a copy of Corel Linux a month ago, but never got around to getting him to install it and set up samba (they wanted a file/print server, but didn't want to shell out the $$$ for NT). He called them the other day, and they had the secretary (secretary, mind you) installing it "just to see how easy it was"! She had made it all the way through the installation and was using Corel's "Windows Filesharing" configuration tool. Now I haven't used Corel, but the secretary didn't appear to have had any trouble installing everything (using the defaults, I assume). He is going to go over there after a few days to see how secure and "correct" the installation and setup are.
Nuts to the "Installing Windows is easy...Installing Linux is hard!" crowd!
That was my first thought as well!
That's why the product is called J++, the name _can't_ have the word "java" in it at all because it doesn't follow the Sun standard. The IDE produces code that looks suspiciously like java (and can run in MS's JVM). If you _carefully_ read their ad-speak, the only place it mentions "real java" (as opposed to their bastard version) is where it says it can produce "Pure Java" code that will run cross-platform. A _very_ fine point of distinction, and some lawyers and ad-men got paid a lot of money for figuring out how to sell the product, implying that it is a true Java IDE, without getting sued (they got sued anyway, but that was for the JVM, not the IDE).
testes, you ass...
Why do parents across the US dress their daughters up like little prostitutes and parade them in little miss beauty pageants, like JonBenet Ramsey?
It is the same thing.
No, you are reminiscing about the times when Walt himself pushed for the _stories_ and animation to be the best that they could be. These days Disney is just another group of money men keeping their stockholders happy pushing inferior movies to kids (along with all of the associated toy/product licensing that follows). Pixar has actually been able to work around that, though, and make some great movies under the Disney thumb.
If any of the posts from people who didn't like the movie had any content as to why they didn't like it, then they wouldn't get moderated down. All of the ones I've seen have just been "Toy Story/Pixar/Disney sucks!" type of posts (and the stupid NAKED AND PETRIFIED type post). If you don't like it, then post a reasonable explanation as to why, don't use profanity, and try to spell all of the words correctly.
I happen to have really liked the movie for its own sake. The story was good, the animation (in places, like Al's face when he was sleeping) are astounding. My 8 year old son and I both laughed through the whole thing (and we usually laughed in the same places, showing he got most of the jokes and references aimed at the adults), and the whole audience cheered at the end. The "buddy picture" aspect of the first carried over nicely, and I was glad to see that the rest of the main characters got to go along for the adventure this time.
As for Rob moderating this site with an iron fist (as you seem to think), Rob doesn't even moderate the stories, the readers do! Since Toy Story is not some controversial subject, a serious discussion about the film is just not possible. You either liked it or you didn't. If you liked it, then fine. If you liked all but one or two parts, then fine. If you thought is was the worst movie ever, then say so in a manner than doesn't draw flames and you will not get marked "Flamebait"!
Because Netscape sucks ass.
It is a single bloated binary that crashes on a regular basis, and (more importantly) it doesn't follow a large chunk of the CSS, DOM, and DHTML standards. They keep saying "5.0 will fix all that!!" but it isn't even in beta.
Mozilla, while starting to look nice, is still not anywhere near the quality I would expect to use on a daily basis. The latest Milestone release is still very flawed, and what's worse, seems to be heading even more to being Windows-centric than Navigator 4.xx!! I mean come on, it starts a "Profile Manager" to set up a "Profile" for you when you first run it. Um, helllooo!!!! This is Unix, thank you, I already have a Profile, it is called my user account. Just create a ~/.mozilla directory for your settings and cache, and pop up a nice little window telling me so. Also, I tried both the Windows and Linux versions of M11, and the Windows version seemed more "complete", in that portions of the program that would crash the Linux version would work just fine on the Windows version.
Even KFM is a better browser than Netscape, and the Konq rewrite, with the added fully compliant Java/Javascript and CSS/DOM/DHTML standards will blow Mozilla out of the water.
I'm sorry to say this, being a long-time Netscape user (on both Windows and Linux) but Konq has working now what Mozilla will have six months from now, and given Netscape's track record of slowly but surely eroding the functionality and stability of the Unix versions of Navigator/Communicator, I for one will be happy to see Konq come out so that we can really have a decent browser.
From the info I've seen, Konqueror actually seems more complete in the support for the DHTML standards than Mozilla right now. I don't like to admit it, but IE is light years ahead of Communicator in most respects (except for the swiss-cheese that MS calls security). Konq actually seems like it will catch up to the DOM and CSS standards and maybe surpass both Mozilla and IE.
KDE as a "desktop environment" I could care less about. I really can't wait for a nice, standards compliant web browser on Unix (instead of Netscape's buggy, incomplete, just-plain-sucks offerings). I wonder if (since it based on QT) Konq can run on Windows? I have a project that I am working on that would be much easier if I could take advantage of some of the CSS and DOM features that are lacking in either IE or NN (or both). It would be nice to be able to just install Konq on the user's desktop, make a nice little icon for them to click on, and target the full specs instead of just the subset that NN or IE follows (or worse, the "extensions" to the standard that MS so gratefully put in to tempt developers). This is the first of several "browser-based" apps we are going to do in the near future, and I was shocked at the curent state of affairs in the standards compliance of the various browers.
The quad Xeon _will_ kick a single dual celeron's ass. But will the Xeon kick the ass of the ten dual celeron's in a server farm that costs the same as the quad Xeon? I think not.
HTTP, SMTP, and POP protocols all lend themselves to parallelization, since they are inherently stateless connections. A properly designed farm scales nicely and handles fault tolerance better than a single machine ever can (even with RAID arrays and hot-swap power and hard drives).
Save the quad Xeon for an application, database, or a TP server and split the web stuff off into a farm.
In order for my mom and my son to use the programs that they want to use, have already puchased, and learned to use (Quicken, a lot of educational games, etc...), then the computers that they must use must have intel-based hardware and a Win32-based OS.
These programs could run just fine on an AMD/Cyrix machine, but where else am I going to find another Win32-based OS for them? I would like the same freedom to choose which Win32 OS I use that I have in the Intel-based Unix realm. Hrm, I have Unix programs, and I can choose between the free/nonfree BSD's, Solaris x86, SCO (ugh!), Linux, etc... A simple recompile (if in source form) or using iBCS if in binary-only form and I can run the program with a minimum of fuss.
How about putting the Win32 API into the hands of an independant not-for-profit body and allowing "Windows Clones" to be created that must be certified. Gee, this sounds rather like POSIX or X or something.....
x2x is at here and x2vnc is located here
This has got to be one of the most ridiculous things that I have ever heard about. I guess the fact that their technique almost perfectly matches with the 32-bit unix "Y38" problem is mere coincidence. Many, many people (including myself) have used 1938/2038 as the "window" in which old data gets updated to new dates for this very reason, and as "y2k" compliant date mechanisms and 64-bit dates become the norm, quite a few "pre2K" fixes are probably based on this technique.
Software patents are evil, and must be stopped before it is too late!!
Just what I need, medical advice from Al "Like a Surgeon" Yankovic!
Those are very good links, and proof that the guy from the original story isn't really a crackpot.
I can answer that question!
.cf file is so complicated is so that Sendmail, Inc. can sell Sendmail Pro to everyone, with their nice GUI configuration software. No more editing .cf when you're in a hurry, no more running m4 to update the .cf, just click that there Sendmail logo on the desktop and go clicky-clacky till it works!
(conspiracy theory)
The _real_ reason that the
(/conspiracy theory)
(old man voice) Back in my day, we didn't have none o' tham gooeys these kids are all crazy about...we used a hex editor on a VT100 to modify the sendmail binary directly...AND WE LIKED IT!"
The problem with the sendmail.cf file is that the .cf file was designed from the start to be parsed quickly by the program, not to be human readable. The only concession to human-readibility that the .cf makes is the comments. For a long-lived server, startup time is neglible compared to the runing time of the process, so for the long-lived instance of sendmail (the one that receives mail and processes the queue at regular intervals) why not make it a human readable file anyway? Plus the fact that servers are much faster now means that the overhead of the parsing time is very small.
/etc/mail. If speed in loading config info is so important, slurping the data in directly from a DB file would be faster than parsing text.
Why doesn't sendmail use Berkeley DB for the main config info anyway? It already uses it for the aliases and other config files in