Prism 2.5/3 based cards have supported scanning for years on Linux BUT (there had to be a but becaues your experience is different right?) it depended on the driver (HostAP and linux-wlan had could by default, orinoco needed drivers). Since the 2.4.13 kernel the orinoco drivers have supported scanning in the mainstream Linux kernel without patches.
Now by comparison a Linux distro (e.g. Ubuntu) can have support for the following chipsets (list taken from Linux WLAN Howto cross referenced against Ubuntu) on 19th November 2005:
A pretty similar list wouldn't you say (OpenBSD has Realtek which Ubuntu doesn't, Ubuntu has Ralink 2400, Prism GT, ACX100/110 and Wavelan which OpenBSD doesn't)? In fact, Linux has Realtek drivers too but as they aren't shipped in Ubuntu I left them off the list. Linux also has ndiswrapper and Linuxant Driverloader allowing the use of Win32 drivers but I'm discounting non native drivers.
Chipsets that aren't supported by either OS that are significant: Broadcom - There's a heck of a lot of Broadcom stuff out there and no sign of open drivers ever. That's their perogative but this stuff is all over the place... Marvel - a new 802.11g player as far as I can tell. No open source drivers that I know of so far.
For the meantime, chipsets like Broadcom mean that open source OSes will always have a more troublesome than Windows with random wireless drivers. Choose carefully and don't reward vendors with non free drivers where possible.
I'm amazed you haven't come across the linux wlan chipset list. Although it hasn't been updated for nearly two years it's about the most exhaustive list of what is in each card that I've seen. If 802.11b is fine then there are still Prism 2.5/3 based cards out there that work very well. I also recently tested a Atheros based PCMCIA card (I'm purposely not naming manufacturers) which also worked well but required a small binary lump.
As mentioned elsewhere, support for wifi isn't spotty - it's support for certain chipsets that is (alas this is also extends to various USB wifi devices too). If you buy (for example) a Broadcom based card I'm afraid you're in for a rough ride because Broadcom don't want to release open source drivers. There's no point getting upset - Broadcom are within their rights to do so and Linux isn't binary only friendly. It's the way things are.
The best advice I can give is get a peek inside the box so you know which chipset you are buying. Manufacturers are lazy and try to avoid changing model numbers significantly even if they swap chipsets because it means all the other materials can stay exactly the same. If you are going on someone else's information be extremly weary of ANY deviation to the model name/number/revison. Things like a +, extra letters or revision increase of any amount can mean chipset changes.
I'm not sure if it's simply down to starting less stuff but Mandriva 2006.0 boots much faster than Ubuntu Breezy or OpenSUSE 10 (50 seconds versus over 1 minute 10 seconds) on my machine here (it's a fairly old slow machine). It's is far far faster than Mandriva 2005.0 and currently seems to have some sort of lead here. My hat (Fedora?) goes off to the Mandriva developers for improving (and integrating other's improvements) so much over their previous release in this area (that's not to say that further improvements won't be forthcoming from the other distros though).
However boot up and shutdown vary enourmously from machine to machine (and also by how fragmented your partition has become) and I don't have a printer or bluetooth etc. attached to this machine.
I have been using Mandrake/Mandriva for years and my Netgear MA311 has worked since around 2002. This is a Prism 2/2.5/3 chipset card.
In 2005.0, for some reason the wifi drivers for such chipsets was defaulted to hostap rather than orinoco. Due to driver bugs this meant that the ncessary extra magic did not ncessarily make it into modules.conf causing wifi to fail to come after first boot until the necessary line was added. See this Mandriva bugzilla bug.
2006 now appears to default to orinoco and required no extra prodding to get working in my tests of the Beta releases. Of course I don't have your exact machine so no one will be able to tell you for sure.
As others have said, make sure that your DWL 520 really is prism based. That model of card has been through lots of chipsets (upping the model number each time would have saved the world a lot of pain but I guess marketing and cost to reprint stuff won over).
It's a step up from ASCII in that you have formatting but I have no idea how well it copes with non basic ASCII characters. You will always have the option to grovel some text out of the document by opening it up in a text editor if the worst comes to the worst though.
These days RTF is a defacto MS format rather than a good interchange format between differing products but that's an aside from archiving (unless you change to a different platform).
I don't want to detract from your main point (which is that configuration is hard) but can't they pay someone else to do the forking? Who says that everyone has to be programmer in order to have their own fork?
I've just done a few tests and the only time the charset in content-type was not the specified one was when apache returned a redirect to another document (e.g. the case when you have a directory but you leave off the trailing slash). To the best of my knowledge Apache does not look within files before deciding what HTTP charset to send - it is purely done by file extension.
If the HTTP charset IS changing by document contents it might be because you have a proxy trying to do something clever to the headers between you and the server.
I didn't know so that's why I asked : ) That original Latin-1 behaviour sounds "correct" though. You might want to see my other comment on how HTTP charsets are supposed to take priority.
I think the point is many people run their (desktop) firewalls in a stateful "let anything out, and let anything in associated with outgoing connection back" manner so passive does appear to solve the problem for them.
Just to be awkward though, I prefer my huge files to be served via rsync.
I'm not 100% clear on what you are after but don't things like suphp and cgiwrap allow you to use a system binary but have users run the scripts under their own uid?
BTW, another option is 3) Escalate priviledge for scripts through setuid binaries (but this carries its own risks).
From the spec you linked, we read that the HTTP charset should take prescedence over any other charset:
To sum up, conforming user agents must observe the following priorities when determining a document's character encoding (from highest priority to lowest):
1. An HTTP "charset" parameter in a "Content-Type" field.
2. A META declaration with "http-equiv" set to "Content-Type" and a value set for "charset".
3. The charset attribute set on an element that designates an external resource.
As it turns out this isn't the whole truth (there can be heuristics) but what you are doing is rather dicey (read as: may give different results in different browsers or in the future). Make your charsets agree or you might cause yourself problems.
Have you checked to see whether uppercase charsets were ignored anyway (before you set AddDefaultCharset)? Remember the default charset is iso-8859-1 so you will have to use another one.
Unfortunately there are lots of reasons why it could be going wrong.
Honestly the best thing you can do is to post to the nv forum and then send a message to NVidia themselves (see the "CONTACTING US" part of the README that came with the drivers). You may well have hit a genuine problem (Which only NVidia will be able to confirm) but you don't provide enough information - (which card? What were you doing at the time? Were there any Xids? The list goes on and on). *Don't post the reply to those questions here* though - go to NVidia. That way you can skip the speculation. Good luck.
The reason why your "porn, MP3s and NTFS" aren't supported without extra downloads is because of patents/lack of specifications.
I don't blame you for not knowing this because if you aren't going to go off and write this stuff yourself then it is deadly boring and tedious but it doesn't stop it being true.
MP3s are patented. The worry is if someone produces software to make or play MP3s they have pay royalties or risk being sued. This would ramp up the cost of Linux.
Porn. I have no idea what format your porn is in (I'm guessing it's video). Let's say it's some sort of MPEG4 video (this covers WMF too). MPEG4 is patented too and you need to pay royalties if you produce software that plays or makes them. This would ramp up the cost of Linux.
NTFS writing is off because it is dangerous/limited and can quite easily destroy the partition. This is mostly because NTFS has no documentation outside of MS. It may also be patented by MS. Licencing it would ramp up the cost of Linux.
By the time a company has paid for all these things and passed the cost to its customers (remember your distro can't possibly be free in either sense now) you have something more expensive than Windows which still doesn't run Windows programs. Why wouldn't you just use Windows?
The only system that could possibly win the game of being more Windows than Windows is MacOS and that comes with implicit expectation that you won't try and do things like read NTFS partitions or run Windows software outside of an emulator.
First up the GPL v3 doesn't yet exist yet. To that extent this speculation is moot.
Incompatiblity between the GPL v3 and v2 only occurs if the "or later version" was left out of the licence. This tends to only affect developers trying to swallow other GPL projects which is not common. Most linking cases (e.g. GPL libraries) are already taken care of using the runtime exception.
Anyway, there are plenty of open source licences (many belonging to big projects) that are incompatible with the GPL v2 so I don't think having v3 "incompatible" with v2 is not going to be a big problem.
Sure there are going to be those who don't like the "web app" clause (assuming it goes in and doesn't get commonly exempted) but there are those who don't like the "you must release the source if you redistribute the binary etc" clause that is in v2 today. The only way to please the most users is for programs to be public domain which remarkably few developers do.
I reckon OpenBSD is a better example than FreeBSD because they have replaced more GPL userland bits and pieces with BSD equivalents. However if you are only concerned with running this software rather than contributing to it then it simply doesn't affect you whether the software is non-GPL OSS or GPL so long as its community is vibrant and it does what you need it to.
You are (currently) hard pressed to get the *complete stack* of possible software in only GPL bits just as you are to unable to get one out of only BSD-like bits. Thankfully everyone but the purists have a choice - to use software adhering to more than one licence.
I completely agree with your premise that OS (as in Operating System) = kernel although a lot of the problem is that words and phrases take on new meanings. If enough people think OS = "The software stack required to implement a WIMP GUI and popular apps" (as happens when people say "The Windows Operating System") then the phrase picks up second meaning. This can lead to words and phrases being eventually redefined.
Now the bit I do take issue with is saying that x.org is an essential part of Fedora's OS. I know there are configuration tools for the kernel that require QT and GTK but there's a curses based config tool too. I think Red Hat would have a hard time of saying X11 was a part of the OE but it is "optional" (providing you don't run GUI programs) anyway. I am vaguely aware that on Windows signifcant parts of the GUI subsystem run in kernel space so I reckon you have a good case for saying GUI stuff is part of the "OS" there but not on Linux based distros...
The IE/X11 analogy also breaks down because it's possible to install Fedora without X11 but it's not (at least until Windows 2003?) possible to *install* Windows without IE. Nor is there anything which implements all the same functions as IE (unlike your sh example where you could use sash/ksh/csh). The fact that there COULD be something is somewhat by the by unless THERE IS something. It's like saying something could replace Quicktime (the movie player) on the Mac. The public API is documented so you could just rip out the current libraries and just slot in VLC as replacement right (even in embedded movies)?
I am answering this from a package only point of view and this is MY understanding - it may not be correct (ask someone from Red Hat, it might be that RHEL really is branched from a Fedora). Since you mentioned RHEL we can turn it round and say what is RHEL? RHEL is mostly a branch of rawhide + non Red Hat packages. Fedora Core (note the Core) is completely a branch of rawhide (as in "all Fedora Core packages are/were in rawhide", not "all of rawhide is in Fedora Core") and the RHEL branch is often taken at a similar (but not identical) point to that of the most recent Fedora. Where appropriate branch fixes make their way back to the trunk and other branches. Parts of the Fedora branch may also "rebase".
Fedora is basically (gross simplification) a more branched version of rawhide than RHEL.
There are bunch of other things that make Fedora which I deleted from this post (I cut it down considerably because I don't have room to cover them all).
Actually it's quite common in an algorithm to trade space for time. Algorithms tend to work on abstract levels where there's lots of one resource in order to trade off. In fact many algorithms are concerned precisely about abstraction over anything else.
The other catch is you've got to be wary of optimisations that complicate code. It used to be the case that gotos could provide a fast way to short circuit your way out of code at a cost to a reader trying to follow the code's logic.
Examples of what? Examples of them *selling* non-proprietary software (which raises the question "does RH sell software or do they sell support?")? Or examples of them *developing* non-proprietary software?
Remember though, for software to be proprietary you need to be able to (legally) *prevent sharing and use of the source*. If you can do that it's not proprietary because ownership is no longer under only your control.
RHEL4 has plenty of its packages based off those in FC3 but my understanding is that is that the packages for both FC and RHEL come from rawhide. Now because FC3 isn't a rolling distro (in the same way as something like Gentoo or Debian unstable) there are things in rawhide that MAY never make their way into FC3 because they came out after FC3 was released. However those packages (or some future version) will turn up in FC4.
So if one assumes that RHEL4 was based off rawhide and rawhide had progressed past FC3 then you would wind up with later packages in RHEL4. No conspiracy I'm afraid (try a different studying technique next time;). It would also suggest that the packages in FC4 will be the same or newer than those in RHEL4...
A RHEL's purpose is to be a static platform for X years. It gets a few security and bug fixes and that's about it. Nowt to do with oxygen, rail travel or eating ones own kind - just that subscription paid support allows the expensive and boring process of backporting to be done for yearly releases.
Oh yeah, don't say "closed" by itself in reference to software unless you mean the source code is closed - you are muddying the waters. Having the source code available is precisely why projects like CentOS and Whitebox exist and saying RHEL is closed comes across purposefully confusing with the intent to mislead.
Back in the days of Winamp 2.91 there was a bug that meant when Winamp started on shuffle it would always play the first track. I think the behaviour of shuffle was also tweaked by the legendary Ryan Geess to not repeat a track until 50% of the other songs had been played. This may well have had statistical side effects.
Slashdot Mirror
Prism 2.5/3 based cards have supported scanning for years on Linux BUT (there had to be a but becaues your experience is different right?) it depended on the driver (HostAP and linux-wlan had could by default, orinoco needed drivers). Since the 2.4.13 kernel the orinoco drivers have supported scanning in the mainstream Linux kernel without patches.
As an OpenBSD user I can tell you that it does not have the broadest support. Let me back this counterclaim up.
OpenBSD supports the following chipsets (as taken from the OpenBSD i386 hardware compatibility page on 19th November 2005:
ADMtek, Aironet, Atheros, Atmel, Centrino (2100, 2200), Prism 2.5/3, Ralink (2500), Raytheon and Realtek
Now by comparison a Linux distro (e.g. Ubuntu) can have support for the following chipsets (list taken from Linux WLAN Howto cross referenced against Ubuntu) on 19th November 2005:
ADMtek, Aironet, Atheros, Atmel, Centrino (2100, 2200), Prism 2.5/3, Ralink (2400, 2500), Prism GT, Raytheon, Texas Instruments ACX100/110, Wavelan
A pretty similar list wouldn't you say (OpenBSD has Realtek which Ubuntu doesn't, Ubuntu has Ralink 2400, Prism GT, ACX100/110 and Wavelan which OpenBSD doesn't)? In fact, Linux has Realtek drivers too but as they aren't shipped in Ubuntu I left them off the list. Linux also has ndiswrapper and Linuxant Driverloader allowing the use of Win32 drivers but I'm discounting non native drivers.
Chipsets that aren't supported by either OS that are significant:
Broadcom - There's a heck of a lot of Broadcom stuff out there and no sign of open drivers ever. That's their perogative but this stuff is all over the place...
Marvel - a new 802.11g player as far as I can tell. No open source drivers that I know of so far.
For the meantime, chipsets like Broadcom mean that open source OSes will always have a more troublesome than Windows with random wireless drivers. Choose carefully and don't reward vendors with non free drivers where possible.
I'm amazed you haven't come across the linux wlan chipset list. Although it hasn't been updated for nearly two years it's about the most exhaustive list of what is in each card that I've seen. If 802.11b is fine then there are still Prism 2.5/3 based cards out there that work very well. I also recently tested a Atheros based PCMCIA card (I'm purposely not naming manufacturers) which also worked well but required a small binary lump.
As mentioned elsewhere, support for wifi isn't spotty - it's support for certain chipsets that is (alas this is also extends to various USB wifi devices too). If you buy (for example) a Broadcom based card I'm afraid you're in for a rough ride because Broadcom don't want to release open source drivers. There's no point getting upset - Broadcom are within their rights to do so and Linux isn't binary only friendly. It's the way things are.
The best advice I can give is get a peek inside the box so you know which chipset you are buying. Manufacturers are lazy and try to avoid changing model numbers significantly even if they swap chipsets because it means all the other materials can stay exactly the same. If you are going on someone else's information be extremly weary of ANY deviation to the model name/number/revison. Things like a +, extra letters or revision increase of any amount can mean chipset changes.
I'm not sure if it's simply down to starting less stuff but Mandriva 2006.0 boots much faster than Ubuntu Breezy or OpenSUSE 10 (50 seconds versus over 1 minute 10 seconds) on my machine here (it's a fairly old slow machine). It's is far far faster than Mandriva 2005.0 and currently seems to have some sort of lead here. My hat (Fedora?) goes off to the Mandriva developers for improving (and integrating other's improvements) so much over their previous release in this area (that's not to say that further improvements won't be forthcoming from the other distros though).
However boot up and shutdown vary enourmously from machine to machine (and also by how fragmented your partition has become) and I don't have a printer or bluetooth etc. attached to this machine.
I have been using Mandrake/Mandriva for years and my Netgear MA311 has worked since around 2002. This is a Prism 2/2.5/3 chipset card.
In 2005.0, for some reason the wifi drivers for such chipsets was defaulted to hostap rather than orinoco. Due to driver bugs this meant that the ncessary extra magic did not ncessarily make it into modules.conf causing wifi to fail to come after first boot until the necessary line was added. See this Mandriva bugzilla bug.
2006 now appears to default to orinoco and required no extra prodding to get working in my tests of the Beta releases. Of course I don't have your exact machine so no one will be able to tell you for sure.
As others have said, make sure that your DWL 520 really is prism based. That model of card has been through lots of chipsets (upping the model number each time would have saved the world a lot of pain but I guess marketing and cost to reprint stuff won over).
It's a step up from ASCII in that you have formatting but I have no idea how well it copes with non basic ASCII characters. You will always have the option to grovel some text out of the document by opening it up in a text editor if the worst comes to the worst though.
The main problem with RTF is that there are several mildly incompatible implementations to choose from and consequently formatting (along with other features) may not be preserved/read properly between differing programs.
These days RTF is a defacto MS format rather than a good interchange format between differing products but that's an aside from archiving (unless you change to a different platform).
I don't want to detract from your main point (which is that configuration is hard) but can't they pay someone else to do the forking? Who says that everyone has to be programmer in order to have their own fork?
I've just done a few tests and the only time the charset in content-type was not the specified one was when apache returned a redirect to another document (e.g. the case when you have a directory but you leave off the trailing slash). To the best of my knowledge Apache does not look within files before deciding what HTTP charset to send - it is purely done by file extension.
If the HTTP charset IS changing by document contents it might be because you have a proxy trying to do something clever to the headers between you and the server.
Am I right?
I didn't know so that's why I asked : ) That original Latin-1 behaviour sounds "correct" though. You might want to see my other comment on how HTTP charsets are supposed to take priority.
I think the point is many people run their (desktop) firewalls in a stateful "let anything out, and let anything in associated with outgoing connection back" manner so passive does appear to solve the problem for them.
Just to be awkward though, I prefer my huge files to be served via rsync.
I'm not 100% clear on what you are after but don't things like suphp and cgiwrap allow you to use a system binary but have users run the scripts under their own uid?
BTW, another option is 3) Escalate priviledge for scripts through setuid binaries (but this carries its own risks).
As it turns out this isn't the whole truth (there can be heuristics) but what you are doing is rather dicey (read as: may give different results in different browsers or in the future). Make your charsets agree or you might cause yourself problems.
Have you checked to see whether uppercase charsets were ignored anyway (before you set AddDefaultCharset)? Remember the default charset is iso-8859-1 so you will have to use another one.
Unfortunately there are lots of reasons why it could be going wrong.
Honestly the best thing you can do is to post to the nv forum and then send a message to NVidia themselves (see the "CONTACTING US" part of the README that came with the drivers). You may well have hit a genuine problem (Which only NVidia will be able to confirm) but you don't provide enough information - (which card? What were you doing at the time? Were there any Xids? The list goes on and on). *Don't post the reply to those questions here* though - go to NVidia. That way you can skip the speculation. Good luck.
The reason why your "porn, MP3s and NTFS" aren't supported without extra downloads is because of patents/lack of specifications.
I don't blame you for not knowing this because if you aren't going to go off and write this stuff yourself then it is deadly boring and tedious but it doesn't stop it being true.
MP3s are patented. The worry is if someone produces software to make or play MP3s they have pay royalties or risk being sued. This would ramp up the cost of Linux.
Porn. I have no idea what format your porn is in (I'm guessing it's video). Let's say it's some sort of MPEG4 video (this covers WMF too). MPEG4 is patented too and you need to pay royalties if you produce software that plays or makes them. This would ramp up the cost of Linux.
NTFS writing is off because it is dangerous/limited and can quite easily destroy the partition. This is mostly because NTFS has no documentation outside of MS. It may also be patented by MS. Licencing it would ramp up the cost of Linux.
By the time a company has paid for all these things and passed the cost to its customers (remember your distro can't possibly be free in either sense now) you have something more expensive than Windows which still doesn't run Windows programs. Why wouldn't you just use Windows?
The only system that could possibly win the game of being more Windows than Windows is MacOS and that comes with implicit expectation that you won't try and do things like read NTFS partitions or run Windows software outside of an emulator.
First up the GPL v3 doesn't yet exist yet. To that extent this speculation is moot.
Incompatiblity between the GPL v3 and v2 only occurs if the "or later version" was left out of the licence. This tends to only affect developers trying to swallow other GPL projects which is not common. Most linking cases (e.g. GPL libraries) are already taken care of using the runtime exception.
Anyway, there are plenty of open source licences (many belonging to big projects) that are incompatible with the GPL v2 so I don't think having v3 "incompatible" with v2 is not going to be a big problem.
Sure there are going to be those who don't like the "web app" clause (assuming it goes in and doesn't get commonly exempted) but there are those who don't like the "you must release the source if you redistribute the binary etc" clause that is in v2 today. The only way to please the most users is for programs to be public domain which remarkably few developers do.
I reckon OpenBSD is a better example than FreeBSD because they have replaced more GPL userland bits and pieces with BSD equivalents. However if you are only concerned with running this software rather than contributing to it then it simply doesn't affect you whether the software is non-GPL OSS or GPL so long as its community is vibrant and it does what you need it to.
You are (currently) hard pressed to get the *complete stack* of possible software in only GPL bits just as you are to unable to get one out of only BSD-like bits. Thankfully everyone but the purists have a choice - to use software adhering to more than one licence.
I completely agree with your premise that OS (as in Operating System) = kernel although a lot of the problem is that words and phrases take on new meanings. If enough people think OS = "The software stack required to implement a WIMP GUI and popular apps" (as happens when people say "The Windows Operating System") then the phrase picks up second meaning. This can lead to words and phrases being eventually redefined.
Now the bit I do take issue with is saying that x.org is an essential part of Fedora's OS. I know there are configuration tools for the kernel that require QT and GTK but there's a curses based config tool too. I think Red Hat would have a hard time of saying X11 was a part of the OE but it is "optional" (providing you don't run GUI programs) anyway. I am vaguely aware that on Windows signifcant parts of the GUI subsystem run in kernel space so I reckon you have a good case for saying GUI stuff is part of the "OS" there but not on Linux based distros...
The IE/X11 analogy also breaks down because it's possible to install Fedora without X11 but it's not (at least until Windows 2003?) possible to *install* Windows without IE. Nor is there anything which implements all the same functions as IE (unlike your sh example where you could use sash/ksh/csh). The fact that there COULD be something is somewhat by the by unless THERE IS something. It's like saying something could replace Quicktime (the movie player) on the Mac. The public API is documented so you could just rip out the current libraries and just slot in VLC as replacement right (even in embedded movies)?
Sounds like Shuan is getting his own show.
I am answering this from a package only point of view and this is MY understanding - it may not be correct (ask someone from Red Hat, it might be that RHEL really is branched from a Fedora). Since you mentioned RHEL we can turn it round and say what is RHEL? RHEL is mostly a branch of rawhide + non Red Hat packages. Fedora Core (note the Core) is completely a branch of rawhide (as in "all Fedora Core packages are/were in rawhide", not "all of rawhide is in Fedora Core") and the RHEL branch is often taken at a similar (but not identical) point to that of the most recent Fedora. Where appropriate branch fixes make their way back to the trunk and other branches. Parts of the Fedora branch may also "rebase".
Fedora is basically (gross simplification) a more branched version of rawhide than RHEL.
There are bunch of other things that make Fedora which I deleted from this post (I cut it down considerably because I don't have room to cover them all).
Actually it's quite common in an algorithm to trade space for time. Algorithms tend to work on abstract levels where there's lots of one resource in order to trade off. In fact many algorithms are concerned precisely about abstraction over anything else.
The other catch is you've got to be wary of optimisations that complicate code. It used to be the case that gotos could provide a fast way to short circuit your way out of code at a cost to a reader trying to follow the code's logic.
Examples of what? Examples of them *selling* non-proprietary software (which raises the question "does RH sell software or do they sell support?")? Or examples of them *developing* non-proprietary software?
Remember though, for software to be proprietary you need to be able to (legally) *prevent sharing and use of the source*. If you can do that it's not proprietary because ownership is no longer under only your control.
RHEL4 has plenty of its packages based off those in FC3 but my understanding is that is that the packages for both FC and RHEL come from rawhide. Now because FC3 isn't a rolling distro (in the same way as something like Gentoo or Debian unstable) there are things in rawhide that MAY never make their way into FC3 because they came out after FC3 was released. However those packages (or some future version) will turn up in FC4.
;). It would also suggest that the packages in FC4 will be the same or newer than those in RHEL4...
So if one assumes that RHEL4 was based off rawhide and rawhide had progressed past FC3 then you would wind up with later packages in RHEL4. No conspiracy I'm afraid (try a different studying technique next time
A RHEL's purpose is to be a static platform for X years. It gets a few security and bug fixes and that's about it. Nowt to do with oxygen, rail travel or eating ones own kind - just that subscription paid support allows the expensive and boring process of backporting to be done for yearly releases.
Oh yeah, don't say "closed" by itself in reference to software unless you mean the source code is closed - you are muddying the waters. Having the source code available is precisely why projects like CentOS and Whitebox exist and saying RHEL is closed comes across purposefully confusing with the intent to mislead.
Back in the days of Winamp 2.91 there was a bug that meant when Winamp started on shuffle it would always play the first track. I think the behaviour of shuffle was also tweaked by the legendary Ryan Geess to not repeat a track until 50% of the other songs had been played. This may well have had statistical side effects.