This could well be a blow to Linux's increasing popularity as it addresses some fundamental concerns about NT.
No, It's a blow to NT - it will forstall the potential deployment of a bastardized, not-quite-standard Microsoft imitation of sendmail. It will also drive home the point to many a PHB that you can get sendmail on NT, but not for free. On Linux it's free. Contrary to popular belief, such things do matter to PHB's. [
if all commerce occurs online, then where will state governments get their funding from?
Sounds like a convincing argument, but it's wrong:
sin taxes (booze etc)
state income tax (do you yankees have that?)
fuel tax
lotteries (tax the stupid)
fees for government services
other taxes I forgot
The truth is that taxes are, right now, too high. Way too high. As commerce moves wholesale to the net it's a perfect time to give Joe Overburdened Taxpayer a break.
The "else" is redundant since the object of the if is an unconditional jump. Sorry to be pedantic, and I realize the comment was the point. Still, flawed code posted here ought not to escape silently, lest novice programmers come to believe that such sloppy style has been tacitly blessed. --
Can't do it. The laws say that not only can't you have encryption, you can't even have any hooks that can be used for encryption.
Any computer has "hooks" that can be used for this purpose, and therefore should be illegal to export. Consider that you can take an email program, and patch in encryption hooks with a debugger if you have to. That means the program has hooks in it because it has places where you can patch in the encryption code.
Ok, now that we can see how silly and unenforceable the "no hooks" policy is (as long as you don't put in hooks that are specifically for encryption everything should be allright), lets consider how our encryption program could hook itself into mozilla. Hmm, remember, Mozilla is all held together with scripty-goo, and consists to a large extend of dynamically loaded modules. There's a way, for sure, and even an elegant way that fits nicely with the Mozilla architecture. Or, maybe there should be a law that browsers with scriptable components are illegal to export? --
It's apparently the "hip" thing to bash on Linux and Linux users now...
Bashing on linux is hip because it's useful. In fact, the criticisms of Linux here are a whole lot more useful than the praises, since they help get important issues addressed. On the other hand, preaching to the choir accomplishes very little of value. --
Yes, there should be a core group of moderators whose specific task is to demote trolls, with an unlimited budget for demotion. These people should be trusted and permanent, have lots of free time, and, importantly, should have the same limited budget for promotion as any other moderator. These people would be our firemen - they're there for the specific purpose of putting out fiers. --
We need more moderator points all right. For example, I normally read at threshold 2, otherwise I'd never be able to get through all the articles I'm interested in in a reasonable time, and when I moderate I typically read an entire article in flat mode with threshold of 0 or 1, looking for articles to promote (and occasionally demoting a troll). Even for a single article this takes a significant amount of time, and I *never* seem to have quite enough points to do the job, which for me, mainly consists of moderating up enough articles so that people like me can read the way I do. I'd say a doubling would do the trick, and leave a few points over for troll-hunting. I really don't care how *often* the points show up as long as it's not so often that moderating time starts getting taken out of reading and contributing time. --
How much more do they benefit from improving the development libraries and utilities than they would from using those development libraries and utilities to do something?
They benefit from both. (1) Better development tools and platform = better apps = more Red Hat CD's sold (2) Better apps = more Red Hat Cd's sold. It makes most sense to support both, since they now can. Of the two, apps are the big weak point (though better source level debugging is a crying need on the development front). Linux needs KOffice and GnomeOffice. It needs those both to be better, lighter, tighter, faster, more useful, more stable, more scriptable and better integrated than Microsoft Office. Both groups could do worse than look closely at Microsoft Works - this is a seriously slickly designed package. Then go beyond it. Gnome already has a spreadsheet that looks and acts like Excel although it's far from finished. Good start - now make the ui scriptable and expose the internal api via Corba - do whatever it takes to let the rest of us geeks take it further. I'm starting to ramble, I'll stop now. --
because I'm automating workflow, rather than handling financial transactions, I don't miss the features like commits or rollbacks that MySQL leaves out (specifically to increase the db's speed)
But wouldn't you rather have transactions that not, if there were no speed hit?
I've been working on the problem of how to do transaction commit/rollback for a long time now (years) and I'm pretty sure I've got a solution. The basic idea has been tested extensively in a dbms I wrote in, hmm, 1989. The goal then was to have automatic rollback to the last commited state of the db in the event of an unexpected termination (crash/reset). In the event, there were thousands of intended and unintended termination incidents none of which ever broke the automatic rollback algorithm. This algorithm, which I call "sync tree" has a few other nice properties including:
No journal file to maintain (although journalling could be used at the same time)
Fast - the algorithm performs in near-linear time (vs db size and transaction count) with a very small K.
Simple - hehe, kind of.
Can work transparently at the file system level
Instant recovery - after unexpected termination the system reawakens at the state of the last commit.
Goodbye auto-fsck - when implemented at the file system level, the file system likewise reawakens in a consistent state and doesn't need to be fsck'd (can you say "instant off"?).
My next step with this is to hack it into ext2 and try it out, myself as the guinea pig. Is anybody else interested in this project? You can email me at: danielphillips (at) yahoo (dot) com or phillips (at) dowco (dot) com. --
The GPL'd version is 3.20.32a; the version I got with Redhat 6.0 and am still using is 3.21.30. This is so recent that I'd think the logical next step would be just to put the current CVS version under GPL.
Meanwhile, the PostgreSQL team is adding new features at a truly impressive rate. The good effects of having TWO full blown database systems competing with each other in open source can't be underestimated.
I think I know:) It's because on Linux you're supposed to use shared libraries to make your code smaller, but because the shared libraries change so often, it's unsafe to release a big monster like Mozilla and hope it will somehow get lucky enough to have used NO obsolete, changed, recently broken library functions. So it winds up having to be statically linked, or using its own custom-compiled libraries. WP8 is like this too, and Netscape.
That sucks of course, but at least it means the program usually runs, and that doesn't suck. Getting it to work with real shared libraries is a kind of a final development phase thing and I really hope that it works out, because who need yet another bloated monster stomping its great big lizard feet all over their memory? --
I've been evaluating both of them as a replacement for MS SQL and MS Access - they're both great pieces of work, and the competition between them is causing both to move forward rapidly. However, I'd say that of the two PostgreSQL is the only one I can sell to management at this time, mainly because of one feature: transactions.
MySQL developers will have to wake up to the fact that the need for transactions isn't going to go away, ever. Granted, there may be some performance loss with this feature, but it COULD be optional.
A couple years ago when Microsoft was in the process of killing off DOS by wooing game developers away from DOS to Windows, the games developer news groups suddenly became flooded with rabid pro-microsoft posters - if you go look in deja news you'll see such posts amounted to as much as half the traffic at the time. This happened again around the time Microsoft was trying to bury OpenGL as a 3D gaming platform (it didn't work, only because of ID's John Carmack). Is this part of a pattern? Personally, I'm dead convinced it is.
On slashdot, the technique doesn't work because of moderation. And because of the general level of cluefullness. And because, hmm, not many plainclothes microsoft agents would want to take the risk of being pecked to death by a flock of crazed penquins.
A new IM protocol can catch on in the same way that ICQ caught on in the first placing... by harnessing the power of teen chat. The client binaries have to be small, tight, and easy to send to your friends by IRC DCC or email. The user interface has to be cute... and how about *less clunky* at the same time. Free servers have to be in place, and somebody has to pay for that bandwidth. Maybe the chat client should include a link back to the server's web page so the server operator can put ads there, or something. Just an idea. There has to be a way to get free servers going.
While we're at it, lets put a Linux kernel into the BIOS flash too:)
Re:While on the subject of X windows...
on
Fifteen Years of X
·
· Score: 1
The fact that it doesn't happen in Windows is a little puzzling. Perhaps the gain is turned down more in Windows? It could also be the driver. The driver can actually make a difference in the amount of interference you get.
It most certainly can. Some instructions generate a lot of RFI, for example, multiplies and divides... because these instructions generate activity on a vast expanses of chip real-estate that normally like dormant. Not only do they generated RFI, but heat as well, which you'll notice if you're running a laptop (the fan speeds up.)
Seems to me that the mouse wheel (which I love) could do that job quite intuitively. Wheel up to push, wheel down to pull.
Well, actually, the wheel is better used for rate-of-climb control... you can grab with the left button, then use the mouse to drive around in 3D space using an Ultima-Underworld style control system, towing your 3D objects with you. (I've been doing it for a few years now:) Err, that's if you want to fly around inside your windows. If you want to sit in one place and pull them around, then when you want is thumbwheel towards you for pull and away for push, leaving your fingers free for grabbing.
BTW, flying around inside a windowing system is LOTS of fun, once you've done it you'll find 2D graphics kinda dull...
Of course, if the machine is too hosed to reboot, this obviously won't work. A hardware solution is required that turns the break signal into a hard reset.
The software solution won't work any worse than Ctrl-Alt-Del (also handled in software). Also, sure, you can reboot from an interrupt context, in fact, it's often hard to *avoid* rebooting when you're in there:) (remember the good old triple fault?)
I'm typing this on a VAIO F150 running Red Hat 6.0, kernel 2.2.9 (yes, I know I should be running.10 by now).
This is what's wrong with my VAIO:
- Built in Winmodem - it hits high tranfer rates, up to 55kbytes/sec on download of compressed files (I've seen it) - BUT it sucks because the specs are closed, forcing me to use an extra modem when running Linux, and also, the winmodem *heats up the processor* (no kidding - the fan comes on high when a download is in progress).
- the advertised uptime on batteries is just a lie - you will get an hour at most off one battery. The power-hungery winmodem and big display just make this worse. **You will need that second battery*** (it fits into the floppy bay).
- the DSTN screen is a little iffy - it's nice and bright, but you get rectangular bleeding from bright colored regions across the whole screen, both horizontally and vertically
- the power management bios support is very questionable... it worked for a couple of days under Windows, then just started crashing after that. Under Linux, suspend/restart works the first time after booting, the 2nd time you'll die a horrible death from which only a hard power-down will get you back. Once I even had to pull the battery to bring it back to life.
- There is some strange interaction with LILO or kernel (I haven't isolated it yet) that causes root to fail to mount (100% repeatable) if LILO times out - to boot Linux you have to hit Tab, then enter.
- There are some very curious omissions in the bios setup, such as not being able to turn off the cute little sound it makes every time it boots (yes, you can turn the volume down, but you can't turn it off). You can't get rid of the "SONY" screen either, which hangs around far longer than necessary. Apparently, MS knows how to get rid of the bios preboot screens, because the annoying Windows preboot screen goes away after you add logo=0 to MSDOS.SYS and start/exit windows, or else the VAIO knows how to read the msdos.sys file (yikes!).
- No way in the bios to turn off pressure-sensitive mouse click emulation - this would be great if it worked reliably but the truth is it doesn't - you get a *lot* of spurious clicks, even when hardly touching the pad.
- The speakers really suck.
Um, that's about it with my gripes, the bottom line is, with a 300 MHz Pentium II processor (reported as a celeron by proc) it really kicks butt, recompiling the kernel in about 5 minutes. It has a nice, fullsize keyboard and a large bright screen (mine has 1 dead pixel, I'll have to talk to Sony about that before my warrantee expires). Now, if Sony would just do something about the warts - lose the winmodem, debug the power management, etc., I'd be a really happy camper.
(if someone from Sony is reading this, feel free to drop me a line at phillips (at) dowco (dot) com)
All you're doing is taking what would've been called 2.2.25 and calling it 2.4.0
Not at all. According to the usual procedure, the goodies currently under development in the 2.3.xx tree will make it into 2.4, not into 2.2.xx (with few exceptions). Then guess what? MORE goodies will start to appear, in 2.5.xx.
Soooooo... the bottom line is: the sooner we see 2.4, the better.
the problem is not backwards compatibility with deprecated DOM extensions (which were rarely used), but backwards compatiblity with (soon to be obselete) NS4 browsers trying to render sites that it can't.
Caveat: I haven't yet checked to see whether the following suggestions have already been implemented...
All support for all nonstandard extensions should be hived off into their own DLL/Module. Any use of such extensions should cause a warning (suppressable by the user, in the same manner as a security warning).
For marketing and pragmatic reasons it is important to support some nonstandard extensions. It is equally important to support them in a way that clearly marks them as deprecated. Purists can simply turn off support for all nonstandard extensions and deal with them by emailing the webmasters of sites using such extensions. Purists can even delete the relevant DLL/module.
it may be time to fork the tree (or be ready to) and proceed with Open Kaffe (GNU Kaffe? some other name?)
Yes, with due respect to Transvirtual, it's way to dangerous to leave the majority of the the JVM code copyrighted by a company that can so easily be attacked/controlled by openSW-unfriendly forces, be they from Redmond or anywhere else. The copyright holder decides the form of licence, and right now it's GNU, but will it stay that way after a controlling interest in Transvirtual changes hands? Even if it stays GNU, it won't do to leave control of the CVS in the hands of anything other than a not-for-profit and democratic organization. Forking the tree now is the easiest way to get what we want and need.
How about LAVA (Linux jAVA; Liberated jAVA for non-linux platforms)?
Slashdot Mirror
This could well be a blow to Linux's increasing popularity as it addresses some fundamental concerns about NT.
No, It's a blow to NT - it will forstall the potential deployment of a bastardized, not-quite-standard Microsoft imitation of sendmail. It will also drive home the point to many a PHB that you can get sendmail on NT, but not for free. On Linux it's free. Contrary to popular belief, such things do matter to PHB's. [
Sounds like a convincing argument, but it's wrong:
sin taxes (booze etc)
state income tax (do you yankees have that?)
fuel tax
lotteries (tax the stupid)
fees for government services
other taxes I forgot
The truth is that taxes are, right now, too high. Way too high. As commerce moves wholesale to the net it's a perfect time to give Joe Overburdened Taxpayer a break.
if(x == LASTCHAR) goto thats_it; else ++x;
The "else" is redundant since the object of the if is an unconditional jump. Sorry to be pedantic, and I realize the comment was the point. Still, flawed code posted here ought not to escape silently, lest novice programmers come to believe that such sloppy style has been tacitly blessed.
--
Can't do it. The laws say that not only can't you have encryption, you can't even have any hooks that can be used for encryption.
Any computer has "hooks" that can be used for this purpose, and therefore should be illegal to export. Consider that you can take an email program, and patch in encryption hooks with a debugger if you have to. That means the program has hooks in it because it has places where you can patch in the encryption code.
Ok, now that we can see how silly and unenforceable the "no hooks" policy is (as long as you don't put in hooks that are specifically for encryption everything should be allright), lets consider how our encryption program could hook itself into mozilla. Hmm, remember, Mozilla is all held together with scripty-goo, and consists to a large extend of dynamically loaded modules. There's a way, for sure, and even an elegant way that fits nicely with the Mozilla architecture. Or, maybe there should be a law that browsers with scriptable components are illegal to export?
--
It's apparently the "hip" thing to bash on Linux and Linux users now...
Bashing on linux is hip because it's useful. In fact, the criticisms of Linux here are a whole lot more useful than the praises, since they help get important issues addressed. On the other hand, preaching to the choir accomplishes very little of value.
--
Yes, there should be a core group of moderators whose specific task is to demote trolls, with an unlimited budget for demotion. These people should be trusted and permanent, have lots of free time, and, importantly, should have the same limited budget for promotion as any other moderator. These people would be our firemen - they're there for the specific purpose of putting out fiers.
--
We need more moderator points all right. For example, I normally read at threshold 2, otherwise I'd never be able to get through all the articles I'm interested in in a reasonable time, and when I moderate I typically read an entire article in flat mode with threshold of 0 or 1, looking for articles to promote (and occasionally demoting a troll). Even for a single article this takes a significant amount of time, and I *never* seem to have quite enough points to do the job, which for me, mainly consists of moderating up enough articles so that people like me can read the way I do. I'd say a doubling would do the trick, and leave a few points over for troll-hunting. I really don't care how *often* the points show up as long as it's not so often that moderating time starts getting taken out of reading and contributing time.
--
PNG
MNG
IMHO, MNG is what we need to be promoting, because image animation is a checkbox item, however irritating some of us old codgers may find it.
--
How much more do they benefit from improving the development libraries and utilities than they would from using those development libraries and utilities to do something?
They benefit from both. (1) Better development tools and platform = better apps = more Red Hat CD's sold (2) Better apps = more Red Hat Cd's sold. It makes most sense to support both, since they now can. Of the two, apps are the big weak point (though better source level debugging is a crying need on the development front). Linux needs KOffice and GnomeOffice. It needs those both to be better, lighter, tighter, faster, more useful, more stable, more scriptable and better integrated than Microsoft Office. Both groups could do worse than look closely at Microsoft Works - this is a seriously slickly designed package. Then go beyond it. Gnome already has a spreadsheet that looks and acts like Excel although it's far from finished. Good start - now make the ui scriptable and expose the internal api via Corba - do whatever it takes to let the rest of us geeks take it further. I'm starting to ramble, I'll stop now.
--
But wouldn't you rather have transactions that not, if there were no speed hit?
I've been working on the problem of how to do transaction commit/rollback for a long time now (years) and I'm pretty sure I've got a solution. The basic idea has been tested extensively in a dbms I wrote in, hmm, 1989. The goal then was to have automatic rollback to the last commited state of the db in the event of an unexpected termination (crash/reset). In the event, there were thousands of intended and unintended termination incidents none of which ever broke the automatic rollback algorithm. This algorithm, which I call "sync tree" has a few other nice properties including:
No journal file to maintain (although journalling could be used at the same time)
Fast - the algorithm performs in near-linear time (vs db size and transaction count) with a very small K.
Simple - hehe, kind of.
Can work transparently at the file system level
Instant recovery - after unexpected termination the system reawakens at the state of the last commit.
Goodbye auto-fsck - when implemented at the file system level, the file system likewise reawakens in a consistent state and doesn't need to be fsck'd (can you say "instant off"?).
My next step with this is to hack it into ext2 and try it out, myself as the guinea pig. Is anybody else interested in this project? You can email me at: danielphillips (at) yahoo (dot) com or phillips (at) dowco (dot) com.
--
The GPL'd version is 3.20.32a; the version I got with Redhat 6.0 and am still using is 3.21.30. This is so recent that I'd think the logical next step would be just to put the current CVS version under GPL.
Meanwhile, the PostgreSQL team is adding new features at a truly impressive rate. The good effects of having TWO full blown database systems competing with each other in open source can't be underestimated.
--
I think I know :) It's because on Linux you're supposed to use shared libraries to make your code smaller, but because the shared libraries change so often, it's unsafe to release a big monster like Mozilla and hope it will somehow get lucky enough to have used NO obsolete, changed, recently broken library functions. So it winds up having to be statically linked, or using its own custom-compiled libraries. WP8 is like this too, and Netscape.
That sucks of course, but at least it means the program usually runs, and that doesn't suck. Getting it to work with real shared libraries is a kind of a final development phase thing and I really hope that it works out, because who need yet another bloated monster stomping its great big lizard feet all over their memory?
--
I've been evaluating both of them as a replacement for MS SQL and MS Access - they're both great pieces of work, and the competition between them is causing both to move forward rapidly. However, I'd say that of the two PostgreSQL is the only one I can sell to management at this time, mainly because of one feature: transactions.
MySQL developers will have to wake up to the fact that the need for transactions isn't going to go away, ever. Granted, there may be some performance loss with this feature, but it COULD be optional.
--
A couple years ago when Microsoft was in the process of killing off DOS by wooing game developers away from DOS to Windows, the games developer news groups suddenly became flooded with rabid pro-microsoft posters - if you go look in deja news you'll see such posts amounted to as much as half the traffic at the time. This happened again around the time Microsoft was trying to bury OpenGL as a 3D gaming platform (it didn't work, only because of ID's John Carmack). Is this part of a pattern? Personally, I'm dead convinced it is.
On slashdot, the technique doesn't work because of moderation. And because of the general level of cluefullness. And because, hmm, not many plainclothes microsoft agents would want to take the risk of being pecked to death by a flock of crazed penquins.
--
All installations will have to go through the MS Install Wizard...
Will this work as well as MS's Hardware Installation Wizard?
*shudders*
Isn't that the one where the best strategy is to avoid any button marked "recommended"?
--
A new IM protocol can catch on in the same way that ICQ caught on in the first placing... by harnessing the power of teen chat. The client binaries have to be small, tight, and easy to send to your friends by IRC DCC or email. The user interface has to be cute... and how about *less clunky* at the same time. Free servers have to be in place, and somebody has to pay for that bandwidth. Maybe the chat client should include a link back to the server's web page so the server operator can put ads there, or something. Just an idea. There has to be a way to get free servers going.
While we're at it, lets put a Linux kernel into the BIOS flash too :)
The fact that it doesn't happen in Windows is a little puzzling. Perhaps the gain is turned down more in Windows? It could also be the driver. The driver can actually make a difference in the amount of interference you get.
It most certainly can. Some instructions generate a lot of RFI, for example, multiplies and divides... because these instructions generate activity on a vast expanses of chip real-estate that normally like dormant. Not only do they generated RFI, but heat as well, which you'll notice if you're running a laptop (the fan speeds up.)
(following up my own post) ooops... I meant 55 KBits/sec, or 5.5 KBytes/sec -> darn good for an analog modem on a normal (clean) telco line.
Seems to me that the mouse wheel (which I love) could do that job quite intuitively. Wheel up to push, wheel down to pull.
Well, actually, the wheel is better used for rate-of-climb control... you can grab with the left button, then use the mouse to drive around in 3D space using an Ultima-Underworld style control system, towing your 3D objects with you. (I've been doing it for a few years now:) Err, that's if you want to fly around inside your windows. If you want to sit in one place and pull them around, then when you want is thumbwheel towards you for pull and away for push, leaving your fingers free for grabbing.
BTW, flying around inside a windowing system is LOTS of fun, once you've done it you'll find 2D graphics kinda dull...
Of course, if the machine is too hosed to reboot, this obviously won't work. A hardware solution is required that turns the break signal into a hard reset.
:) (remember the good old triple fault?)
The software solution won't work any worse than Ctrl-Alt-Del (also handled in software). Also, sure, you can reboot from an interrupt context, in fact, it's often hard to *avoid* rebooting when you're in there
I still think I'd prefer a VAIO.
.10 by now).
I'm typing this on a VAIO F150 running Red Hat 6.0, kernel 2.2.9 (yes, I know I should be running
This is what's wrong with my VAIO:
- Built in Winmodem - it hits high tranfer rates, up to 55kbytes/sec on download of compressed files (I've seen it) - BUT it sucks because the specs are closed, forcing me to use an extra modem when running Linux, and also, the winmodem *heats up the processor* (no kidding - the fan comes on high when a download is in progress).
- the advertised uptime on batteries is just a lie - you will get an hour at most off one battery. The power-hungery winmodem and big display just make this worse. **You will need that second battery*** (it fits into the floppy bay).
- the DSTN screen is a little iffy - it's nice and bright, but you get rectangular bleeding from bright colored regions across the whole screen, both horizontally and vertically
- the power management bios support is very questionable... it worked for a couple of days under Windows, then just started crashing after that. Under Linux, suspend/restart works the first time after booting, the 2nd time you'll die a horrible death from which only a hard power-down will get you back. Once I even had to pull the battery to bring it back to life.
- There is some strange interaction with LILO or kernel (I haven't isolated it yet) that causes root to fail to mount (100% repeatable) if LILO times out - to boot Linux you have to hit Tab, then enter.
- There are some very curious omissions in the bios setup, such as not being able to turn off the cute little sound it makes every time it boots (yes, you can turn the volume down, but you can't turn it off). You can't get rid of the "SONY" screen either, which hangs around far longer than necessary. Apparently, MS knows how to get rid of the bios preboot screens, because the annoying Windows preboot screen goes away after you add logo=0 to MSDOS.SYS and start/exit windows, or else the VAIO knows how to read the msdos.sys file (yikes!).
- No way in the bios to turn off pressure-sensitive mouse click emulation - this would be great if it worked reliably but the truth is it doesn't - you get a *lot* of spurious clicks, even when hardly touching the pad.
- The speakers really suck.
Um, that's about it with my gripes, the bottom line is, with a 300 MHz Pentium II processor (reported as a celeron by proc) it really kicks butt, recompiling the kernel in about 5 minutes. It has a nice, fullsize keyboard and a large bright screen (mine has 1 dead pixel, I'll have to talk to Sony about that before my warrantee expires). Now, if Sony would just do something about the warts - lose the winmodem, debug the power management, etc., I'd be a really happy camper.
(if someone from Sony is reading this, feel free to drop me a line at phillips (at) dowco (dot) com)
All you're doing is taking what would've been called 2.2.25 and calling it 2.4.0
Not at all. According to the usual procedure, the goodies currently under development in the 2.3.xx tree will make it into 2.4, not into 2.2.xx (with few exceptions). Then guess what? MORE goodies will start to appear, in 2.5.xx.
Soooooo... the bottom line is: the sooner we see 2.4, the better.
the problem is not backwards compatibility with deprecated DOM extensions (which were rarely used), but backwards compatiblity with (soon to be obselete) NS4 browsers trying to render sites that it can't.
Caveat: I haven't yet checked to see whether the following suggestions have already been implemented...
All support for all nonstandard extensions should be hived off into their own DLL/Module. Any use of such extensions should cause a warning (suppressable by the user, in the same manner as a security warning).
For marketing and pragmatic reasons it is important to support some nonstandard extensions. It is equally important to support them in a way that clearly marks them as deprecated. Purists can simply turn off support for all nonstandard extensions and deal with them by emailing the webmasters of sites using such extensions. Purists can even delete the relevant DLL/module.
it may be time to fork the tree (or be ready to) and proceed with Open Kaffe (GNU Kaffe? some other name?)
Yes, with due respect to Transvirtual, it's way to dangerous to leave the majority of the the JVM code copyrighted by a company that can so easily be attacked/controlled by openSW-unfriendly forces, be they from Redmond or anywhere else. The copyright holder decides the form of licence, and right now it's GNU, but will it stay that way after a controlling interest in Transvirtual changes hands? Even if it stays GNU, it won't do to leave control of the CVS in the hands of anything other than a not-for-profit and democratic organization. Forking the tree now is the easiest way to get what we want and need.
How about LAVA (Linux jAVA; Liberated jAVA for non-linux platforms)?