In terms of time, a SuSE 9.1 ought to be taking shape about now. Has anyone heard anything about a new release? Will SuSE's acquisition by Novell delay the next release? Will it include Ximian Desktop? If not, can we assume that a new release of XD2 will follow soon after? Sorry if asking about this offends you (it usually offends _somebody_), I'm just excited about the prospect of getting some new toys to play with, most of all a production-level 2.6 kernel as the default! I got wind that the desktop guys at my Fortune 250 company -- the people who mandate what the users run -- are thinking about taking a look at Linux, mostly because of how hard we've been hit by the latest email viruses. Obviously, I'll be recommending SuSE for such a thing, but I'd sure love to show them a new version with the speed improvements in the 2.6 kernel I keep hearing about. Regards, dk P.S. Yes, I know there was a pre-release 2.6 kernel included in 9.0. I don't want to try it. Robert Love said in his web log that one of his buddies is working on a new one that will slip into 9.0 through red carpet, but warned against using it.
Hi, On Monday 01 March 2004 13:05, David Krider wrote:
In terms of time, a SuSE 9.1 ought to be taking shape about now. Has anyone heard anything about a new release? Will SuSE's acquisition by Novell delay the next release? Will it include Ximian Desktop? If not, can we assume that a new release of XD2 will follow soon after?
On http://www.suse.com/de/company/press/press_releases/archive04/cebit.html (unfortunately only available in German) you find the information that around end of April the next release (9.1) is expected to be ready. It will be based on kernel 2.6 and KDE 3.2 (those are the only detais given).
Sorry if asking about this offends you (it usually offends _somebody_), I'm just excited about the prospect of getting some new toys to play with, most of all a production-level 2.6 kernel as the default!
I got wind that the desktop guys at my Fortune 250 company -- the people who mandate what the users run -- are thinking about taking a look at Linux, mostly because of how hard we've been hit by the latest email viruses. Obviously, I'll be recommending SuSE for such a thing, but I'd sure love to show them a new version with the speed improvements in the 2.6 kernel I keep hearing about.
If your company is considering deploying Linux on the desktop, then I'd strongly advise to go for the desktop business product (SUSE LINUX Desktop). http://www.suse.de/en/business/products/sld/index.html A version of SLD based on kernel 2.6 will be released later this year. You wouldn't/shouldn't consider a product that gives you no enterprise ready support option. The normal SUSE LINUX distribution is aiming directly at the home user and does *not* offer anything near to what an enterprise might need when it comes to support. Greetings from Bremen hartmut
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 01 March 2004 01:08 pm, Hartmut Meyer wrote:
(unfortunately only available in German) you find the information that around end of April the next release (9.1) is expected to be ready. It will be based on kernel 2.6 and KDE 3.2 (those are the only detais given).
After my very bad experience I just had backing out of an aborted 2.6 upgrade, I have to say, sometimes waiting is not all that bad. Having the latest release ensures you have most of the latest functionality, and /all/ of the latest bugs. It really depends on what you want out of Linux. I've decided compiling my own kernel is a bit too risky for the gain. I've been pretty slow at trying the 2.6 from the kernelmeister. And I think it won't be until it's on a CD that I try again.
Sorry if asking about this offends you (it usually offends _somebody_), I'm just excited about the prospect of getting some new toys to play with, most of all a production-level 2.6 kernel as the default!
Why should asking about the latest release of a company's product on the company's mailing list dedicated to that product offend anybody? I think it tends to generate interest in what's coming down the pike, and is good for SuSE's business.
If your company is considering deploying Linux on the desktop, then I'd strongly advise to go for the desktop business product (SUSE LINUX Desktop).
http://www.suse.de/en/business/products/sld/index.html
A version of SLD based on kernel 2.6 will be released later this year.
I sure hope they don't start subtracting from the Professional distribution to differentiate it form the SLD.
You wouldn't/shouldn't consider a product that gives you no enterprise ready support option. The normal SUSE LINUX distribution is aiming directly at the home user and does *not* offer anything near to what an enterprise might need when it comes to support.
As far as /technical/ support goes, that is correct. SuSE provides far more than MSXP when it comes to productivity software. And with the (proprietary) Ximian Connector it will interface with MS-Exchange more easily, which is the one place where I found problems using Open Source in a corporate network.
Greetings from Bremen hartmut
STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFAQ6duwX61+IL0QsMRAjUQAKCxr4Igk4LSzb1c5X69+0+4vRFjMQCgrzFh vGzDjLRanXWRZAv0US7iYDE= =PNeR -----END PGP SIGNATURE-----
On Mon, 2004-03-01 at 21:13, Steven T. Hatton wrote:
After my very bad experience I just had backing out of an aborted 2.6 upgrade, I have to say, sometimes waiting is not all that bad. Having the latest release ensures you have most of the latest functionality, and /all/ of the latest bugs. It really depends on what you want out of Linux. I've decided compiling my own kernel is a bit too risky for the gain. I've been pretty slow at trying the 2.6 from the kernelmeister. And I think it won't be until it's on a CD that I try again.
I found that going to a 2.6 kernel (2.6.0-test7 was my first one) was a
fair bit of work and did require booting old 2.4 kernel (home-cooked as
well) a good few times to get the initrd sorted for the 2.6 kernel. Once
that was working, 2.6 all the way up to 2.6.3 has been working fine.
2.6.0 and 2.6.1 required PreEmpt to be switched off, but since 2.6.2-rc2
something, that has been working well.
Compiling own kernel is not as hard as people make out but I will freely
admit that it isn't a walk in the park sometimes. Usually, it works
first time though. Bleeding edge people do try the -rc kernels, or -mm
or -ck or any of the plethora of additional patch sets there are, as
they might have fixes for their specific hardware or a bug they have
hit. Win some and lose some. :-)
I have in general been okay with -rc and -pre kernels, as well as -ac
and -aa patches. Only time I have ended up in a pickle was using
BitKeeper to pull down the kernel tree, clone it and add linux-sound and
2.6-acpi patches from the respective experimental trees. That bit me,
but not badly.
Happy Hacking Dudes,
--
Anders Karlsson
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 01 March 2004 05:05 pm, Anders Karlsson wrote:
On Mon, 2004-03-01 at 21:13, Steven T. Hatton wrote:
After my very bad experience I just had backing out of an aborted 2.6 upgrade, I have to say, sometimes waiting is not all that bad. Having the latest release ensures you have most of the latest functionality, and /all/ of the latest bugs. It really depends on what you want out of Linux. I've decided compiling my own kernel is a bit too risky for the gain. I've been pretty slow at trying the 2.6 from the kernelmeister. And I think it won't be until it's on a CD that I try again.
I found that going to a 2.6 kernel (2.6.0-test7 was my first one) was a fair bit of work and did require booting old 2.4 kernel (home-cooked as well) a good few times to get the initrd sorted for the 2.6 kernel. Once that was working, 2.6 all the way up to 2.6.3 has been working fine. 2.6.0 and 2.6.1 required PreEmpt to be switched off, but since 2.6.2-rc2 something, that has been working well.
Compiling own kernel is not as hard as people make out but I will freely admit that it isn't a walk in the park sometimes. Usually, it works first time though. Bleeding edge people do try the -rc kernels, or -mm or -ck or any of the plethora of additional patch sets there are, as they might have fixes for their specific hardware or a bug they have hit. Win some and lose some. :-)
I have in general been okay with -rc and -pre kernels, as well as -ac and -aa patches. Only time I have ended up in a pickle was using BitKeeper to pull down the kernel tree, clone it and add linux-sound and 2.6-acpi patches from the respective experimental trees. That bit me, but not badly.
Happy Hacking Dudes,
Oh, it's not all that "hard". The hard part is figuring out what all the options mean, and which ones you really want/need. I haven't done it for quite some time simply because I am focused on other things, and don't want to risk screwing my system up. What happened this time with the rpm was the 2.6 locked up during the first boot, and the rollback was less than pretty. Partly because I decided to learn Grub in crisis mode. I know LILO well enough to have recovered much more gracefully. The big thing that threw me was (hd0,1)==/dev/hda2. But, I ended up fixing other things that were broken, learning lots of stuff about my system, and spending lots of time thumbing through the latest Linux in a Nutshell while waiting for the patches to reload from the server. I'm glad others are still compiling their own kernel. It needs to happen. STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFAQ756wX61+IL0QsMRAg3EAJ95K3+JMA/BdfMRIZa69fbuc4nhLgCeMwsC suH8gRoZL4p0Ljnw53RcobQ= =TYSS -----END PGP SIGNATURE-----
On Monday 01 March 2004 05:51 pm, Steven T. Hatton wrote:
On Monday 01 March 2004 05:05 pm, Anders Karlsson wrote:
On Mon, 2004-03-01 at 21:13, Steven T. Hatton wrote:
After my very bad experience I just had backing out of an aborted 2.6 upgrade, I have to say, sometimes waiting is not all that bad. Having the latest release ensures you have most of the latest functionality, and /all/ of the latest bugs. It really depends on what you want out of Linux. I've decided compiling my own kernel is a bit too risky for the gain. I've been pretty slow at trying the 2.6 from the kernelmeister. And I think it won't be until it's on a CD that I try again.
I found that going to a 2.6 kernel (2.6.0-test7 was my first one) was a fair bit of work and did require booting old 2.4 kernel (home-cooked as well) a good few times to get the initrd sorted for the 2.6 kernel. Once that was working, 2.6 all the way up to 2.6.3 has been working fine. 2.6.0 and 2.6.1 required PreEmpt to be switched off, but since 2.6.2-rc2 something, that has been working well.
Compiling own kernel is not as hard as people make out but I will freely admit that it isn't a walk in the park sometimes. Usually, it works first time though. Bleeding edge people do try the -rc kernels, or -mm or -ck or any of the plethora of additional patch sets there are, as they might have fixes for their specific hardware or a bug they have hit. Win some and lose some. :-)
I have in general been okay with -rc and -pre kernels, as well as -ac and -aa patches. Only time I have ended up in a pickle was using BitKeeper to pull down the kernel tree, clone it and add linux-sound and 2.6-acpi patches from the respective experimental trees. That bit me, but not badly.
Happy Hacking Dudes,
Oh, it's not all that "hard". The hard part is figuring out what all the options mean, and which ones you really want/need. I haven't done it for quite some time simply because I am focused on other things, and don't want to risk screwing my system up. What happened this time with the rpm was the 2.6 locked up during the first boot, and the rollback was less than pretty. Partly because I decided to learn Grub in crisis mode. I know LILO well enough to have recovered much more gracefully. The big thing that threw me was (hd0,1)==/dev/hda2. But, I ended up fixing other things that were broken, learning lots of stuff about my system, and spending lots of time thumbing through the latest Linux in a Nutshell while waiting for the patches to reload from the server.
I'm glad others are still compiling their own kernel. It needs to happen.
Yes it does.... and in my view, it is far less problematic than trying to use the RPM's. Once you learn how to compile a kernel and install it side-by-side with *any* other kernel, you are home free. And it really isn't that difficult. I would recommend it to anyone....
STH
-- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/01/04 17:54 + +----------------------------------------------------------------------------+ "Interchangeable parts won't."
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 01 March 2004 05:55 pm, Bruce Marshall wrote:
On Monday 01 March 2004 05:51 pm, Steven T. Hatton wrote:
I'm glad others are still compiling their own kernel. It needs to happen.
Yes it does.... and in my view, it is far less problematic than trying to use the RPM's. Once you learn how to compile a kernel and install it side-by-side with *any* other kernel, you are home free. And it really isn't that difficult. I would recommend it to anyone....
STH
Don't bet too much on that. I know SuSE tend to back-port important features which you might not get from the source. I guess as long as you leave the SuSE kernel in place, you probably don't need to worry about rpm dependencies. I'm not sure exactly what I'd have to change these days to point to the correct boot image and initrd, but I'm loath to change things that SuSEconfig has under management, unless I use YaST to do it. If you do touch those files, you better pay close attention when applying patches, etc. SuSEconfig won't fix them up for you as it would its own work.
-- +-------------------------------------------------------------------------- --+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/01/04
STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFAQ+nZwX61+IL0QsMRAgm0AJ9EW6irk1Fvn0cr8Nr+I/obI0qxxgCcC/VD LwBjnt7Xrta6QyRGIvnc8Qo= =53mb -----END PGP SIGNATURE-----
On Mon, 2004-03-01 at 20:56, Steven T. Hatton wrote:
On Monday 01 March 2004 05:55 pm, Bruce Marshall wrote:
Once you learn how to compile a kernel and install it side-by-side with *any* other kernel, you are home free. And it really isn't that difficult. I would recommend it to anyone....
I'm not sure exactly what I'd have to change these days to point to the correct boot image and initrd, but I'm loath to change things that SuSEconfig has under management, unless I use YaST to do it. If you do touch those files, you better pay close attention when applying patches, etc. SuSEconfig won't fix them up for you as it would its own work.
I've built a few kernels over the years, and I've come to the conclusion that I, personally, would rather wait for new kernels to show up as the default in new distributions. While I'm comfortable putting my own kernel into place beside my other(s), what bugs me about using a 2.6 kernel these days is that no one seems to use the garden-variety sources. Everyone seems to add this or that patch. In addition, while researching how to fix the recent 192 kernel's breakage of VMware, I accidently ran across a post on kerneltrap.org (I think) that talked about getting VMware (and the binary Nvidia driver) working under 2.6. That convinced me that the Linux world in general isn't ready for this, especially not me. Shoot, it's bad enough keeping VMware and the Nvidia driver working on production kernels shipped by one of the "Big 3!" dk P.S. "Big 3" refers to Red Hat, SuSE, and Debian, in my mind. But Gentoo keeps making strides. If they'd just make a binary reference platform for each (popular) optimization level so that people wouldn't have to recompile the whole thing if they didn't want to, I think they'd garner a lot more attention.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Tuesday 02 March 2004 08:30 am, David Krider wrote:
On Mon, 2004-03-01 at 20:56, Steven T. Hatton wrote: [snip] I've built a few kernels over the years, and I've come to the conclusion that I, personally, would rather wait for new kernels to show up as the default in new distributions. While I'm comfortable putting my own kernel into place beside my other(s), what bugs me about using a 2.6 kernel these days is that no one seems to use the garden-variety sources. Everyone seems to add this or that patch.
Well, I have a hard time taking my own advice. See below.
In addition, while researching how to fix the recent 192 kernel's breakage of VMware, I accidently ran across a post on kerneltrap.org (I think) that talked about getting VMware (and the binary Nvidia driver) working under 2.6. That convinced me that the Linux world in general isn't ready for this, especially not me. Shoot, it's bad enough keeping VMware and the Nvidia driver working on production kernels shipped by one of the "Big 3!"
I'm sorry. The whole notion of VMware seems somehow perverse to me. I guess it works for people, but I expect it would be more problem than solution. The face that people have two OS's running on the same chip at the same time amazes me orders of magnitude more than the KDE running on XP.
dk
P.S. "Big 3" refers to Red Hat, SuSE, and Debian, in my mind. But Gentoo keeps making strides.
It really is vertigo inducing to observe the rate of change in Open Source. I'm not really joking when I say released software is obsolete. The stuff at xml.apache.org seems that way. They always have the features I really want in the cooker, rather than on the plate.
If they'd just make a binary reference platform for each (popular) optimization level so that people wouldn't have to recompile the whole thing if they didn't want to, I think they'd garner a lot more attention.
I'm not really sure that's their goal. And perhaps that's good. All philosophical musings asside, I have been at this all night. I've got something built, and it only took one reconfigure to get there. I haven't booted it, for a couple reasons. The make output is telling me it's looking for an unknown symbol. I suspect that is a bad thing. Another reason is because I really don't know what the heck I have. I got two file from the Kernelmeister: inux-2.6.3.SUSE.tar.bz2 and suse-2.6.3-0.bz2 The unpack as the full body of the Linux Kernel source at some point in time, and what appears to be an 18 Meg patch file. compressed down to 2.9 Meg mind you. I started to apply that patch, and I got this: hattons@ljosalfr:/download/com/suse/kernel-source/ Tue Mar 02 09:09:19:> cd linux-2.6.3.SUSE/ hattons@ljosalfr:/download/com/suse/kernel-source/linux-2.6.3.SUSE/ Tue Mar 02 09:09:28:> patch -i ../suse-2.6.3-0 patching file lirc_it87 patching file scsi-changer.txt patching file MAINTAINERS Reversed (or previously applied) patch detected! Assume -R? [n] That suggests that the patch has already been applied. Or that the patch is not a straight diff between the starting point, and the SuSEfied version. I believe I can throw a switch and force all the patches in the patch file to be applied. That doesn't seem like the right thing to do. Has anybody figured out what the heck he intends others to do with these? STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFARJmawX61+IL0QsMRAlOZAJ0azSDuReeb1jG46vJ3naBtl5XqSwCgsSDz cwE17U0bBVPr045bPZtwMD4= =5eYe -----END PGP SIGNATURE-----
On Tuesday 02 March 2004 15.26, Steven T. Hatton wrote:
I got two file from the Kernelmeister: inux-2.6.3.SUSE.tar.bz2 and suse-2.6.3-0.bz2
I don't know where you found those or who the "kernelmeister" is supposed to be, but the latest version of the suse-ified 2.6.3 is -16 and can be found in people/kraxel/i386/i586/ Also, the tag "suse" in the filename seems to suggest that it is a prepatched version of the kernel you have there. The patches would likely apply to linux-2.6.3 as retreived from kernel.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Tuesday 02 March 2004 09:32 am, Anders Johansson wrote:
On Tuesday 02 March 2004 15.26, Steven T. Hatton wrote:
I got two file from the Kernelmeister: inux-2.6.3.SUSE.tar.bz2 and suse-2.6.3-0.bz2
I don't know where you found those or who the "kernelmeister" is supposed to be, but the latest version of the suse-ified 2.6.3 is -16 and can be found in people/kraxel/i386/i586/
Also, the tag "suse" in the filename seems to suggest that it is a prepatched version of the kernel you have there. The patches would likely apply to linux-2.6.3 as retreived from kernel.org
Yup, Kraxel's are newer. Guess Mantel's not *the* Kerenelmeister anymore? ;-) I'm really just being silly. I know Mantel has been hacking the Kernel for years. I saw Kraxel's (I'm not sure how to untangle that) name on a lot of the patches as well as Mantel's. I had just been conditioned to go Mantel's files for the latest Kernels. I will also observe not all mirrors are up to date. Which is a constant frustration. Thanks for the pointer. I suspect the -16 rpm's will slip right into the work I've already done. The hard part is reading all that stuff, and making sure the 100% you have left matches the hardware on your system, and has all the prerequisites for your modules. STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFARLAcwX61+IL0QsMRAvoiAJ9nFlbl0qob8mjjb89XZsZZvmOYUACgl3tb 1prnCNJgiJ5U0a2AcLj8aiY= =fRLQ -----END PGP SIGNATURE-----
On Tuesday 02 March 2004 09:26 am, Steven T. Hatton wrote:
I'm sorry. The whole notion of VMware seems somehow perverse to me. I guess it works for people, but I expect it would be more problem than solution. The face that people have two OS's running on the same chip at the same time amazes me orders of magnitude more than the KDE running on XP
Better get used to the idea... I've been running virtual machines and OS's since 1969. I think it's here to stay..... -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04 09:36 + +----------------------------------------------------------------------------+ "When you're sending someone Styrofoam, what do you pack it in?"
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Tuesday 02 March 2004 09:37 am, Bruce Marshall wrote:
On Tuesday 02 March 2004 09:26 am, Steven T. Hatton wrote:
I'm sorry. The whole notion of VMware seems somehow perverse to me. I guess it works for people, but I expect it would be more problem than solution. The face that people have two OS's running on the same chip at the same time amazes me orders of magnitude more than the KDE running on XP
Better get used to the idea... I've been running virtual machines and OS's since 1969. I think it's here to stay.....
Is it really a virtual machine architecture? I guess that's why they call it _V_M_ ware, eh? I assume VMware sits on top of the chip, and intercepts the calls from the two OS's and parcels out the resources as it sees fit. If everybody involved were playing nicely together, it wouldn't bother me. It's just that Bill always seem to be trying to break such things. About a year back, Sun filed suit for a $Billion U.S. because XP broke Java. That /is/ 10.00^9. How does it deal with network connections, and other potentially competing resources?
--+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04
STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFARLKVwX61+IL0QsMRAo/qAJ4+6A0IjsnzCSTmAH5+TYgSmCNBagCgjrQD Hel0KFJFqz3R14bC2XaoAo4= =Jm2Y -----END PGP SIGNATURE-----
On Tuesday 02 March 2004 11:12 am, Steven T. Hatton wrote:
Is it really a virtual machine architecture? I guess that's why they call it _V_M_ ware, eh? I assume VMware sits on top of the chip, and intercepts the calls from the two OS's and parcels out the resources as it sees fit. If everybody involved were playing nicely together, it wouldn't bother me. It's just that Bill always seem to be trying to break such things.
About a year back, Sun filed suit for a $Billion U.S. because XP broke Java. That /is/ 10.00^9.
How does it deal with network connections, and other potentially competing resources?
I think you're making it much more complicated than it really is. Many hardware architectures provide for a virtual mode, where the memory is mapped/translated so that it appears to the VM that it has been loaded at address 0. The host (linux in this case) pages the virtual memory and the hardware does the address translations. So far so good... and the only thing (a big generalization) that needs to be added is the ability to do 'low level' stuff such as disk IO, network stuff, etc. This is where VMware comes in. Whenever Win98 (for example) tries to make a call to do hardware types of things, i.e. things that a normal user program wouldn't be allowed to do, an interrupt is created and that interrupt goes to VMware which can simulate the disk IO as an example. Thus, Win98 can be writing to a reiserfs disk file which in fact, it knows nothing about reiserfs. VMware handles all the disk calls. Same for networking and other calls. But yes, it is truely a virtual architecture which can be seen when installing windows into VMware. You have to do all of the normal things you would do on the real hardware... make partitions, format them, boot from a floppy, etc. -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04 11:17 + +----------------------------------------------------------------------------+ "Neuroses are red, Melancholia's blue, I'm schizophrenic, What are you?"
On Tuesday 02 March 2004 17.22, Bruce Marshall wrote:
Many hardware architectures provide for a virtual mode, where the memory is mapped/translated so that it appears to the VM that it has been loaded at address 0. The host (linux in this case) pages the virtual memory and the hardware does the address translations.
Yeah, that's how the memory is handled for all user space processes
But yes, it is truely a virtual architecture
As I understand it it isn't fully virtual. For performance, they don't virtualise the x86 instruction set like for instance bochs does, it runs on the actual hardware, which is why you don't get VMware versions for non-x86 platforms
On Tuesday 02 March 2004 11:27 am, Anders Johansson wrote:
As I understand it it isn't fully virtual. For performance, they don't virtualise the x86 instruction set like for instance bochs does, it runs on the actual hardware, which is why you don't get VMware versions for non-x86 platforms
You are right.... but I thought my explanation made that clear. A true emulator of say a Sparc machine running on an X86 would really be slow. Essentially Win98 running under VMware runs like any other program, using the X86 hardware to do its instructions until it needs to do something that normally wouldn't be allowed (an OS call for IO for example). Then the VMware routines that surround the Win98 'space' take over to do the service needed for the call. -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04 12:26 + +----------------------------------------------------------------------------+ "Advice is seldom welcome; and those who want it the most always like it the least." - Earl of Chesterfield
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Tuesday 02 March 2004 11:22 am, Bruce Marshall wrote:
On Tuesday 02 March 2004 11:12 am, Steven T. Hatton wrote:
How does it deal with network connections, and other potentially competing resources?
I think you're making it much more complicated than it really is.
Many hardware architectures provide for a virtual mode, where the memory is mapped/translated so that it appears to the VM that it has been loaded at address 0. The host (linux in this case) pages the virtual memory and the hardware does the address translations.
I'm used to that. Mathematica is something of its own OS, and was more so under WinDOS. As long as VMWare (Linux?) intercepts all the system calls from NT, and responds in such a way NT thinks it's a CPU, then there's not a problem. That doesn't sound all that trivial to me. You have to emulate a bit more than just virtual memory. What happens to Windows drivers if Linux is actually the OS controlling the hardware? Drivers get out there on the busses an start manipulating the register sets on the hardware, or writing to the video ram. Does VM Ware provide pseudo drivers that are installed in the windows configuration?
So far so good... and the only thing (a big generalization) that needs to be added is the ability to do 'low level' stuff such as disk IO, network stuff, etc. This is where VMware comes in.
Whenever Win98 (for example) tries to make a call to do hardware types of things, i.e. things that a normal user program wouldn't be allowed to do, an interrupt is created and that interrupt goes to VMware which can simulate the disk IO as an example. Thus, Win98 can be writing to a reiserfs disk file which in fact, it knows nothing about reiserfs. VMware handles all the disk calls.
Does it support NT based OS's? If it's just WinDOS, that's can be done with a bit more than an 8088 emulation.
Same for networking and other calls.
But yes, it is truely a virtual architecture which can be seen when installing windows into VMware. You have to do all of the normal things you would do on the real hardware... make partitions, format them, boot from a floppy, etc.
What happens to NTFS? I guess that's not really a problem if you start out with VMWare, but what if your hardrive is already NTFS, full and huge?
--+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04
STH -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQFARMHZwX61+IL0QsMRAlOtAKDrIcNTYvNw+h9H1MqN4Io9d3iy4gCgpgEs q/lu6u5av2nRn1sshdgC6wM= =Dgkp -----END PGP SIGNATURE-----
On Tuesday 02 March 2004 12:18 pm, Steven T. Hatton wrote:
I'm used to that. Mathematica is something of its own OS, and was more so under WinDOS. As long as VMWare (Linux?) intercepts all the system calls from NT, and responds in such a way NT thinks it's a CPU, then there's not a problem. That doesn't sound all that trivial to me. You have to emulate a bit more than just virtual memory. What happens to Windows drivers if Linux is actually the OS controlling the hardware? Drivers get out there on the busses an start manipulating the register sets on the hardware, or writing to the video ram. Does VM Ware provide pseudo drivers that are installed in the windows configuration?
I'm not all that sure how they let Win98 handle things, but essentially Win98 (for example) is given an IDE disk, and maybe an IDE CDrom if you want, and a floppy drive if you want, and now days you can also have USB devices. All pretty simple stuff which is supported by the VMware routines. The use of REAL hardware is pretty restricted. For instance, if you want to have a SCSI drive, it must be with a particular SCSI controller.
So far so good... and the only thing (a big generalization) that needs
Does it support NT based OS's? If it's just WinDOS, that's can be done with a bit more than an 8088 emulation.
I believe it does. And Linux running under Linux too.
What happens to NTFS? I guess that's not really a problem if you start out with VMWare, but what if your hardrive is already NTFS, full and huge?
In general, the use of RAW harddrives is limited. Not sure if a real NTFS drive is possible... but you can have an NTFS drive in your WINXp machine, and it would still be simulated onto a linux drive. You supposedly *can* use a Win98 partition and boot it... but again, such use is pretty restricted and has never been necessary in my case. For what little windows most people need, they build a simple version of Winxx and run a few applications on it. -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04 12:31 + +----------------------------------------------------------------------------+ Ruckert's Law: "There is nothing so small that it can't be blown out of proportion."
On Monday 01 March 2004 08:56 pm, Steven T. Hatton wrote:
Don't bet too much on that. I know SuSE tend to back-port important features which you might not get from the source. I guess as long as you leave the SuSE kernel in place, you probably don't need to worry about rpm dependencies. I'm not sure exactly what I'd have to change these days to point to the correct boot image and initrd, but I'm loath to change things that SuSEconfig has under management, unless I use YaST to do it. If you do touch those files, you better pay close attention when applying patches, etc. SuSEconfig won't fix them up for you as it would its own work.
--
Some valid points there but.... I think an off-hand kernel from SuSE is going to also miss some things that people may need... note the problems with vmware and some of the newer SuSE kernels. It's a mixed bag but I'd rather go the vanilla route. There's never been anything I know of that I needed and the vanilla kernel didn't have (but SUSE did). -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04 08:59 + +----------------------------------------------------------------------------+ "It seems like the less a statesman amounts to, the more he loves the flag."
* Bruce Marshall
It's a mixed bag but I'd rather go the vanilla route. There's never been anything I know of that I needed and the vanilla kernel didn't have (but SUSE did).
I guess you've neevr tried to install on recent laptops (or for that matter most recent hardware ). Examples include the IDE controller and network card in recent dell laptops (m60/d800) Kind regards, -- Gerhard den Hollander Phone :+31-10.280.1515 ICT manager Direct:+31-10.280.1539 Fugro-Jason Fax :+31-10.280.1511 gdenhollander@Fugro-Jason.com POBox 1573 visit us at http://www.Fugro-Jason.com 3000 BN Rotterdam JASON.......#1 in Reservoir Characterization The Netherlands This e-mail and any files transmitted with it are confidential and intended solely for the use of the addressee. This e-mail shall not be deemed binding unless confirmed in writing. If you have received it by mistake, please let us know by e-mail reply and delete it from your system; you may not copy this message or disclose its contents to anyone. Please note that any views or opinions presented in this e-mail are solely those of the author and do not necessarily represent those of the company. E-mail transmission cannot be guaranteed to be secure or error-free. The sender therefore does not accept liability for any errors or omissions in the contents of this message, which arise as a result of e-mail transmission.
On Tuesday 02 March 2004 09:34 am, Gerhard den Hollander wrote:
* Bruce Marshall
(Tue, Mar 02, 2004 at 09:01:29AM -0500) It's a mixed bag but I'd rather go the vanilla route. There's never been anything I know of that I needed and the vanilla kernel didn't have (but SUSE did).
I guess you've neevr tried to install on recent laptops (or for that matter most recent hardware ).
I've done both, many times, including Dells. Don't know what your point is.
Examples include the IDE controller and network card in recent dell laptops (m60/d800)
If the hardware isn't supported, it doesn't much matter what kernel you use. -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/02/04 09:45 + +----------------------------------------------------------------------------+ "I've learned- that it takes years to build up trust, and only suspicion, not proof, to destroy it."
Bruce Marshall wrote:
Yes it does.... and in my view, it is far less problematic than trying to use the RPM's. Once you learn how to compile a kernel and install it side-by-side with *any* other kernel, you are home free. And it really isn't that difficult. I would recommend it to anyone....
Do I understand correctly that it is possible to install two different kernels side by side by compiling it but that it is not possible to do this by using the RPM's? Patrick
On Thursday 11 March 2004 05:18 pm, Patriiiiiiiiiick wrote:
Bruce Marshall wrote:
Yes it does.... and in my view, it is far less problematic than trying to use the RPM's. Once you learn how to compile a kernel and install it side-by-side with *any* other kernel, you are home free. And it really isn't that difficult. I would recommend it to anyone....
Do I understand correctly that it is possible to install two different kernels side by side by compiling it but that it is not possible to do this by using the RPM's?
Patrick
Since I've never used the RPM's (except for initial install) I can't tell you about putting two or more of them on, but with compiled kernels, you can have as many as you have the disk space for.... -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 03/11/04 17:44 + +----------------------------------------------------------------------------+ "The important thing is not to stop questioning."
Quoting Patriiiiiiiiiick
Bruce Marshall wrote:
Yes it does.... and in my view, it is far less problematic than trying to use the RPM's. Once you learn how to compile a kernel and install it side-by-side with *any* other kernel, you are home free. And it really isn't that difficult. I would recommend it to anyone....
Do I understand correctly that it is possible to install two different kernels side by side by compiling it but that it is not possible to do this by using the RPM's?
Patrick
It is possible, if the kernels are named differently. Install the 2nd kernel, don't upgrade it (rpm -U). The installation will probably change the vmlinuz and initrd symlinks to point to the new kernel, but you can change them back. There can be problems if they require different /etc/modules.conf. HTH, Jeffrey
participants (9)
-
Anders Johansson
-
Anders Karlsson
-
Bruce Marshall
-
David Krider
-
Gerhard den Hollander
-
Hartmut Meyer
-
Jeffrey L. Taylor
-
Patriiiiiiiiiick
-
Steven T. Hatton