I bought the family a new car yesterday. A slightly old (born 96) Volkswagen Caravelle TDI, the looong version. In that car, 8 persons can have a comfy ride, together with a lot of stuff.
So that's what occupied me the last few days. And the day after tomorrow, I'll travel to Madrid. Check out my photoblog, pictures of the new car, and from Madrid will appear!
Monday, April 24, 2006
Friday, March 31, 2006
Dr. Scient
Yesterday, I defended my thesis, and everything went well. So, finally, after 9 3/4 years, I have finished my Dr.Scient studies, and am now a dr. in scientific computing.
Well done, Mandus.
Well done, Mandus.
Wednesday, March 29, 2006
The trial lecture and the defense
Maybe some of you wonder what is going on - I promised to report on the progress of the preparations for my trial lecture, and said a few words. After that complete silence...
First, I have worked very hard the last 2 weeks - and tomorrow is the big day. And I think I have the presentations ready by now, although I have to test them a few times to check that I can stick to the schedule.
Next, I have to say that I didn't came up with extremely much interesting content. I tried to investigate the "formal verification" field to see whether there was something there that could be of any use, and find quite some material to read. Unfortunately, most of what I found is still to preliminary to be of any real value for practical verification of scientific codes. So I ended up with the standard pile of tricks, method of manufactured solutions beeing the most prominent, or the most advanced.
The presentation of my thesis is the toughest one, though, as I have to present what I am been doing the last couple of years (I started in -96...) in about 30 minutes. Though call, but I'll manage.
Take care, and wish me good luck tomorrow!
First, I have worked very hard the last 2 weeks - and tomorrow is the big day. And I think I have the presentations ready by now, although I have to test them a few times to check that I can stick to the schedule.
Next, I have to say that I didn't came up with extremely much interesting content. I tried to investigate the "formal verification" field to see whether there was something there that could be of any use, and find quite some material to read. Unfortunately, most of what I found is still to preliminary to be of any real value for practical verification of scientific codes. So I ended up with the standard pile of tricks, method of manufactured solutions beeing the most prominent, or the most advanced.
The presentation of my thesis is the toughest one, though, as I have to present what I am been doing the last couple of years (I started in -96...) in about 30 minutes. Though call, but I'll manage.
Take care, and wish me good luck tomorrow!
Saturday, March 18, 2006
Preparation for the trial lecture
Maybe just for my own future reference... I will now briefly document the preparations for my trial lecture and defense of my Dr.Scient thesis.
So far I have worked on a "mindmap" using kdissert to organize the different aspects of the topic for the trial lecture that come to my mind. I have also search google and other places for relevant information. During thursday and friday I discovered at least 2 books and about 20 papers that seems to be relevant. I am now reading a paper by Salari and Knupp which are very relevant, as it discuss code verification by the method of manufactured solutions, which is a common approach in computational science and engineering.
As I move forward I also want to investigate to some extent the field of more formal code verification in computer science. From what I know from before, it is not possible to prove correctness of a scientific code (in the multi-million lines of code range) using such formal theories, as usually it is required that some special programming language is used, as well as the complexity can not be infinitely large. I am also not sure whether or not these to fields are talking about the same thing, as the interest in comp. sci & eng. is to verify that the output of the code is correct, and does not care that much about how it become correct (ok, I do simplify things here...), while traditional CS probably are more concerned about the correctness of the code itself.
Another topic is how we can relate verification of scientific codes to verification of experiments in other natural sciences, like physics or biology. I have to look further into this as well.
Now on to the reading to see what more I can discover!
So far I have worked on a "mindmap" using kdissert to organize the different aspects of the topic for the trial lecture that come to my mind. I have also search google and other places for relevant information. During thursday and friday I discovered at least 2 books and about 20 papers that seems to be relevant. I am now reading a paper by Salari and Knupp which are very relevant, as it discuss code verification by the method of manufactured solutions, which is a common approach in computational science and engineering.
As I move forward I also want to investigate to some extent the field of more formal code verification in computer science. From what I know from before, it is not possible to prove correctness of a scientific code (in the multi-million lines of code range) using such formal theories, as usually it is required that some special programming language is used, as well as the complexity can not be infinitely large. I am also not sure whether or not these to fields are talking about the same thing, as the interest in comp. sci & eng. is to verify that the output of the code is correct, and does not care that much about how it become correct (ok, I do simplify things here...), while traditional CS probably are more concerned about the correctness of the code itself.
Another topic is how we can relate verification of scientific codes to verification of experiments in other natural sciences, like physics or biology. I have to look further into this as well.
Now on to the reading to see what more I can discover!
Thursday, March 16, 2006
The topic for my trial lecture
Today, I have received the topic for my trial lecture. The title of the lecture is:
Verification of Scientific Codes
Verification of Scientific Codes
Not a bad topic at all. Alltough this is something we don't think about every day, it sure is an important part of what I do. If we can't say anything about the correctness of the programs that we write inorder to say something about scientific problems, how can we then use the program to draw any conclusions for the problem that we investigate.
Also, a few years back we discussed this topic within the group, and one of my coworkers even produced some software that target some of the issues within this field. So I have someone to ask, and do have some ideas my self as well.
A trial lecture is never a walk in the park, but at least this is something that I can relate to, so I think it is an ok topic.
Now, on to diggin' out references, reading material and start producing slides for the talk. You wan't see much of me the next two weeks...
Take care folks,
Also, a few years back we discussed this topic within the group, and one of my coworkers even produced some software that target some of the issues within this field. So I have someone to ask, and do have some ideas my self as well.
A trial lecture is never a walk in the park, but at least this is something that I can relate to, so I think it is an ok topic.
Now, on to diggin' out references, reading material and start producing slides for the talk. You wan't see much of me the next two weeks...
Take care folks,
Wednesday, March 08, 2006
Samsung Q1 - infoSync World - origami
Samsung Q1 - infoSync World
Is it just me, or isen't this very much like the previous U50/70/71 models from Sony? I thought we should expect something new. Ok, you know that I'm not a big Microsoft fan, but for all the hype I at least expected something actually new...
By the way, why can't any pda or small laptop maker consider making something with a formfactor comparable with the Psion models - a decent keyboard, ok display, good batterylife, and software and communication hardware (bluetooth/umts/wifi) up to todays standards. That will be a killer piece of hardware.
In other news, I am at the winterschool at Geilo, on Navier stokes and automation of PDE/FEM software (http://www.fenics.org), this week. Cool stuff!
Is it just me, or isen't this very much like the previous U50/70/71 models from Sony? I thought we should expect something new. Ok, you know that I'm not a big Microsoft fan, but for all the hype I at least expected something actually new...
By the way, why can't any pda or small laptop maker consider making something with a formfactor comparable with the Psion models - a decent keyboard, ok display, good batterylife, and software and communication hardware (bluetooth/umts/wifi) up to todays standards. That will be a killer piece of hardware.
In other news, I am at the winterschool at Geilo, on Navier stokes and automation of PDE/FEM software (http://www.fenics.org), this week. Cool stuff!
Wednesday, March 01, 2006
My Ph.D thesis hot off the press
Not exactly true, but I have paid unipub a visit today, to hand over the press ready pdf-version of my thesis.
Now I only have to prepare a couple of presentation, and give them, as well as answering a few questions from the comitee, before I have earned the right to call my self a Dr. It's been a long and interesting journey, and I'm glad I can see the end.
The important date to remember is 30. March.
Take care folks!
Now I only have to prepare a couple of presentation, and give them, as well as answering a few questions from the comitee, before I have earned the right to call my self a Dr. It's been a long and interesting journey, and I'm glad I can see the end.
The important date to remember is 30. March.
Take care folks!
Tuesday, February 28, 2006
Monday, February 27, 2006
Ekspertene slakter Norge - Aftenposten.no
Ekspertene slakter Norge - Aftenposten.no
Hallo, liksom. Norge er et lite, bitte lite, land. Vi kan ikke forvente at vi skal ta en haug med gull hver gang det er vinter-ol.
Jeg synes vi har gjort en grei innsats jeg, og de som har flydd på ski og hoppa og kosta, de har gjort en hederlig innsats.
Dessuten har jeg fått vite at min dr.grads avhandlig er funnet "worthy of defending for the dr. scient degree". Så nå er det nesten bare tut og kjør igjen...
Take care,
Hallo, liksom. Norge er et lite, bitte lite, land. Vi kan ikke forvente at vi skal ta en haug med gull hver gang det er vinter-ol.
Jeg synes vi har gjort en grei innsats jeg, og de som har flydd på ski og hoppa og kosta, de har gjort en hederlig innsats.
Dessuten har jeg fått vite at min dr.grads avhandlig er funnet "worthy of defending for the dr. scient degree". Så nå er det nesten bare tut og kjør igjen...
Take care,
Tuesday, February 14, 2006
digi.no - Rapport hevder Linux-servere er billigst
digi.no - Rapport hevder Linux-servere er billigst
Som man roper i skogen får man svar vil nå jeg si. Som ihuga linux tilhenger er jeg seff enig, men det er nå en gang sånn at Linux og Windows er to helt forskjellige systemer, så jeg skjønner ikke hvorfor man på død og liv skal sammenlikne.
Dersom du har sys.admins som kan både Windows og Linux, vil som regel Linux bli valgt på serversiden dersom applikasjonen man skal implementere tillater valg. Grunnen er opplagt - fra sys.admin siden er som regel Linux et langt mer behagelig valg.
Dersom du har sys.admins som bare kan Windows har du ikke noe valg. Bortsett fra å sparke staben og ansette nye da. Og dersom de bare kan Linux har du heller ikke noe valg.
På desktop siden er systemene såpass forskjellige, at dersom ikke ledelse og sys.admins ønsker å overkjøre sine brukere, og dermed få lite produktive, fiendlige innstilte medarbeidere, så har man heller ikke noe valg i særlig grad. Da snakker jeg om avanserte, oppgående brukere selvsagt - kveget får bruke det de får. Men de andre må altså få velge, og da vil hver enkelt bruker velge det saom passer best for seg, enten det er Linux, Mac OSX, Windows, eller et av de andre OS som finnes.
Omtrent sånn ser nå jeg det.
Som man roper i skogen får man svar vil nå jeg si. Som ihuga linux tilhenger er jeg seff enig, men det er nå en gang sånn at Linux og Windows er to helt forskjellige systemer, så jeg skjønner ikke hvorfor man på død og liv skal sammenlikne.
Dersom du har sys.admins som kan både Windows og Linux, vil som regel Linux bli valgt på serversiden dersom applikasjonen man skal implementere tillater valg. Grunnen er opplagt - fra sys.admin siden er som regel Linux et langt mer behagelig valg.
Dersom du har sys.admins som bare kan Windows har du ikke noe valg. Bortsett fra å sparke staben og ansette nye da. Og dersom de bare kan Linux har du heller ikke noe valg.
På desktop siden er systemene såpass forskjellige, at dersom ikke ledelse og sys.admins ønsker å overkjøre sine brukere, og dermed få lite produktive, fiendlige innstilte medarbeidere, så har man heller ikke noe valg i særlig grad. Da snakker jeg om avanserte, oppgående brukere selvsagt - kveget får bruke det de får. Men de andre må altså få velge, og da vil hver enkelt bruker velge det saom passer best for seg, enten det er Linux, Mac OSX, Windows, eller et av de andre OS som finnes.
Omtrent sånn ser nå jeg det.
Gratulerer Sverige!!!
Som noen vet pågår det et OL for tiden. I dag var det sprint-stafett, en langrennsøvelse hvor ihvertfall de fleste normenn nok hadde forventet norske gullmedaljer. Men idag vant Sverige, både herre og dame stafetten. Derfor vil jeg gjerne få lov til å gratulere våre gode naboer på det hjerteligste!
Wednesday, February 08, 2006
I will soon be a doctor!
Just a short notice to inform my faithful crowd of readers that I soon will be a Dr.Scient. All the required documents are now accepted. The only thing left is that the committee must accept my thesis, and then the defense of the thesis, of course. But that should be walk in the park. not...
Anyway, wish me good luck on the last mile.
And as always, take care out there. There's lot of snow, and the roads are slippy. At least in my area in Norway. Just have a look out of my window:
Tuesday, February 07, 2006
A silly post. Mostly.
Ok, sorry folks, this post isn't very interesting. I am testing out a program called gnome-blog, running on linux, which can be used for posting directly to the blog from my desktop. As this is a gtk/gnome program, as should be evident from the name, it is not optimal for me. The optimal solution will be something I can run directly from the shell, or maybe even from a macro in the editor vim, but I will just test this program for now.
By the way, I have to report on an extraordinary service from our regular hardware-pusher, pc-express. As reported yesterday, the NIC on my desktop computer at work broke down. As the manufacturer in question, Shuttle, are known by our dealer to not exactly handle guarantee matters in a prompt fashion, I was told today that the just ordered a new box for me, and will handle the return of the old one without bothering me. In effect, I will get a computer back very soon, which is important to me. I think this is more than I could expect, so I am really satisfied.
PE2 2:10-11: But chiefly them that walk after the flesh in the lust of uncleanness, and despise government. Presumptuous are they, selfwilled, they are not afraid to speak evil of dignities.
Whereas angels, which are greater in power and might, bring not railing accusation against them before the Lord.
Take care,
Monday, February 06, 2006
Me and my computers
The regular reader knows that I have right now 3 computers that I use regulary. With two of them there has been some events recently that I can mention here, and for the last one, the P1510, I have promised you some comments on the linux installation.
First, the AMD64 computer that I regularly use at work. The Ubuntu for AMD64 is working flawlessly on that computer. Everything that I need is running just fine. But on friday the NIC fell apart. Usually, that is not a big deal, you just pull the card out and slip a new one in to the box. Problem is that this is a Shuttle SN25p, with the excelent NIC built right into the mainboard. Combined with noe free PCI slot on the mainboard, I do not have any other option but sending the whole box in for service. So, I am happy with the box, it works quite well, and performance is more than adequate for my work (which includes some really heavy simulations of partial differential equations), but no free slot for a emergency NIC can easily be a showstopper in the future. I am thinking on finding other use for the computer within our lab, and get another one for myself.
The PowerMac at home is also doing fine these days. As you know, I ended up installing gentoo on that computer, because the installers of both debian and ubuntu for powerpc failed to work, due to problem with disk partitioning. The installer is just unable to detect the geometry of my partitions, and give me only the option to wipe out the entire disk, including my MacOSX installation, and put in linux. Not what I want, as I use MacOSX for different purposes.
The problem was that I never used the gentoo installed. It just take to much time to compile everything, and a whole lot of packages is masked out for powerpc64, as I guess not very many gentoo users play with these animals. But then I figured out that I already have a working disk geometry and working yaboot for booting the operating system I want, so I just tried to install ubuntu for powerpc64 into or on top of the gentoo installation. For this purpose I chosed the pre-release of the next ubuntu (Dapper Drake), and run the installer in expert modus. In stead of working through the partitioning and mounting step in the installer, I switched to another console, mounted what I wanted as '/target', run the necessary swapon command, and wiped out the original gentoo stuff from /target. Most of the installer worked just fine from that point, except making the installation bootable and writing a correct fstab. The fstab I could easily handle by hand, and the configuration for the 'yaboot' boot-loader I have been clever enough to back up to a USB drive. So, after copying in the yaboot config, running 'ybin -v' and doing some editing of fstab using vim, I have a working ubuntu. Now with most of what I need available in binary form. There are some issues with the airport extreme wireless card, my firewire soundcard and the built in soundcard, but I can live with this.
With the Fujitsu P1510, the history is not very interesting. I booted from Debian sarge net-install cd, performed a regular installation, upgraded to etch. Everything is just fine. Bluetooth, sound, WiFi, Xorg - everything works. There are only a few minior things. Suspend/resume is not working out of the box, so I patched the kernel with software suspend 2. Using this I can hibernate the computer using the swap partition to store the state, and bring it back up and running in a matter of a few seconds. I also have some issues with cpu throttling, I have not manage to get neither cupfreqd nor powernowd to run, this is something I have to look a bit more into.
Take care, folks!
First, the AMD64 computer that I regularly use at work. The Ubuntu for AMD64 is working flawlessly on that computer. Everything that I need is running just fine. But on friday the NIC fell apart. Usually, that is not a big deal, you just pull the card out and slip a new one in to the box. Problem is that this is a Shuttle SN25p, with the excelent NIC built right into the mainboard. Combined with noe free PCI slot on the mainboard, I do not have any other option but sending the whole box in for service. So, I am happy with the box, it works quite well, and performance is more than adequate for my work (which includes some really heavy simulations of partial differential equations), but no free slot for a emergency NIC can easily be a showstopper in the future. I am thinking on finding other use for the computer within our lab, and get another one for myself.
The PowerMac at home is also doing fine these days. As you know, I ended up installing gentoo on that computer, because the installers of both debian and ubuntu for powerpc failed to work, due to problem with disk partitioning. The installer is just unable to detect the geometry of my partitions, and give me only the option to wipe out the entire disk, including my MacOSX installation, and put in linux. Not what I want, as I use MacOSX for different purposes.
The problem was that I never used the gentoo installed. It just take to much time to compile everything, and a whole lot of packages is masked out for powerpc64, as I guess not very many gentoo users play with these animals. But then I figured out that I already have a working disk geometry and working yaboot for booting the operating system I want, so I just tried to install ubuntu for powerpc64 into or on top of the gentoo installation. For this purpose I chosed the pre-release of the next ubuntu (Dapper Drake), and run the installer in expert modus. In stead of working through the partitioning and mounting step in the installer, I switched to another console, mounted what I wanted as '/target', run the necessary swapon command, and wiped out the original gentoo stuff from /target. Most of the installer worked just fine from that point, except making the installation bootable and writing a correct fstab. The fstab I could easily handle by hand, and the configuration for the 'yaboot' boot-loader I have been clever enough to back up to a USB drive. So, after copying in the yaboot config, running 'ybin -v' and doing some editing of fstab using vim, I have a working ubuntu. Now with most of what I need available in binary form. There are some issues with the airport extreme wireless card, my firewire soundcard and the built in soundcard, but I can live with this.
With the Fujitsu P1510, the history is not very interesting. I booted from Debian sarge net-install cd, performed a regular installation, upgraded to etch. Everything is just fine. Bluetooth, sound, WiFi, Xorg - everything works. There are only a few minior things. Suspend/resume is not working out of the box, so I patched the kernel with software suspend 2. Using this I can hibernate the computer using the swap partition to store the state, and bring it back up and running in a matter of a few seconds. I also have some issues with cpu throttling, I have not manage to get neither cupfreqd nor powernowd to run, this is something I have to look a bit more into.
Take care, folks!
Thursday, December 15, 2005
Waiting is over!
The Fujitsu P1520 arrived today! It's a very sweet, little thing.
I am installing Linux on it right now, a report on this will follow shortly.
Take care,
I am installing Linux on it right now, a report on this will follow shortly.
Take care,
NRK.no - EU - Store all phone and net traffic for two years
NRK.no - EU
The european parliment have just approved a new directive, which state that all phone and internet traffic should be stored for two years. That is just crazy. Apparently, only phonenumbers you call, where you are at what time ip-addresses for communication, etc. But that is pretty extensive. Doesn't sound very well thought about and very crazy.
Hey, I have nothing to hide, but I don't like that idea at all, so I better go dig up those urls again that describes software that can be used to hide your tracks.
Take care - and be scared.
The european parliment have just approved a new directive, which state that all phone and internet traffic should be stored for two years. That is just crazy. Apparently, only phonenumbers you call, where you are at what time ip-addresses for communication, etc. But that is pretty extensive. Doesn't sound very well thought about and very crazy.
Hey, I have nothing to hide, but I don't like that idea at all, so I better go dig up those urls again that describes software that can be used to hide your tracks.
Take care - and be scared.
Thursday, December 08, 2005
Waiting for Fujitsu Siemens P1510
This is a story in progress!
Back in early october I ordered the new Fujitsu Siemens P1510, a new very cool and very small laptop, from our regular dealer. I was promised that delivery will be around 15. november.
When we approached 15. november, I was told that there was a slight delay, 1. december will be the great day. Somehow I don't belive what I was told. And very correct, there was a further delay, 7. december should be the delivery date. Now I was also told that the computer was shipped from germany, probably from some Fujitsu Siemens manufacturing facility there, so now further delay was expected.
But, hey, thats too simple, isn't it? Yesterday, which was the 7. december, I was again told that a new delay was in progress. New delivery date is 14. december. But then I have some questions. My computer was shipped, right? So, who in this world got my computer? A more important customer? Does someone paid off somebody in the delivery chain to get my computer? Maybe it wasn't shipped at all?
The important thing is that this happens all the time. Should we just take this any longer? When I order an item, and get a delivery date, I expect the item to show up. I guess that if you are a craftsman, and order a new tool, a chainsaw or something like that, and get a delivery date, you expect the item to be around at that time, and tell your customer that you have to wait until that date with delivery. Could the craftsman live with one month or more delay? I guess not. He turn to another vendor. Why can't I do the same? Because every single company in this business is more or less the same.
So, buyers in all countries, speak up. Don't take this any longer. Give the message to your dealer that you want the item on the day they promise. It's much better that the dealer give a correct but long into the future delivery date , than specifying delivery dates that he can not keep up with.
I think I in the future will request a 1% reduction in price for each day the delivery is delayed. If we all request this, maybe we can get a healtier attitude in the business.
Back in early october I ordered the new Fujitsu Siemens P1510, a new very cool and very small laptop, from our regular dealer. I was promised that delivery will be around 15. november.
When we approached 15. november, I was told that there was a slight delay, 1. december will be the great day. Somehow I don't belive what I was told. And very correct, there was a further delay, 7. december should be the delivery date. Now I was also told that the computer was shipped from germany, probably from some Fujitsu Siemens manufacturing facility there, so now further delay was expected.
But, hey, thats too simple, isn't it? Yesterday, which was the 7. december, I was again told that a new delay was in progress. New delivery date is 14. december. But then I have some questions. My computer was shipped, right? So, who in this world got my computer? A more important customer? Does someone paid off somebody in the delivery chain to get my computer? Maybe it wasn't shipped at all?
The important thing is that this happens all the time. Should we just take this any longer? When I order an item, and get a delivery date, I expect the item to show up. I guess that if you are a craftsman, and order a new tool, a chainsaw or something like that, and get a delivery date, you expect the item to be around at that time, and tell your customer that you have to wait until that date with delivery. Could the craftsman live with one month or more delay? I guess not. He turn to another vendor. Why can't I do the same? Because every single company in this business is more or less the same.
So, buyers in all countries, speak up. Don't take this any longer. Give the message to your dealer that you want the item on the day they promise. It's much better that the dealer give a correct but long into the future delivery date , than specifying delivery dates that he can not keep up with.
I think I in the future will request a 1% reduction in price for each day the delivery is delayed. If we all request this, maybe we can get a healtier attitude in the business.
Wednesday, November 30, 2005
Slashdot | Microsoft Open Document Standard Not So Open
Slashdot | Microsoft Open Document Standard Not So Open: "Microsoft Open Document Standard Not So Open"
Please, can someone explain to me why I am not shocked? I guess that as long as water flows, sun raises and sets, Microsoft will never be really open. It's just against their nature, they can't be something they aren't.
Live free and prosper!
Please, can someone explain to me why I am not shocked? I guess that as long as water flows, sun raises and sets, Microsoft will never be really open. It's just against their nature, they can't be something they aren't.
Live free and prosper!
Sunday, October 30, 2005
Linux on a G5 - Followup
You might wonder how it went with the Linux install on my G5. Please read the previous post on this if you are one of my new readers!
It turned out that the Debian installer still couldn't figure out where to install itself, even when the free space ready for Linux was first on the disk. There is certainly something strange with the partition scheme from Apple. As a last resort I went to gentoo. What you can't do with gentoo, I guess you can't do at all. To make the story short, gentoo works as a charm. During my tests with the other distributions, I figured out how to create partitions and mount them using mac-fdisk, mkfs and mount directly. The problem was that even when I mounted partitions, the installer couldn't find them. And if I got beyond that point, I still wasn't abel to install the bootloader properly. With gentoo, everything is manual and up to me, so it was much more easy to perform the install.
The only problem I really had was with yaboot configuration. The yabootconfig script that should have created a yaboot.conf file was nonfunctional with my setup, so I had to write the config manually, based on some examples I found online. That wasn't that hard, but it took me some time to figure out the ofpath id's I needed to put in there, as well as which partition number the Tiger actually should boot from. It turned out that I should not specify the partition where the Tiger lives, but the one just infront of it.
So - end of the story is that I now have a working Tiger and a working gentoo, though it will still take some time to get all the software I need on gentoo. That's why I don't like gentoo that much. The power of the gentoo install, coupled with the power of apt-get/dpkg, that's really what I want. But I guess I stick with gentoo for now - it will certainly give me a lean, mean machine!
A short note on the Shuttle SN25P case is also appropriate here. I can tweak the ubuntu really to what I want. I got NIS and automount working, I have sound and DVD burning working, and most of the packages I want. Actually, I just think about it as Debian, and it works quite fine. Guess I can live with that!
Take care!
It turned out that the Debian installer still couldn't figure out where to install itself, even when the free space ready for Linux was first on the disk. There is certainly something strange with the partition scheme from Apple. As a last resort I went to gentoo. What you can't do with gentoo, I guess you can't do at all. To make the story short, gentoo works as a charm. During my tests with the other distributions, I figured out how to create partitions and mount them using mac-fdisk, mkfs and mount directly. The problem was that even when I mounted partitions, the installer couldn't find them. And if I got beyond that point, I still wasn't abel to install the bootloader properly. With gentoo, everything is manual and up to me, so it was much more easy to perform the install.
The only problem I really had was with yaboot configuration. The yabootconfig script that should have created a yaboot.conf file was nonfunctional with my setup, so I had to write the config manually, based on some examples I found online. That wasn't that hard, but it took me some time to figure out the ofpath id's I needed to put in there, as well as which partition number the Tiger actually should boot from. It turned out that I should not specify the partition where the Tiger lives, but the one just infront of it.
So - end of the story is that I now have a working Tiger and a working gentoo, though it will still take some time to get all the software I need on gentoo. That's why I don't like gentoo that much. The power of the gentoo install, coupled with the power of apt-get/dpkg, that's really what I want. But I guess I stick with gentoo for now - it will certainly give me a lean, mean machine!
A short note on the Shuttle SN25P case is also appropriate here. I can tweak the ubuntu really to what I want. I got NIS and automount working, I have sound and DVD burning working, and most of the packages I want. Actually, I just think about it as Debian, and it works quite fine. Guess I can live with that!
Take care!
Wednesday, October 26, 2005
Linux on a G5 PowerMac
I finally wanted to put linux on my G5 PowerMac, as I see my main linux computer today, a laptop, soon will be gone, replaced with a smaller and sweater laptop (the Fujitsu Siemens P1510). As I now have the Shuttle SN25P at the office, I'll also need a stationary computer at home running Linux. Don't get me wrong, the Mac OS X system is a decent system. Unfortunately, it's not up to my standards for doing real work.
This should be an easy task, right. I already have a free external firewire disk, and my intention was to load Linux on that one. Inserted the ubuntu cd (read the story on the Shuttle to get why I just started out with ubuntu), and keep going. Right. Tap, tap, tap, enter, enter, enter, ooopss. I accidentally erased the MacOSX partition instead of repartitioning the external drive. So, thats an extreme makeover for my MacOSX installation. Anyway, most important stuff I have elsewhere, so no big deal. So, took two steps back, installed Tiger on the internal drive, and went back to my original quest. Installation on the external drive went just fine, up to the point where the bootloader, yaboot, is to be installed. It can't be installed on the external firewire drive, at least not from the ubuntu installer. Another couple of steps back.
Next try, go for a split of the internal system. So, I install Tiger again on a smaller partition, leaving a spaceious 80GB towards the end of the drive for Linux. Bahh.. The ubuntu installers is unable to find the free partition and install there. So, I figured out I should start with partitioning from the linux side. So I did - removed the Tiger again, created partitions for Linux at the end of the disk, installed ubuntu. Only to discover that the mac disk utility is unable to create HFS+ partition in the free space for Tiger. I lost counting, but had to destroy everything once more.
This time I did some googling, especially on the Debian installer documentation for PPC. There I got the following recipe: Start with installation of MacOSX, leave free space for Debian at the beginning of the disk. Then, install MacOSX, and finally install Debian.
So, now I will try that and see how it goes. Btw, the one time I actually got ubuntu installed on the internal drive (alone, no MacOSX present) I was unable to boot it, because the SATA drivers was not loaded in the kernel at boottime. I guess I need to use initrd or something, but have to figure that out later on.
On to the next install, I'll see you all later :-)
This should be an easy task, right. I already have a free external firewire disk, and my intention was to load Linux on that one. Inserted the ubuntu cd (read the story on the Shuttle to get why I just started out with ubuntu), and keep going. Right. Tap, tap, tap, enter, enter, enter, ooopss. I accidentally erased the MacOSX partition instead of repartitioning the external drive. So, thats an extreme makeover for my MacOSX installation. Anyway, most important stuff I have elsewhere, so no big deal. So, took two steps back, installed Tiger on the internal drive, and went back to my original quest. Installation on the external drive went just fine, up to the point where the bootloader, yaboot, is to be installed. It can't be installed on the external firewire drive, at least not from the ubuntu installer. Another couple of steps back.
Next try, go for a split of the internal system. So, I install Tiger again on a smaller partition, leaving a spaceious 80GB towards the end of the drive for Linux. Bahh.. The ubuntu installers is unable to find the free partition and install there. So, I figured out I should start with partitioning from the linux side. So I did - removed the Tiger again, created partitions for Linux at the end of the disk, installed ubuntu. Only to discover that the mac disk utility is unable to create HFS+ partition in the free space for Tiger. I lost counting, but had to destroy everything once more.
This time I did some googling, especially on the Debian installer documentation for PPC. There I got the following recipe: Start with installation of MacOSX, leave free space for Debian at the beginning of the disk. Then, install MacOSX, and finally install Debian.
So, now I will try that and see how it goes. Btw, the one time I actually got ubuntu installed on the internal drive (alone, no MacOSX present) I was unable to boot it, because the SATA drivers was not loaded in the kernel at boottime. I guess I need to use initrd or something, but have to figure that out later on.
On to the next install, I'll see you all later :-)
Subscribe to:
Posts (Atom)