Yeah, because I also had some bug reports filed.
However, for the moment I do not use wmaker much.
On my small portable (233 MHz) I installed XFCE4 for my wife, on my big portable (AMD64 3000+) I use KDE,
on my server I use Enlightenment with the Aqua theme, come to think of it, the workstation I use to connect to my server still runs wmaker.
I hope it makes it quickly into Debian unstable/testing.
I do play with it. I set up a local mirror at my home, in which I download every night the updates of the packages.
I use this to automatically upgrade all my UML instances that I am running, and also to easily upgrade my portables and my father's computer (from my portable).
I haven't seen much breakage in more than 6 months of mirroring, occasionally postfix which cannot be automatically configured, but that's all.
I think that the community around Fedora should do the same, or maybe they do that already.
Patent costs are larger. I work for Philips, and we had a request from the IP group, if we please could file some more (software) patent requests.
In the lecture we got, the complete cost for entering a valid patent was mentioned, it is about one million euro per patent.
Not sure if this is because Philips is a global company, but I can be sure that it is because much investigations need to be done. When more software patents are approved, I can only assume that the total cost of filing a patent will only get larger and larger, because of the amount of patents to be investigated.
This reminds me of Terry Pratchett's 'The Truth', where he states that there are two ways of making gold from lead, the first one by using alchemy, what he calls the easy way, the second one the hard way, and the hard way is the printing press.
The same can be said for making money from software, the easy way is by selling software, the hard way is by making your money around the software.
Being a European (Belgium) myself, I think the premise was that the people who went on strike in the 19th century, had not much to lose anyway at that time.
I think the wave of solidarity that struck then has never been matched again.
I have been employed between workers. When someone says strike, then they all strike together. It has also to do with peer pressure. If you don't strike, you are not part of the group.
But I think the strikes where more effective as a signal by a large mass of people, than on the economic impacts they had on the industry back then.
I do think that we should be able to pull off something similar like the 19th century strikes.
We have computers and we have a world-spanning network.
These can enable us to connect together in large groups and do something likewise (like Downhill Battle) : mobilise people to take action, to print standard letters and sign them personally and send them to selected targets : news stations, papers, politicians and create signals from large masses of people.
Fax bombing is forbidden and unpolite, and spamming too, but image that certain targets get a flow of all individually addressed snail mail, in all kinds of envelopes and with all kinds of stamps. They have to open and read them, because they can't know what is inside.
When such a campaign starts, at the same time a press release should be sent out.
It should be easy for people to send a letter in the mail. Someone with the necessary skills should draft a letter, post it on the web and make it easy to print out and sign manually. Other people should find out addresses (or one address) and post it likewise. The senders should write (with a pen) the destination.
One stamp does not cost much, but consider the collective power of 10000 stamps.
It should also be easy to reach people when new campaigns start, and guidelines given on how to convince new people to join.
Actually, I have several books of Niven, and of Niven and Pournelle, and of Pournelle only, and it seems to me that Pournelle is more the right-wing type than Niven.
If you have read Pournelle's monthly columns in Byte, you can gather his right-wing stances from the comments in between too.
I have done fair administrative programming, and from my experience the best languages for administrative programming where COBOL and the xBase dialects, e.g. Clipper and FoxPro.
The main reasons for this are :
1) Datastructures can be declared and used consistently, and do not have much clutter.
2) They work fantastic with fixed point values for currency.
From all the other languages I know, only C, C++ and Ada are good at point 1. In script languages you cannot define datastructures and address the consistently without clutter (Perl : $rec{field}, $rec->field; Python rec['field'], I am sure you have your own examples).
If they are good at point 1, then they mostly suck at point 2.
I learned COBOL 10 years after I learned C, Pascal, xBase and I find it not so bad. However, I had the opportunity to use compilers which did not need as much red tape as older implementations.
I think that this configuration gives the best proce performance. On one end of the spectrum you have a completely interconnected mesh, on the other hand you have all systems on a single bus, in between you have a whole lot of possible topographies.
In the hypercube system, you need in N dimensions N communication channels per processor, and the maximum distance that any packet has two travel is N hops.
In a complete interconnected mesh with P processors, you need P communication channels per processor. While your maximum hop is in that case only 1, you need (P - 1)*P communication channels, which is quadratic.
The limited liability here in Belgium (probably also Europe) is in case of bankruptcy, but does not count in the first three years of the existence of the corporation, and neither when fraudulent bankruptcy can be proven.
A long time ago I read an article (I think it was in an IEEE pub.) about supercomputers, and their general conclusion about the current ones on the market then, was that most speed ratings giving where burst ratings (eg. Cray-1, was 80 Mflops burst), but most could only give only about 10% sustained performance of their burst performance.
The talk of the panel about parallellisation is the same for normal supercomputers. If you want high sustained performance, then you need an algorithm that is highly vectorisable, and if you look at how such a computing unit is built, it is also a highly parallel system.
What's the death of your high speed ? Many decision making tasks. It always comes down to the same, or you have a high speed integer system which is good at decision making, or you have one that can run continuus floating-point computations, but not both together.
At the end of last year I bought two computers, one for me and one for my father.
These are the specs and the prices :
2 Ghz Celeron, 256 MB, 40 GB, 240EUR
2 GHz Athlon, 2 GB, 250 GB RAID, 2500EUR
Now use Moore's law to extrapolate (I did it backwards with all the systems I ever bought and it mostly fits).
This means that at the end of 2006 I should be able to get a system that is half the capacity of my current high-end system for 240EUR (but with 8 GHz speed), and mid 2008 a system that has the same capacity as mine now.
Extrapolating for the high-end system I should get for 2500EUR a system with 8GB RAM and 1TB storage, re. 16 GB RAM and 2TB storage.
What to do with such systems ? The low-end system currently runs Linux, and runs it pretty well.
The high-end system runs Red Hat as base system, and currently also 5 UML virtual machines with Debian, which I hope to expand next month to 8, for VM's with SuSE, Mandrake and Slackware Linux.
Interesting ? Even more than that I think. With a hopeful rising of bandwidth, within a few years community clusters should be possible with HPC and Terabyte storage performance.
I am looking out for the growth of what I call the collective memory, a massively parallel, loosely coupled community system where people can store and find information, and be sure that information is preserved, because of the redundancy of the system.
I have been running since two months a virtual network of 5 UML instances on my Red Hat 9 system. They all run Debian into memory spaces of 64, 128 and 256 Mb of memory.
Their functions are separated into :
A mirror for Debian testing
A Debian 'unstable' configuration
A Zope application server
A postgreSQL database server
A workstation setup
All UML instances update themselves automatically every day from the mirror server.
The current uptime for the complete system is now almost 23 days (mostly due to maintenance and some cable problems).
Future plans : move my mail hub from my Linux router/firewall/proxy combo to a UML instance, add NFS and NIS to share common home directories, and add filesystem images for Red Hat, Mandrake, SuSe, Slackware.
Slashdot Mirror
Yeah, because I also had some bug reports filed. However, for the moment I do not use wmaker much. On my small portable (233 MHz) I installed XFCE4 for my wife, on my big portable (AMD64 3000+) I use KDE, on my server I use Enlightenment with the Aqua theme, come to think of it, the workstation I use to connect to my server still runs wmaker. I hope it makes it quickly into Debian unstable/testing.
I do play with it. I set up a local mirror at my home, in which I download every night the updates of the packages.
I use this to automatically upgrade all my UML instances that I am running, and also to easily upgrade my portables and my father's computer (from my portable).
I haven't seen much breakage in more than 6 months of mirroring, occasionally postfix which cannot be automatically configured, but that's all.
I think that the community around Fedora should do the same, or maybe they do that already.
Is 'Core N' a stable snapshot, or what ?
I already used sidebar sliders on my Sinclair QL in 1987, and I am sure that these things were already in Smalltalk-80 (1980).
Patent costs are larger. I work for Philips, and we had a request from the IP group, if we please could file some more (software) patent requests.
In the lecture we got, the complete cost for entering a valid patent was mentioned, it is about one million euro per patent.
Not sure if this is because Philips is a global company, but I can be sure that it is because much investigations need to be done. When more software patents are approved, I can only assume that the total cost of filing a patent will only get larger and larger, because of the amount of patents to be investigated.
I thought that CORBA already made arrangements for all these things.
This reminds me of Terry Pratchett's 'The Truth', where he states that there are two ways of making gold from lead, the first one by using alchemy, what he calls the easy way, the second one the hard way, and the hard way is the printing press.
The same can be said for making money from software, the easy way is by selling software, the hard way is by making your money around the software.
Being a European (Belgium) myself, I think the premise was that the people who went on strike in the 19th century, had not much to lose anyway at that time.
I think the wave of solidarity that struck then has never been matched again.
I have been employed between workers. When someone says strike, then they all strike together. It has also to do with peer pressure. If you don't strike, you are not part of the group.
But I think the strikes where more effective as a signal by a large mass of people, than on the economic impacts they had on the industry back then.
I do think that we should be able to pull off something similar like the 19th century strikes.
We have computers and we have a world-spanning network.
These can enable us to connect together in large groups and do something likewise (like Downhill Battle) : mobilise people to take action, to print standard letters and sign them personally and send them to selected targets : news stations, papers, politicians and create signals from large masses of people.
Fax bombing is forbidden and unpolite, and spamming too, but image that certain targets get a flow of all individually addressed snail mail, in all kinds of envelopes and with all kinds of stamps. They have to open and read them, because they can't know what is inside.
When such a campaign starts, at the same time a press release should be sent out.
It should be easy for people to send a letter in the mail. Someone with the necessary skills should draft a letter, post it on the web and make it easy to print out and sign manually. Other people should find out addresses (or one address) and post it likewise. The senders should write (with a pen) the destination.
One stamp does not cost much, but consider the collective power of 10000 stamps.
It should also be easy to reach people when new campaigns start, and guidelines given on how to convince new people to join.
Actually, I have several books of Niven, and of Niven and Pournelle, and of Pournelle only, and it seems to me that Pournelle is more the right-wing type than Niven.
If you have read Pournelle's monthly columns in Byte, you can gather his right-wing stances from the comments in between too.
Just like in Jakarta, also Thursday.
I have done fair administrative programming, and from my experience the best languages for administrative programming where COBOL and the xBase dialects, e.g. Clipper and FoxPro.
The main reasons for this are :
1) Datastructures can be declared and used consistently, and do not have much clutter.
2) They work fantastic with fixed point values for currency.
From all the other languages I know, only C, C++ and Ada are good at point 1. In script languages you cannot define datastructures and address the consistently without clutter (Perl : $rec{field}, $rec->field; Python rec['field'], I am sure you have your own examples).
If they are good at point 1, then they mostly suck at point 2.
I learned COBOL 10 years after I learned C, Pascal, xBase and I find it not so bad. However, I had the opportunity to use compilers which did not need as much red tape as older implementations.
When I started using Linux in 1998 I used olvwm.
I think that this configuration gives the best proce performance. On one end of the spectrum you have a completely interconnected mesh, on the other hand you have all systems on a single bus, in between you have a whole lot of possible topographies.
In the hypercube system, you need in N dimensions N communication channels per processor, and the maximum distance that any packet has two travel is N hops.
In a complete interconnected mesh with P processors, you need P communication channels per processor. While your maximum hop is in that case only 1, you need (P - 1)*P communication channels, which is quadratic.
There is a very nice Aqua theme available for E.
It is called a hypercube because it is a mapping of a four-dimensional (and higher) cube into three-dimensional space.
It has nothing to do with communications or computing, but with topology.
And if it is a religious society, the leaders know damn well how to twist it to use the religion for their political agenda.
Have a look at the wars from the fourteenth, fifteenth and sixteenth century, and the advances of the Turks into Eastern Europe.
I am sure that there are people who know other things.
IIRC, the Japanese did already experiments in the eighties with software factories, in an attempt to beat USA in producing software.
It never worked out, because the Americans where more creative and skilled in programming.
So MS is its usual self, thieving and lying.
The limited liability here in Belgium (probably also Europe) is in case of bankruptcy, but does not count in the first three years of the existence of the corporation, and neither when fraudulent bankruptcy can be proven.
It is all about making money.
A long time ago I read an article (I think it was in an IEEE pub.) about supercomputers, and their general conclusion about the current ones on the market then, was that most speed ratings giving where burst ratings (eg. Cray-1, was 80 Mflops burst), but most could only give only about 10% sustained performance of their burst performance.
The talk of the panel about parallellisation is the same for normal supercomputers. If you want high sustained performance, then you need an algorithm that is highly vectorisable, and if you look at how such a computing unit is built, it is also a highly parallel system.
What's the death of your high speed ? Many decision making tasks. It always comes down to the same, or you have a high speed integer system which is good at decision making, or you have one that can run continuus floating-point computations, but not both together.
Especially since Moore law still seems to hold.
At the end of last year I bought two computers, one for me and one for my father.
These are the specs and the prices :
Now use Moore's law to extrapolate (I did it backwards with all the systems I ever bought and it mostly fits).
This means that at the end of 2006 I should be able to get a system that is half the capacity of my current high-end system for 240EUR (but with 8 GHz speed), and mid 2008 a system that has the same capacity as mine now.
Extrapolating for the high-end system I should get for 2500EUR a system with 8GB RAM and 1TB storage, re. 16 GB RAM and 2TB storage.
What to do with such systems ? The low-end system currently runs Linux, and runs it pretty well.
The high-end system runs Red Hat as base system, and currently also 5 UML virtual machines with Debian, which I hope to expand next month to 8, for VM's with SuSE, Mandrake and Slackware Linux.
Interesting ? Even more than that I think. With a hopeful rising of bandwidth, within a few years community clusters should be possible with HPC and Terabyte storage performance.
I am looking out for the growth of what I call the collective memory, a massively parallel, loosely coupled community system where people can store and find information, and be sure that information is preserved, because of the redundancy of the system.
Jurgen
I agree on the Basic and Windows part, but not on the COBOL.
I presume that many COBOL programs still run because they are too complex for analysis and hence too expensive to port to newer languages.
Btw, if you remove COBOL, does it not seem strange that the remaining parts are heavily MS centric ?
Asking the vendor to open the source about the measurement system and using programs like ethereal or iptraf to compare what is being measured.
You have to design for latency, but bandwidth can be bought.
This means that the engineering problems are in removing the lag and providing a synchronous feed.
If you need more bandwidth, just install more pipes.
I have been running since two months a virtual network of 5 UML instances on my Red Hat 9 system. They all run Debian into memory spaces of 64, 128 and 256 Mb of memory.
Their functions are separated into :
All UML instances update themselves automatically every day from the mirror server.
The current uptime for the complete system is now almost 23 days (mostly due to maintenance and some cable problems).
Future plans : move my mail hub from my Linux router/firewall/proxy combo to a UML instance, add NFS and NIS to share common home directories, and add filesystem images for Red Hat, Mandrake, SuSe, Slackware.
As in Liquid Crystal Diode ?
Or as in Light Emitting Display ?