I have my own very ideas:) I believe that XML is good but very verbose; anyhow, the array of parsers and stuff would make working with it really simple... So it's a naught point that XML config files are hard to edit if an XML-editor becomes as standard as vi.
A good idea I saw on ATG-Dynamo (who uses both XML configs and plain text), is the idea of configurable config-layers, that is, that it checks in order different places for config (kinda like looking first in/etc then in ~/.foo, etc.), letting you choose which 'directories' are read *and* having plenty of flexibility to combine the different files (i.e., concatenate, replace,...).
The only thing I can think of that is missing in that for a general config system would be the addition of 'final' configuration, i.e: when an attribute is finalized, no later layer of configuration can change it.
What I'd like to know is... what is so special about this distro?
Yeah, it doesn't offer binary packages.
I don't know about you, but I apt-get install most packages and build the ones that I want to build (to./configure things and when I want to go farther than Sid). To be fair, their build process looks pretty neat.
Let's remember than even people who build packages won't check all the sources for trojans, that processor optimizations are mostly unnoticeable and that most binaries come appropiately configured and compiled. And some builds can be so slooow...
Oops, I forgot. Another problem of the last point is that the compression/decompression would be unbearably slow of very heavy in memory, at least with the methods I can think of
To sum things up, simple maths tells us that no compression algorithm can compress all files of length n by at least a bit. Having an algorithm like that would yield infinite compression... there's no demostration needed there, that just doesn't make sense.
Second, Information Theory says that you cannot compress data of n-bits of entropy to less than n bits. Data is said to be 'random' if it is n bits long and it has n bits of entropy (that is not accurate, I know).
You can cheat and invent an algorithm that compresses *1* random string of data to a byte and adds one byte to the rest, so you can undo that transformation easily. There you are, you have an algorithm that compresses random data!
The compression faq (and I guess that Kolmogorov says so, but I don't know) evades those tricks asking for compressed size+decompressor must be less than the uncompressed size.
The only point I can find there is, what you call random data. Suppose a text encoded in ASCII in bytes. That is supposed to have a low entropy. Now take that data in 9-bit chunks and measure entropy. It will be higher. Now, entropy and randomness depend of how you look at data.
This takes us to what's a random file. I'm sure that the guy with the compression faq challenge would give you a *very* random file, with a really even distribution of characters, little repeated sequences, no long streams of 77's, etc. Is that a random file...? It has been doctored. In fact, if you count how many possible files he could give you, it would be less than all the possible files of that length, therefore, you wouldn't need as many bits to represent them all.
The problem is (and I'm sure someone who knows more combinatorics than me), is that it mustn't work pretty well... I'd say that the compression must be less than one bit (intuitive reasoning)...
It has been greatly commented that companies tend to depend on external IM (ICQ, Yahoo, MSN, you name it). I'm amazed that noone has mentioned Jabber! That's one server you can deploy (*NIX only for the moment, Win experimental) and it provides some nice features for the business (and jabber.com deals with business); encryption, automatic 'buddies' (so everyone gets as a default everyone in their department) and it's very light and available in nearly all platforms, with a nice variety of clients.
It's quite useful that you can run your own server, no?
First of all, I don't know if it's me, but the modern skin doesn't display well; I solved it by substituting chrome/modern.jar for an old version.
What I'd like to know is how performance is going in Linux and platforms besides Windows. I use the Windows version and it is real fine, but I've heard that other versions are slower.
Hum! Maybe I'm guessing here, but are you using Linux as your desktop? I think a lot of people are using Linux on their desktops and it works quite well.
I don't believe it's that hard to have some configuration tools that are easy enough for the beginner. The other necessity is a good unified environment, things look nice on the gui front (KDE and Gnome are quite good, but perhaps it would be good if one won over the other and we had ONE desktop for newbies), but there are problems in the inner works...
We have some sort of distro hell here, because different distributions work in different ways, put their config files and things in different places, and that is a big tech-support problem. I used to work on this and, believe me, it's good to tell someone to do start->run->winipcfg to learn about their IP address.
Linux needs some standards, at least for config files and directories. One simple way to set up dial-up, for starters!
I don't think we need more OS's, though I would welcome them; laying my hopes on the new Amiga, even; but it's quite difficult to catch up on the established-I find it surprising that Linux has come to this level of popularity.
As a spanish user, cable is not very common here; but the local dumb telco is giving DSL on more territory than I could hope, including my home:)
Though there seem to be loads of problems of performance and reliability, I have a constant flow of info (28'5k/s downstream and 16+ k/s upstream, really good for a 256-128 Kbps setup). I think I've been for three or four months without noticing ANY problem with my connexion.
The only problem is that the modem they gave me is a 3Com Homeconnect PCI modem, still unsupported in Linux (hey! if any of you knows otherwise, mail me at koali@mailandnews.com!); but it seems now they are offering a choice of internal/external modems (still, they are not willing to change mine).
If I have to talk strictly from personal experience, I would recommend ADSL anytime. But the lucky ones with cable over there seem to be ok, too...
The problem with *NIX is that there are a lot of programmers using it (that is also its major benefit). Every coder out there is incapable of accepting than someone's code is better than his (except in some little situations; e.g. they are too lazy to code it, don't know how)... and considering the number of applications that are not that difficult to program, we end up having a thousand of different mail clients.
I believe it is too much to ask people to unify identical projects... I find hard to believe that there aren't more popular alternatives to Gnome and KDE.
That is the problem of Linux, besides some standarized tools, everything else has about twenty different flavours.
I don't see how to get out of this, rather than scraping it all. And I mean all.
Mac OsX is very interesting for me. If everything there is configured by XML and they enforce GUI guidelines (I forgot that's Apple), we could have a real cool OS; if they don't make the command line inaccessible and developing is not frowned upon.
Perhaps even the rumoured new Amiga is the solution, as people can use it on top of another OSes; and maybe the brandname is so strong as to pull people to it if they do the marketing right.
That's why I fear that Linux is going to face a problem in the near-future, as choice is good, but too much of a good thing is always bad
I don't get it. What is the motivation to develop software for a machine that is, frankly speaking, technologically obsolete?
That could be said of the whole PC world? Does 640k ring a bell anywhere? MS-DOS (dead in Windows Millenium?... could be)? Layers and layers of stuff to cover and aging (and awful) base?
I still have my Amiga, and with a 68060, at 50mhz, it sure feels faster than my 500Mhz Athlon, by the way. And it has a decent OS (much more efficient than Windoze and more ellegant and simple than UN*X)... it just doesn't play Quake (oops, I forgot. It does).
Slashdot Mirror
I have my own very ideas :) I believe that XML is good but very verbose; anyhow, the array of parsers and stuff would make working with it really simple... So it's a naught point that XML config files are hard to edit if an XML-editor becomes as standard as vi.
A good idea I saw on ATG-Dynamo (who uses both XML configs and plain text), is the idea of configurable config-layers, that is, that it checks in order different places for config (kinda like looking first in /etc then in ~/.foo, etc.), letting you choose which 'directories' are read *and* having plenty of flexibility to combine the different files (i.e., concatenate, replace, ...).
The only thing I can think of that is missing in that for a general config system would be the addition of 'final' configuration, i.e: when an attribute is finalized, no later layer of configuration can change it.
Yeah, it doesn't offer binary packages.
I don't know about you, but I apt-get install most packages and build the ones that I want to build (to ./configure things and when I want to go farther than Sid). To be fair, their build process looks pretty neat.
Let's remember than even people who build packages won't check all the sources for trojans, that processor optimizations are mostly unnoticeable and that most binaries come appropiately configured and compiled. And some builds can be so slooow...
Oops, I forgot. Another problem of the last point is that the compression/decompression would be unbearably slow of very heavy in memory, at least with the methods I can think of
Second, Information Theory says that you cannot compress data of n-bits of entropy to less than n bits. Data is said to be 'random' if it is n bits long and it has n bits of entropy (that is not accurate, I know).
You can cheat and invent an algorithm that compresses *1* random string of data to a byte and adds one byte to the rest, so you can undo that transformation easily. There you are, you have an algorithm that compresses random data!
The compression faq (and I guess that Kolmogorov says so, but I don't know) evades those tricks asking for compressed size+decompressor must be less than the uncompressed size.
The only point I can find there is, what you call random data. Suppose a text encoded in ASCII in bytes. That is supposed to have a low entropy. Now take that data in 9-bit chunks and measure entropy. It will be higher. Now, entropy and randomness depend of how you look at data.
This takes us to what's a random file. I'm sure that the guy with the compression faq challenge would give you a *very* random file, with a really even distribution of characters, little repeated sequences, no long streams of 77's, etc. Is that a random file...? It has been doctored. In fact, if you count how many possible files he could give you, it would be less than all the possible files of that length, therefore, you wouldn't need as many bits to represent them all.
The problem is (and I'm sure someone who knows more combinatorics than me), is that it mustn't work pretty well... I'd say that the compression must be less than one bit (intuitive reasoning)...
Of course, this has little to do with the matter at hand
It has been greatly commented that companies tend to depend on external IM (ICQ, Yahoo, MSN, you name it). I'm amazed that noone has mentioned Jabber! That's one server you can deploy (*NIX only for the moment, Win experimental) and it provides some nice features for the business (and jabber.com deals with business); encryption, automatic 'buddies' (so everyone gets as a default everyone in their department) and it's very light and available in nearly all platforms, with a nice variety of clients.
It's quite useful that you can run your own server, no?
What I'd like to know is how performance is going in Linux and platforms besides Windows. I use the Windows version and it is real fine, but I've heard that other versions are slower.
Will it come with arms and legs? Fully articulated? Easy disassembly?
Anyway, it's nice to see the Python on the 'net; even appearing besides CowboyNeal in a slashdot poll!
Hum! Maybe I'm guessing here, but are you using Linux as your desktop? I think a lot of people are using Linux on their desktops and it works quite well.
I don't believe it's that hard to have some configuration tools that are easy enough for the beginner. The other necessity is a good unified environment, things look nice on the gui front (KDE and Gnome are quite good, but perhaps it would be good if one won over the other and we had ONE desktop for newbies), but there are problems in the inner works...
We have some sort of distro hell here, because different distributions work in different ways, put their config files and things in different places, and that is a big tech-support problem. I used to work on this and, believe me, it's good to tell someone to do start->run->winipcfg to learn about their IP address.
Linux needs some standards, at least for config files and directories. One simple way to set up dial-up, for starters!
I don't think we need more OS's, though I would welcome them; laying my hopes on the new Amiga, even; but it's quite difficult to catch up on the established-I find it surprising that Linux has come to this level of popularity.
As a spanish user, cable is not very common here; but the local dumb telco is giving DSL on more territory than I could hope, including my home :)
Though there seem to be loads of problems of performance and reliability, I have a constant flow of info (28'5k/s downstream and 16+ k/s upstream, really good for a 256-128 Kbps setup). I think I've been for three or four months without noticing ANY problem with my connexion.
The only problem is that the modem they gave me is a 3Com Homeconnect PCI modem, still unsupported in Linux (hey! if any of you knows otherwise, mail me at koali@mailandnews.com!); but it seems now they are offering a choice of internal/external modems (still, they are not willing to change mine).
If I have to talk strictly from personal experience, I would recommend ADSL anytime. But the lucky ones with cable over there seem to be ok, too...
>The whole of the Amiga's custom hardware was >public, and there was never a clone of that
Well, but I seem to recall it wasn't THAT documented. They had to resort to photographs of the chips to upgrade it, at one point
The problem with *NIX is that there are a lot of programmers using it (that is also its major benefit). Every coder out there is incapable of accepting than someone's code is better than his (except in some little situations; e.g. they are too lazy to code it, don't know how)... and considering the number of applications that are not that difficult to program, we end up having a thousand of different mail clients.
I believe it is too much to ask people to unify identical projects... I find hard to believe that there aren't more popular alternatives to Gnome and KDE.
That is the problem of Linux, besides some standarized tools, everything else has about twenty different flavours.
I don't see how to get out of this, rather than scraping it all. And I mean all.
Mac OsX is very interesting for me. If everything there is configured by XML and they enforce GUI guidelines (I forgot that's Apple), we could have a real cool OS; if they don't make the command line inaccessible and developing is not frowned upon.
Perhaps even the rumoured new Amiga is the solution, as people can use it on top of another OSes; and maybe the brandname is so strong as to pull people to it if they do the marketing right.
That's why I fear that Linux is going to face a problem in the near-future, as choice is good, but too much of a good thing is always bad
Guns? How clever... guns to fight bigger guns. I guess I'm lucky to live in a country where nearly no street punk can pull a gun on me
That could be said of the whole PC world? Does 640k ring a bell anywhere? MS-DOS (dead in Windows Millenium?... could be)? Layers and layers of stuff to cover and aging (and awful) base?
I still have my Amiga, and with a 68060, at 50mhz, it sure feels faster than my 500Mhz Athlon, by the way. And it has a decent OS (much more efficient than Windoze and more ellegant and simple than UN*X)... it just doesn't play Quake (oops, I forgot. It does).