I didn't say it couldn't be done, just that it's not simple. With proper motivation (as exists in Oz), it certainly can be done... per-port prefix aggrigate netflow stats would do. Of course, that's an enormous amount of data to collect, filter, and process. Software does exist to do this -- ala "BGP bandwidth accounting". I know of places that use it network planning, but I don't know of any that monitor every CPE link.
Earthlink blocks for both dialup and broadband customers...
While I cannot speak to dialup (I've not used it lately), they certainly are not doing this for "broadband" (i.e. cablemodem) -- at least not everywhere:
(outbound) [cramer:ttyp0]dominion:~/[5:22pm]:telnet ms9.verisignmail.com 25 Trying 216.168.230.183... Connected to ms9.verisignmail.com. Escape character is '^]'. 220 ms9.verisignmail.com ESMTP Mirapoint 3.2.2-GA; Thu, 8 Jan 2004 17:45:30 -0500 (EST) EHLO foo 250-ms9.verisignmail.com Hello user-XXX.cable.mindspring.com [65.87.180.XXX], pleased to meet you 250-8BITMIME 250-SIZE 5000000 250-DSN 250-ETRN 250-AUTH LOGIN 250-AUTH=LOGIN 250 HELP quit 221 ms9.verisignmail.com closing connection Connection closed by foreign host.
(inbound) [jfbeam:pts/1{1}]gir:~/[5:50pm]:telnet 65.87.180.XXX 25 Trying 65.87.180.XXX... Connected to user-XXX.cable.mindspring.com (65.87.180.XXX). Escape character is '^]'. 220 dominion ESMTP Sendmail 8.8.5/8.8.5; Thu, 8 Jan 2004 17:29:25 -0500 HELO example.com 250 dominion Hello adsl-BLAH.sip.rdu.bellsouth.net [65.81.XXX.XXX], pleased to meet you quit 221 dominion closing connection Connection closed by foreign host.
Bandwidth limitations should only apply to backbone use, not local server use. But Dog only knows, that is to complicated for TW...
That's actually too complicated for almost everyone. It's easy to track the number of bits crossing a link. It's much, MUCH, more difficult to track where all those bits are going. Such accounting would require full, per-session, netflow collection. Everywhere. And that's a lot of data to collect and process. As an example, full netflow data amounted to ~700MB/hr, compressed, at my former employer.
That's all true. However, in this case, we have one company that did something stupid and now it's chewing on their ass. And they made their mistake in 1997 when there were already long standing standards for "infinity" -- namely MAX_INT (well, signed int anyway.) They blindly chose 2 billion. This was very likely just some random constant a programmer stuck in the code without thought or review. (I hate it when people do that.)
Re:Basement [Museums] - They're the Future!
on
Build Your Own NOC
·
· Score: 1
And two apple's (II+?) So where's the machines actually connected to that T1?
YYYY-MM-DD That way, any idiot can sort it. (sep. optional)
Re:For a real opensource NOC
on
Build Your Own NOC
·
· Score: 2, Interesting
(MRTG/RRD, OpenNMS) are mediocre to the point of unusability
I cannot speak to OpenNMS, although I am aware of it. However, MRTG is quite usable and valuable. No, it's not the best, most optimal traffic collection system out there, but it is simple, fast, and gets the jobs done when used appropriately. I've used it for many, many years. It does suck if you try to have one instance monitor thousands of interfaces on hundreds of devices, but more than one instance is perfectly functional. I suspect what you want is far beyond what MRTG was designed to provide. I'll admit, I'd like to have a database filled with millions of data points, but MRTG isn't designed to do that. (And at my previous job, monitoring all the ports I'd've liked to would've consumed a few hundred meg per year without archiving or consolidation.)
(FWIW, I know of one place that uses MRTG instead of HP OpenView, for which they paid $$$$$$, because MRTG is faster and simpler and runs on a 200$ PC.)
Homegrown applications are great... when the company will allow it. Many places simply do not want the responsibility or liability of creating and maintaining their own software. If it doesn't work correctly or fails, who do they have to blame but themselves? Plus, the people who wrote the app may not be there in a year thus creating a support issue. I've created a number of homegrown apps to deal with my job, but I'm the only one who completely understands them; when I'm no longer there, that's a problem. Additionally, let's be real here. Given the quality of commercial software, just how good do you expect internally developed software from one or two programmers (who may not understand the problem they are fixing) will be? The best stuff will be coming from the grunts who have to work with and fix stuff everyday -- shell, perl, tcl, etc. scripts born out of necessity. That stuff will not be "quality" nor will it make much sense to anyone other than the author. (I've been here way too many times.)
Don't forget, most of these reviews are by people who aren't qualified to be giving a review. He claims web start requires X when it doesn't (I don't know how to start it manaully without X, but the boot process does.) The fact that he could not get the network online is very telling -- no knowledge of how networking is configured, no knowledge of which network driver to use (ok, so the names don't aways mean much), and no understanding of how his "cable" network is setup (dhcp? pppoe?)
In short, solaris/x86 is not going to be found on the shelves at Wal-Mart. You need to know what you're doing when dealing with Solaris. (We call that training and experience where I'm from.) It's not designed or intended for inexperienced nuts.
Both work, but I wouldn't recommend it. However, linux on sparc does provide a number of things you'll never get with solaris... support for firewire drives, and drivers for almost any PCI card you can find. I'm really pissed off by the lack of firewire support from a founding member of IEEE 1394.
"Damned if you do, damned if you don't"... better threading in solaris with a horrible java vm, or the best java vm in linux with moderate threading support. I've never liked Solaris/x86 simply because it's always been a kudge.
Not really. Do they actually have the connectivity to receive 50kpps? Previous comments indicating nothing else on their network being effected discounts the possibility of ISP intervention (filter,/32 routes to null, etc.) -- in such cases, the client is powerless to do anything as they cannot place filters on the ISP's router.
I'd guess this is coming down from DirecTV. I don't think abs ever had S2 4.0 images (they aren't in my archive, but I haven't checked in months.) The problem here is 100% Tivo's fault... the S2/SA software contains everything necessary to drive a S2/Combo. And, I told tivo that was a stupid idea many years ago -- the first time I saw dss components on a standalone (I think during one of the betas.)
I'm a little surprised abs has lasted this long. Tivo acted to shutdown a number of other sites providing images. I'd really like to see them do something about all those nuts selling "upgrade" drives complete with the tivo software. (do you really think those people have a distribution license with tivo?)
for the record, those mfstool backups have helped many people bring their tivo back from the dead (harddrive.) I wish Tivo would setup some means for subscribers to fetch a "repair kit", but the software doesn't do any kind of system licensing and validation. (hince the "moron" threads.) Now that I think about it, it wouldn't be too difficult to build a repair system, even with crypto support (via the hardware assuming the SA FPGA is the same as the Combo's)
Have you ever bothered to ask those cobol programmers why they used only two digits instead of four? They didn't do it "just because". There was a conscious design decision... four digits take twice the space. And memory/storage was FUCKING EXPENSIVE and processing was SLOW. Every byte and every instruction added up. Today, we have 3GHz processors and GB's of RAM in cheap desktop PCs, so no one cares about writing anything tight, fast, or efficient.
What examples do you need? Have you been under a rock? Newer versions of software should be getting better -- bugs get fixed, sections of code get better optimized, etc. However, it's only gotten worse -- bugs aren't fixed, code becomes less efficient and more buggy, bloat and feature creep make bloody messes. As an example, look at Microsoft OSes... each subsequent version has become larger, slower, and introduced a great many bugs. Add to that the Microsoft Office suite of applications... what 5% to 15% slower than previous versions. And just to be fair, even Linux is not immune to "bloat".../proc ever more poluted with stuff formated for human reading when humans are never the one's read it (and don't give me that "for scripts" BS.) Tried running and/or compiling a modern kernel on old hardware (386 or 486)? The kernel is a lot larger and slower using the same configured options. Shall I continue?
And just how many VB applications have you purchased and used? I don't even have an warez'd VB crap. (other than VB itself.)
...
nobody writes low-level code anymore...
Very few do. They are called "embedded programmers" these days. And guess what, they don't even write at the low-level anymore! (embeded java anyone.) There aren't "thousands" of people writing gcc or linux or freebsd. There are thousands of "contributors". There's a big difference -- very few could make significant progress with the contributions of others before them. (could you write a compiler or OS from scratch all on you own? I thought not.)
Ok, you're over generalizing... java and "www" are not synonymous. Yes, in the very beginning, java was a toy for playing inside a web browser. Today, entire applications are written in java far, far removed from any browser.
You're describing, basically, a chicken-and-the-egg problem. There are alot of "coders" in the world because many thought there was money in it. I've seen too many CS students with next to zero interest in computers and programming -- they went into CS for the money because it's "easy". (vs. most other well paying jobs.) Those are the people for whom it's just a job. They have border-line skills and zero motivation. They are the burger flippers of the software industry. And it shows... when their "burger" comes out a solid chunk of carbon, they don't care. And management doesn't care either until the customers complain. (and sometimes not even then.) Now that the industry is flooded with low-talent, finding the clueful is difficult and costly. So we're faced with apathy... just hire the stupid because it's cheaper and faster than finding real talent. (I happen to work with a bunch of highly talented people -- despite their years of java poisoning:-))
less seasoned practisioners of the art
Umm... these people are not artisans. And I wouldn't use the word "practisioner" either. Programming is an art. One has to recognize that to become a programmer. And an understand of the basics (the history, theory, and structure of programming) is required to become a good programmer. I'm not saying one has to understand the silicon, but there needs to be some knowledge of how the low levels work -- it's like knowing how to change a tire and check the oil.
You're confusing "progress" with "stupid". One would assume programmers would become smarter in the fullness of time like doctors and scientists. Instead, we've crossed that apex and are speeding headlong towards dumb-as-dirt. I like to pick on java, but it's just an example. If it hadn't cought on, something like would have.
Just to touch on modern medicine because it's a pet peav... the purpose of medical science was to improve the quality of life. That has morphed into an imperitive to keep every damn thing in the universe alive forever, unless it pisses us off and then we shoot it. As such, the principals of survival of the fitest, and natural selection have been suplanted by modern medicine (spured by greed.) Medicine is damaging the human gene pool in alarming ways... all because we can[1] and people have the money to pay for it. Humans have 13 (known) lethal gene combinations; modern medicine can override every damn one of them, and genes that normally would remove themselves from the pool replicate instead. In many respects, a lot of people should've been allowed to die or not exist.
crypt() uses a 56bit key (it's "DES") and a 12bit salt. So there are 2^56 x 4096, or 2^68, possible hash inputs and outputs (minus a handfull of collisions.)
That's a total of, roughly, 295,147,905,179,352,825,856 (295 billion billion) possibilities. Personally, I don't have a few petabytes sitting idle for the purposes of lookups. (however, I've dreamt of doing this for nearly 15 years... to be able to choose what one's password will be in the visible password file.)
The installer is confusing and cryptic and a million years behind...
At least it installs EXACTLY what you've told it to install and doesn't take hours of forward/back to deal with dependencies. I have long bitched at redhat for installing shit I intentionally tell it not to install (like, kudzu, lilo...) See, RH has a list of shit that it will always install and you have no say (short of building your own CDs.)
I've used RedHat, Debian, SuSE, and Mandrake ('tho I threw it away instantly), and even Gentoo. (Well, Slackware and SLS, too, for that matter.) Redhat has been evolving towards "brainless installation" for several years. This is a requirement for the masses as very few people know how to build a car, as you put it. Mandrake takes this to the "mindless zombie" absurdity -- which is why I hate it. Gentoo is just a pain in the ass. SuSE is ok, but doesn't have the breadth of debian and I have zero need for localization for 47 different languages. Which brought me to debian...
Despite the age of the "stable" version, and relatively complex installer, I like debian. It installs exactly what I tell it to. I don't have to hunt for and download rpms. They don't drop arch's. Etc.
Ignore what your users want and they'll go away
If they go away, they aren't "our users". And as your complaints are against the installer, those users would never become debian users -- they'll give up long before a working system is setup. The debian installer may be complex, but it works. And it's obviously not so complex that people cannot use it because there are a lot of people using debian (and it's been around for a decade.)
And most "new users" go with RedHat, mostly because of marketing. People outside Linux rarely make a distinction between "Linux" and "RedHat". Substitute SuSE for RedHat in the EU.
but then most DVD players will only support dual layer
Wrong. ALL DVD players are required to read single and dual layer media to be labeled as a DVD player and carry the officially licensed logo. The reason for 99% of the market being dual layer is simply to reduce piracy -- if you cannot burn 8G per disc, then you obviously cannot duplicate the movie, right. Of course, to actually fill the space, movie houses have to load a bunch of junk and encode the movie at unnecessarily high bit rates.
Reading a multi-layer disc is certainly not a simple task. However, we've gotten pretty good at it over the years. Writing a multi-layer disc is quite difficult. And researchers have only recently devised a scheme to do it at all -- don't expect the first generation technology to be fast, stable, or cheap.
Well, yes and no... DVD+R/RW discs look more like dual layer DVD-ROMs, but are only one layer. Some (most?) old DVD players won't be able to read those things no matter what the booktype is. My "ancient" Sony changer can almost read it as a DVD+RW, but sees a "bad disc" if I mark it as a DVD-ROM -- it looks for the second layer and fails to find it. (If I were so inclined, I can go to the diag settings and tweak the power levels to get it to read the +RW, but it might not be able to read -ROMs any more.)
Disregarding booktype, recordable DVDs are harder to read. My new 20$ KLH DVD player (even cheaper crap than Apex:-)) can read DVD+-R/RW, but not always at 1x+ *sigh* The old Apex players had an actual IDE DVD-ROM in there making this sort of thing moot. (just put the DVD burner in it.)
Are people actually posting DVDs in there now? (eg. not f***ing Divx copies of DVDs.) And good luck finding a USENET server with the group intact... I swear some people are expiring that group three times a day. (and web usenet tends to corrupt the group 30x an hour -- I actually suggested suing their software vendor (highwinds?).)
Bull shit. 20 years ago, today's "modern programmers" would've been executed for the crap they write. Very few of those called programmers today have even heard of a clue much less posess one. Things like java have poluted the world by making everything think they can program. The more intellegence placed in the compiler and language, translates to less in the programmer. In a few decades, society will come crubming down for lack of someone smart enough to write a compiler or VM.
Slashdot Mirror
I didn't say it couldn't be done, just that it's not simple. With proper motivation (as exists in Oz), it certainly can be done... per-port prefix aggrigate netflow stats would do. Of course, that's an enormous amount of data to collect, filter, and process. Software does exist to do this -- ala "BGP bandwidth accounting". I know of places that use it network planning, but I don't know of any that monitor every CPE link.
- ...
- Earthlink blocks for both dialup and broadband customers...
While I cannot speak to dialup (I've not used it lately), they certainly are not doing this for "broadband" (i.e. cablemodem) -- at least not everywhere:(outbound)
[cramer:ttyp0]dominion:~/[5:22pm]:telnet ms9.verisignmail.com 25
Trying 216.168.230.183...
Connected to ms9.verisignmail.com.
Escape character is '^]'.
220 ms9.verisignmail.com ESMTP Mirapoint 3.2.2-GA; Thu, 8 Jan 2004 17:45:30 -0500 (EST)
EHLO foo
250-ms9.verisignmail.com Hello user-XXX.cable.mindspring.com [65.87.180.XXX], pleased to meet you
250-8BITMIME
250-SIZE 5000000
250-DSN
250-ETRN
250-AUTH LOGIN
250-AUTH=LOGIN
250 HELP
quit
221 ms9.verisignmail.com closing connection
Connection closed by foreign host.
(inbound)
[jfbeam:pts/1{1}]gir:~/[5:50pm]:telnet 65.87.180.XXX 25
Trying 65.87.180.XXX...
Connected to user-XXX.cable.mindspring.com (65.87.180.XXX).
Escape character is '^]'.
220 dominion ESMTP Sendmail 8.8.5/8.8.5; Thu, 8 Jan 2004 17:29:25 -0500
HELO example.com
250 dominion Hello adsl-BLAH.sip.rdu.bellsouth.net [65.81.XXX.XXX], pleased to meet you
quit
221 dominion closing connection
Connection closed by foreign host.
- ...
- better than anyone else is offering us...
Go talk to Williams Communications (WilTel, whatever.) They are the best connected, cheapest dudes around.- Bandwidth limitations should only apply to backbone use, not local server use. But Dog only knows, that is to complicated for TW...
That's actually too complicated for almost everyone. It's easy to track the number of bits crossing a link. It's much, MUCH, more difficult to track where all those bits are going. Such accounting would require full, per-session, netflow collection. Everywhere. And that's a lot of data to collect and process. As an example, full netflow data amounted to ~700MB/hr, compressed, at my former employer.traceroute
RoadRunner has their own USENET servers as do Mindspring and/or Earthlink.
#include <stdio.h>
#include <time.h>
int main(void) {
time((void*)0x7fffffff);
}
[jfbeam:pts/2{2}]gir:~/[12:05pm]:gcc -o tt tt.c
[jfbeam:pts/2{2}]gir:~/[12:05pm]:./tt
Segm
That's all true. However, in this case, we have one company that did something stupid and now it's chewing on their ass. And they made their mistake in 1997 when there were already long standing standards for "infinity" -- namely MAX_INT (well, signed int anyway.) They blindly chose 2 billion. This was very likely just some random constant a programmer stuck in the code without thought or review. (I hate it when people do that.)
And two apple's (II+?) So where's the machines actually connected to that T1?
YYYY-MM-DD That way, any idiot can sort it. (sep. optional)
- (MRTG/RRD, OpenNMS) are mediocre to the point of unusability
I cannot speak to OpenNMS, although I am aware of it. However, MRTG is quite usable and valuable. No, it's not the best, most optimal traffic collection system out there, but it is simple, fast, and gets the jobs done when used appropriately. I've used it for many, many years. It does suck if you try to have one instance monitor thousands of interfaces on hundreds of devices, but more than one instance is perfectly functional. I suspect what you want is far beyond what MRTG was designed to provide. I'll admit, I'd like to have a database filled with millions of data points, but MRTG isn't designed to do that. (And at my previous job, monitoring all the ports I'd've liked to would've consumed a few hundred meg per year without archiving or consolidation.)(FWIW, I know of one place that uses MRTG instead of HP OpenView, for which they paid $$$$$$, because MRTG is faster and simpler and runs on a 200$ PC.)
Homegrown applications are great... when the company will allow it. Many places simply do not want the responsibility or liability of creating and maintaining their own software. If it doesn't work correctly or fails, who do they have to blame but themselves? Plus, the people who wrote the app may not be there in a year thus creating a support issue. I've created a number of homegrown apps to deal with my job, but I'm the only one who completely understands them; when I'm no longer there, that's a problem. Additionally, let's be real here. Given the quality of commercial software, just how good do you expect internally developed software from one or two programmers (who may not understand the problem they are fixing) will be? The best stuff will be coming from the grunts who have to work with and fix stuff everyday -- shell, perl, tcl, etc. scripts born out of necessity. That stuff will not be "quality" nor will it make much sense to anyone other than the author. (I've been here way too many times.)
Don't forget, most of these reviews are by people who aren't qualified to be giving a review. He claims web start requires X when it doesn't (I don't know how to start it manaully without X, but the boot process does.) The fact that he could not get the network online is very telling -- no knowledge of how networking is configured, no knowledge of which network driver to use (ok, so the names don't aways mean much), and no understanding of how his "cable" network is setup (dhcp? pppoe?)
In short, solaris/x86 is not going to be found on the shelves at Wal-Mart. You need to know what you're doing when dealing with Solaris. (We call that training and experience where I'm from.) It's not designed or intended for inexperienced nuts.
- Solaris on x86 is like linux on sparc...
Both work, but I wouldn't recommend it. However, linux on sparc does provide a number of things you'll never get with solaris... support for firewire drives, and drivers for almost any PCI card you can find. I'm really pissed off by the lack of firewire support from a founding member of IEEE 1394."Damned if you do, damned if you don't"... better threading in solaris with a horrible java vm, or the best java vm in linux with moderate threading support. I've never liked Solaris/x86 simply because it's always been a kudge.
Not really. Do they actually have the connectivity to receive 50kpps? Previous comments indicating nothing else on their network being effected discounts the possibility of ISP intervention (filter, /32 routes to null, etc.) -- in such cases, the client is powerless to do anything as they cannot place filters on the ISP's router.
I'd guess this is coming down from DirecTV. I don't think abs ever had S2 4.0 images (they aren't in my archive, but I haven't checked in months.) The problem here is 100% Tivo's fault... the S2/SA software contains everything necessary to drive a S2/Combo. And, I told tivo that was a stupid idea many years ago -- the first time I saw dss components on a standalone (I think during one of the betas.)
I'm a little surprised abs has lasted this long. Tivo acted to shutdown a number of other sites providing images. I'd really like to see them do something about all those nuts selling "upgrade" drives complete with the tivo software. (do you really think those people have a distribution license with tivo?)
for the record, those mfstool backups have helped many people bring their tivo back from the dead (harddrive.) I wish Tivo would setup some means for subscribers to fetch a "repair kit", but the software doesn't do any kind of system licensing and validation. (hince the "moron" threads.) Now that I think about it, it wouldn't be too difficult to build a repair system, even with crypto support (via the hardware assuming the SA FPGA is the same as the Combo's)
The Tivo SA's had a 90 day "manufacturer" warranty. The Hughes S2 DTivo originally came with a 1 year warranty; I don't know if they still do.
What examples do you need? Have you been under a rock? Newer versions of software should be getting better -- bugs get fixed, sections of code get better optimized, etc. However, it's only gotten worse -- bugs aren't fixed, code becomes less efficient and more buggy, bloat and feature creep make bloody messes. As an example, look at Microsoft OSes... each subsequent version has become larger, slower, and introduced a great many bugs. Add to that the Microsoft Office suite of applications... what 5% to 15% slower than previous versions. And just to be fair, even Linux is not immune to "bloat"...
And just how many VB applications have you purchased and used? I don't even have an warez'd VB crap. (other than VB itself.)
- ...
- nobody writes low-level code anymore...
Very few do. They are called "embedded programmers" these days. And guess what, they don't even write at the low-level anymore! (embeded java anyone.) There aren't "thousands" of people writing gcc or linux or freebsd. There are thousands of "contributors". There's a big difference -- very few could make significant progress with the contributions of others before them. (could you write a compiler or OS from scratch all on you own? I thought not.)You're describing, basically, a chicken-and-the-egg problem. There are alot of "coders" in the world because many thought there was money in it. I've seen too many CS students with next to zero interest in computers and programming -- they went into CS for the money because it's "easy". (vs. most other well paying jobs.) Those are the people for whom it's just a job. They have border-line skills and zero motivation. They are the burger flippers of the software industry. And it shows... when their "burger" comes out a solid chunk of carbon, they don't care. And management doesn't care either until the customers complain. (and sometimes not even then.) Now that the industry is flooded with low-talent, finding the clueful is difficult and costly. So we're faced with apathy... just hire the stupid because it's cheaper and faster than finding real talent. (I happen to work with a bunch of highly talented people -- despite their years of java poisoning
- less seasoned practisioners of the art
Umm... these people are not artisans. And I wouldn't use the word "practisioner" either. Programming is an art. One has to recognize that to become a programmer. And an understand of the basics (the history, theory, and structure of programming) is required to become a good programmer. I'm not saying one has to understand the silicon, but there needs to be some knowledge of how the low levels work -- it's like knowing how to change a tire and check the oil.You're confusing "progress" with "stupid". One would assume programmers would become smarter in the fullness of time like doctors and scientists. Instead, we've crossed that apex and are speeding headlong towards dumb-as-dirt. I like to pick on java, but it's just an example. If it hadn't cought on, something like would have.
Just to touch on modern medicine because it's a pet peav... the purpose of medical science was to improve the quality of life. That has morphed into an imperitive to keep every damn thing in the universe alive forever, unless it pisses us off and then we shoot it. As such, the principals of survival of the fitest, and natural selection have been suplanted by modern medicine (spured by greed.) Medicine is damaging the human gene pool in alarming ways... all because we can[1] and people have the money to pay for it. Humans have 13 (known) lethal gene combinations; modern medicine can override every damn one of them, and genes that normally would remove themselves from the pool replicate instead. In many respects, a lot of people should've been allowed to die or not exist.
[1] See also: Technological Imperative
crypt() uses a 56bit key (it's "DES") and a 12bit salt. So there are 2^56 x 4096, or 2^68, possible hash inputs and outputs (minus a handfull of collisions.)
That's a total of, roughly, 295,147,905,179,352,825,856 (295 billion billion) possibilities. Personally, I don't have a few petabytes sitting idle for the purposes of lookups. (however, I've dreamt of doing this for nearly 15 years... to be able to choose what one's password will be in the visible password file.)
- The installer is confusing and cryptic and a million years behind...
At least it installs EXACTLY what you've told it to install and doesn't take hours of forward/back to deal with dependencies. I have long bitched at redhat for installing shit I intentionally tell it not to install (like, kudzu, lilo...) See, RH has a list of shit that it will always install and you have no say (short of building your own CDs.)I've used RedHat, Debian, SuSE, and Mandrake ('tho I threw it away instantly), and even Gentoo. (Well, Slackware and SLS, too, for that matter.) Redhat has been evolving towards "brainless installation" for several years. This is a requirement for the masses as very few people know how to build a car, as you put it. Mandrake takes this to the "mindless zombie" absurdity -- which is why I hate it. Gentoo is just a pain in the ass. SuSE is ok, but doesn't have the breadth of debian and I have zero need for localization for 47 different languages. Which brought me to debian...
Despite the age of the "stable" version, and relatively complex installer, I like debian. It installs exactly what I tell it to. I don't have to hunt for and download rpms. They don't drop arch's. Etc.
- Ignore what your users want and they'll go away
If they go away, they aren't "our users". And as your complaints are against the installer, those users would never become debian users -- they'll give up long before a working system is setup. The debian installer may be complex, but it works. And it's obviously not so complex that people cannot use it because there are a lot of people using debian (and it's been around for a decade.)And most "new users" go with RedHat, mostly because of marketing. People outside Linux rarely make a distinction between "Linux" and "RedHat". Substitute SuSE for RedHat in the EU.
- but then most DVD players will only support dual layer
Wrong. ALL DVD players are required to read single and dual layer media to be labeled as a DVD player and carry the officially licensed logo. The reason for 99% of the market being dual layer is simply to reduce piracy -- if you cannot burn 8G per disc, then you obviously cannot duplicate the movie, right. Of course, to actually fill the space, movie houses have to load a bunch of junk and encode the movie at unnecessarily high bit rates.Reading a multi-layer disc is certainly not a simple task. However, we've gotten pretty good at it over the years. Writing a multi-layer disc is quite difficult. And researchers have only recently devised a scheme to do it at all -- don't expect the first generation technology to be fast, stable, or cheap.
Well, yes and no... DVD+R/RW discs look more like dual layer DVD-ROMs, but are only one layer. Some (most?) old DVD players won't be able to read those things no matter what the booktype is. My "ancient" Sony changer can almost read it as a DVD+RW, but sees a "bad disc" if I mark it as a DVD-ROM -- it looks for the second layer and fails to find it. (If I were so inclined, I can go to the diag settings and tweak the power levels to get it to read the +RW, but it might not be able to read -ROMs any more.)
:-)) can read DVD+-R/RW, but not always at 1x+ *sigh* The old Apex players had an actual IDE DVD-ROM in there making this sort of thing moot. (just put the DVD burner in it.)
Disregarding booktype, recordable DVDs are harder to read. My new 20$ KLH DVD player (even cheaper crap than Apex
- see ya all on alt.binaries.dvdr
Are people actually posting DVDs in there now? (eg. not f***ing Divx copies of DVDs.) And good luck finding a USENET server with the group intact... I swear some people are expiring that group three times a day. (and web usenet tends to corrupt the group 30x an hour -- I actually suggested suing their software vendor (highwinds?).)Bull shit. 20 years ago, today's "modern programmers" would've been executed for the crap they write. Very few of those called programmers today have even heard of a clue much less posess one. Things like java have poluted the world by making everything think they can program. The more intellegence placed in the compiler and language, translates to less in the programmer. In a few decades, society will come crubming down for lack of someone smart enough to write a compiler or VM.
The e100 driver is part of the mainline kernel. At issue are proprietary software products -- like realplayer, and motif, and things like that.