I have yet to see any good arguments why they can not ban sending out *thousands* of emails.
Mailing lists.
There is no central authority which counts total emails sent by a given individual or host. The closest that you could come to this is not accepting more than N emails from a given host in a given day.
Go work technical support at an ISP, if you've never done so already. Now imagine having to explain to customers that their grandmother's mail is probably getting rejected because "our mail system has already gotten 1500 emails from AOL today, so it's temporarily rejecting everything from them until sometime after midnight."
Given that spammers often use unreported open relays, you're probably not going to be able to know how many messages a spammer has sent through several open relays to thousands of hosts.
In essence the second reason why "they cannot ban sending out *thousands* of emails" is that there is no "they" to count emails sent.
If you can explain how this would be technically possible without restricting legitimate uses of email (including opt-in mailing lists, such as the Debian Weekly News) please feel free. It may be that you haven't thought through this idea very much, or you may have addressed these problems in your thoughts but omitted the details from your above post.
I know admittedly little about Knoppix, but last time I checked a local Debian mirror (~3 weeks ago), I didn't see anything but the same old boot disks.
At any rate, it'd be cheaper for the average company to hire a Debian developer full-time to act as support for any issues the company may come across than to pay Redhat hundreds of thousands a year. The company would win by saving money, and the community would benefit by having more full-time contributors to Debian.
Why do people create Debian installers and not contribute them back to the project?
Debian's distribution method is great in that to create a derivative distro, you can choose to only distribute your additions, and let Debian continue to manage the main repository. I don't understand why the Debian installer remains dated -- updating the installer would benefit anyone installing Debian, and would provide those who wish to create a Debian-based distro with a very nice start for their own installers.
I personally haven't used the Debian installer directly in quite a while. I know a friend recently (about 3 weeks ago, give or take) tried to install Debian on a ReiserFS root. However, for some reason he was never able to actually boot the machine, even from a rescue disk with ReiserFS support.
As my friend was 240 miles away at the time, I wasn't there to give it a shot directly, but I was on the phone with him during the installation, and he finally gave up and installed on an ext2 partition, then once the system was up and running with a fresh kernel I logged in and created ReiserFS/var,/usr,/home partitions.
I love Debian, once it's installed. However, going from zero knowledge to having a freshly installed system running on ReiserFS isn't something which the documentation covers well, even to the extent of stating that it's well-supported.
If I had a spare system with which to mess around using the installers, I'd be glad to submit a patch to the Debian installation guide which would explain clearly what is required to get a Debian system up and running using ReiserFS.
As it is, however, I consider the fact that someone who has installed Debian a few times in the past finds it harder to get a ReiserFS system running than an ext2 system given the available documentation to be a large bug. Yes, perhaps information is available _somewhere_, but if a quick Google search doesn't reveal something, and the Installation Guide doesn't cover it, there's definitely a problem somewhere. Either the Installation Guide should cover such information, or it should contain a large disclaimer explaining that it may not cover everything a new user would need to get the system installed to exactly meet his or her expectations.
I am a big Debian fan, and agree with your summary of the relevant issues.
For those unfamiliar with Debian, the parent poster is (presumably) referring to Debian's "stable" distribution, which is as close to "guaranteed to work, even after upgrades" as you're likely to get anywhere, Linux or otherwise.
Debian does really suffer from an archaic installer. I'm not referring to "lack of GUI" here -- I actually prefer a text-based installer -- but the installer's lack of ReiserFS or 2.4 support is irritating. My working theory regarding the lack of attention to the installer is the fact that most Debian developers rarely use it -- apt-get (and dpkg behind it) is all that you need to keep Debian running. I've used the same Debian installation for years with a minimum of cruft, whereas (as I understand it) keeping Redhat up to date requires using the installer when upgrades are released.
A Debian-based router I administer even survived my accidental deletion of the entire/var partition. Without backups of any form, I had the packaging system functional within an hour. Debian's "survivability" combined with the incremental nature of upgrades means that the installer is rarely used, especially by those most familiar with the distribution.
A few days ago when the Paul Graham article was posted to the frontpage, I was thinking about the fact that MS hasn't implemented Bayesian filtering (or any powerful filtering) into OE.
The three possibilities I came up with were:
MS wanted to give Hotmail/MSN a competitive edge over other ISPs and mail services.
MS didn't consider it worth the money to add Bayesian filtering to OE.
MS is using Hotmail as a testbed for various versions of filtering software; by making changes and observing user behaviour, they could determine whether people generally agreed with the filters, thus roughly gauging their effectiveness. Since changing source on a central server is a faster method of deploying updates than forcing users to require a new client, Hotmail is the perfect place to test new filtering schemes.
Honestly, I thought that either the first or the third was true. Here, it turns out that the first and the second are true.
I wonder what ISPs will be left to do? I suppose they'll either have to seek out a cheap/free mail client for Windows, or switch to webmail.
It might not be a great idea to trust a company who charges between $8 and $13 for a modem to write code to get a general-purpose CPU to emulate a DSP in kernel mode, yet people purchase $400 computers with HSP or PCTel modems and expect reliability.
Certainly there's a lot of separation possible between in-kernel code and userspace code; however, this doesn't mean that those writing the driver adhere to the safest policies, especially in the FPS-centric world of video drivers.
With most hardware I've seen, being willing to pay a little more for quality has always been well worth the returns. With video cards, this isn't always true, since you're usually paying for "sexy" features, like an extra few FPS, rather than a balance between performance and reliability.
In addition, since IE is meant to run code downloaded from the Internet, shouldn't it be written in such a way as to minimize the impact of unresponsive plugins?
Fuckin' A, man -- leave it to Michael to post drivel like this.
Seriously, it was more amusing browsing John Katz articles trying to figure out which bandwagon he was trying to pretend he knew enough to jump on this week. If there's a particularly unsubstantiated claim made on Slashdot's homepage, 9 times out of 10 it's something Michael posted.
Not necessarily. At least, not for "insider trading".
The most significant information was the lawsuit, which had been announced four days previously. Hell, Google probably had links to about 100 "Why SCO is probably full of shit" pages four days after the announcement.
Any time an investor hears about a decision made by a company in which they own stock, it is in their best interests to think about the likely short and long-term affects of that decision.
SCO made information public. Then SCO executives sold stock. SCO leaders may have fraud charges coming, but as a seemingly well-informed (much more so than I, at least) mentioned the charges have a much greater chance of coming from a class-action lawyer than from "your friends" at the SEC.
I do indent my code, but it's quite possible that I'm a fucking idiot nonetheless (which, I am aware, wouldn't conflict with your statement at all). I always wonder if and how Python manages to handle tabs of various sizes, though. Interpreting tabs as 3 spaces leads to formatting quite different than assuming that they represent 8 spaces.
Personally, I use vi, so they're all tab characters, but I've seen editors which make use of mixed whitespace characters.
Besides, it's easier to delay looking into Python by claiming to be horribly offended by syntactically significant whitespace than to start learning it and really want to know more but not have the time to really get into it, which sadly is my current relationship to LISP.
Being American leaves me at a disadvantage here, as I can't say I have a developed appreciation for the art of serving ale.
I've only had Guinness from bottles -- I avoid bars, since too many Americans consider bars a place to go slug back as much alcohol as possible, and don't really take the time to appreciate what they're drinking. Of course, bars try to capitalize on this idea by employing the most attractive bartenders they can. The more you drink, the easier it is to convince yourself to relax and enjoy the view while drinking yet more.
I've never made it to the UK, Ireland[1], or Germany, a fact which saddens me further. What I've seen of Europe has been beautiful, both the people[2] and the land.
At any rate, thanks much for the response.
Drink a pint for me.
[1] Am I correct in my understanding that none of Ireland is currently under British rule?
[2] Certain nasty fellows in France excepted, unfortunately -- my friend's grandfather had his wallet stolen on the subway in Paris. However, I did meet a few nice French people, and I felt more welcome in Spain than in my own home.
Slashdot Mirror
Go work technical support at an ISP, if you've never done so already. Now imagine having to explain to customers that their grandmother's mail is probably getting rejected because "our mail system has already gotten 1500 emails from AOL today, so it's temporarily rejecting everything from them until sometime after midnight."
Given that spammers often use unreported open relays, you're probably not going to be able to know how many messages a spammer has sent through several open relays to thousands of hosts.
In essence the second reason why "they cannot ban sending out *thousands* of emails" is that there is no "they" to count emails sent.
If you can explain how this would be technically possible without restricting legitimate uses of email (including opt-in mailing lists, such as the Debian Weekly News) please feel free. It may be that you haven't thought through this idea very much, or you may have addressed these problems in your thoughts but omitted the details from your above post.
Assume that my use of the word "or" in the above post was exclusive ;-)
I know admittedly little about Knoppix, but last time I checked a local Debian mirror (~3 weeks ago), I didn't see anything but the same old boot disks.
At any rate, it'd be cheaper for the average company to hire a Debian developer full-time to act as support for any issues the company may come across than to pay Redhat hundreds of thousands a year. The company would win by saving money, and the community would benefit by having more full-time contributors to Debian.
That's retarded. There's no reason something like that shouldn't be documented in the Installation Guide. :-(
Why do people create Debian installers and not contribute them back to the project?
Debian's distribution method is great in that to create a derivative distro, you can choose to only distribute your additions, and let Debian continue to manage the main repository. I don't understand why the Debian installer remains dated -- updating the installer would benefit anyone installing Debian, and would provide those who wish to create a Debian-based distro with a very nice start for their own installers.
I personally haven't used the Debian installer directly in quite a while. I know a friend recently (about 3 weeks ago, give or take) tried to install Debian on a ReiserFS root. However, for some reason he was never able to actually boot the machine, even from a rescue disk with ReiserFS support.
/var, /usr, /home partitions.
As my friend was 240 miles away at the time, I wasn't there to give it a shot directly, but I was on the phone with him during the installation, and he finally gave up and installed on an ext2 partition, then once the system was up and running with a fresh kernel I logged in and created ReiserFS
I love Debian, once it's installed. However, going from zero knowledge to having a freshly installed system running on ReiserFS isn't something which the documentation covers well, even to the extent of stating that it's well-supported.
If I had a spare system with which to mess around using the installers, I'd be glad to submit a patch to the Debian installation guide which would explain clearly what is required to get a Debian system up and running using ReiserFS.
As it is, however, I consider the fact that someone who has installed Debian a few times in the past finds it harder to get a ReiserFS system running than an ext2 system given the available documentation to be a large bug. Yes, perhaps information is available _somewhere_, but if a quick Google search doesn't reveal something, and the Installation Guide doesn't cover it, there's definitely a problem somewhere. Either the Installation Guide should cover such information, or it should contain a large disclaimer explaining that it may not cover everything a new user would need to get the system installed to exactly meet his or her expectations.
I am a big Debian fan, and agree with your summary of the relevant issues.
/var partition. Without backups of any form, I had the packaging system functional within an hour. Debian's "survivability" combined with the incremental nature of upgrades means that the installer is rarely used, especially by those most familiar with the distribution.
For those unfamiliar with Debian, the parent poster is (presumably) referring to Debian's "stable" distribution, which is as close to "guaranteed to work, even after upgrades" as you're likely to get anywhere, Linux or otherwise.
Debian does really suffer from an archaic installer. I'm not referring to "lack of GUI" here -- I actually prefer a text-based installer -- but the installer's lack of ReiserFS or 2.4 support is irritating. My working theory regarding the lack of attention to the installer is the fact that most Debian developers rarely use it -- apt-get (and dpkg behind it) is all that you need to keep Debian running. I've used the same Debian installation for years with a minimum of cruft, whereas (as I understand it) keeping Redhat up to date requires using the installer when upgrades are released.
A Debian-based router I administer even survived my accidental deletion of the entire
This gets "+5, Interesting"? He didn't even say who "we" actually is. "We" could be a Fortune 500 company, or "we" could be him and his pet fish.
A few days ago when the Paul Graham article was posted to the frontpage, I was thinking about the fact that MS hasn't implemented Bayesian filtering (or any powerful filtering) into OE.
The three possibilities I came up with were:
MS wanted to give Hotmail/MSN a competitive edge over other ISPs and mail services.
MS didn't consider it worth the money to add Bayesian filtering to OE.
MS is using Hotmail as a testbed for various versions of filtering software; by making changes and observing user behaviour, they could determine whether people generally agreed with the filters, thus roughly gauging their effectiveness. Since changing source on a central server is a faster method of deploying updates than forcing users to require a new client, Hotmail is the perfect place to test new filtering schemes.
Honestly, I thought that either the first or the third was true. Here, it turns out that the first and the second are true.
I wonder what ISPs will be left to do? I suppose they'll either have to seek out a cheap/free mail client for Windows, or switch to webmail.
Apparently you've never used your iBook with an iPod, then -- iPods do implement DRM.
You must not have support for MPEG4ish technologies, such as Quicktime, either -- it seems that Apple has been working on copy protection for MPEG4.
In other words, "you get what you pay for".
It might not be a great idea to trust a company who charges between $8 and $13 for a modem to write code to get a general-purpose CPU to emulate a DSP in kernel mode, yet people purchase $400 computers with HSP or PCTel modems and expect reliability.
Certainly there's a lot of separation possible between in-kernel code and userspace code; however, this doesn't mean that those writing the driver adhere to the safest policies, especially in the FPS-centric world of video drivers.
With most hardware I've seen, being willing to pay a little more for quality has always been well worth the returns. With video cards, this isn't always true, since you're usually paying for "sexy" features, like an extra few FPS, rather than a balance between performance and reliability.
In addition, since IE is meant to run code downloaded from the Internet, shouldn't it be written in such a way as to minimize the impact of unresponsive plugins?
Fuckin' A, man -- leave it to Michael to post drivel like this.
Seriously, it was more amusing browsing John Katz articles trying to figure out which bandwagon he was trying to pretend he knew enough to jump on this week. If there's a particularly unsubstantiated claim made on Slashdot's homepage, 9 times out of 10 it's something Michael posted.
I do use Vim.
Java? Ugh!
GnuCash actually uses Guile to implement a good bit of functionality in LISP, as I understand it.
A quarter of a million lines of Java code is something I really would rather not think about, though.
Ick ick ick.
Not necessarily. At least, not for "insider trading".
The most significant information was the lawsuit, which had been announced four days previously. Hell, Google probably had links to about 100 "Why SCO is probably full of shit" pages four days after the announcement.
Any time an investor hears about a decision made by a company in which they own stock, it is in their best interests to think about the likely short and long-term affects of that decision.
SCO made information public. Then SCO executives sold stock. SCO leaders may have fraud charges coming, but as a seemingly well-informed (much more so than I, at least) mentioned the charges have a much greater chance of coming from a class-action lawyer than from "your friends" at the SEC.
That assumption is about as realisting as buying a single share in McDonald's and expecting free Big Macs for life.
That's essentially my point, that RMS insists that GNU forms a good bit of the OS simply because it really does.
People like to rag on RMS and the GNU thing, but if you take away GNU, you really don't have much of an OS left.
Don't forget,
IBM shall take up the keyboard of mighty heft, and strike mine enemies until they are smitten.
I don't want to use a decent editor, I want to use Vi!
Now that I've managed to piss off as many people as possible, why Python?
I do indent my code, but it's quite possible that I'm a fucking idiot nonetheless (which, I am aware, wouldn't conflict with your statement at all). I always wonder if and how Python manages to handle tabs of various sizes, though. Interpreting tabs as 3 spaces leads to formatting quite different than assuming that they represent 8 spaces.
Personally, I use vi, so they're all tab characters, but I've seen editors which make use of mixed whitespace characters.
Besides, it's easier to delay looking into Python by claiming to be horribly offended by syntactically significant whitespace than to start learning it and really want to know more but not have the time to really get into it, which sadly is my current relationship to LISP.
On that note, whitespace? Are you nuts?!?
Being American leaves me at a disadvantage here, as I can't say I have a developed appreciation for the art of serving ale.
I've only had Guinness from bottles -- I avoid bars, since too many Americans consider bars a place to go slug back as much alcohol as possible, and don't really take the time to appreciate what they're drinking. Of course, bars try to capitalize on this idea by employing the most attractive bartenders they can. The more you drink, the easier it is to convince yourself to relax and enjoy the view while drinking yet more.
I've never made it to the UK, Ireland[1], or Germany, a fact which saddens me further. What I've seen of Europe has been beautiful, both the people[2] and the land.
At any rate, thanks much for the response.
Drink a pint for me.
[1] Am I correct in my understanding that none of Ireland is currently under British rule?
[2] Certain nasty fellows in France excepted, unfortunately -- my friend's grandfather had his wallet stolen on the subway in Paris. However, I did meet a few nice French people, and I felt more welcome in Spain than in my own home.
I completely agree that C isn't the right tool for many tasks. Two words: garbage collection!
;-)
But come on, whitespace? Are you serious?
It's funny because RMS claims that GNU wrote the libraries that help to make Linux a UNIX-like operating system!
Hahaha!