Slashdot can be expected to know that, for example, Tripod put bandwidth limits, so there's no point linking to them (they normally don't, presumably for this very reason)
They can also expect to link to amazon.com, microsoft.com, sun.com, etc, and be confident it can handle the/.'ing. If it's "Jimmy Bob Joe's Page about My New PC" then it's pretty certain to be unable to cope with the load.
It's not rocket science, and the/. editors claim to be techies - they can make some basic assumptions. Maybe even apologise to people when they inadvertently post a link to some poor sap's DSL line.
Admins need to understand their server and configure them to refuse requests when overloaded, so that they don't crash
Apache does that; I assume other decent web servers do, too.
That is, of course, exactly the problem - I remember a month or two ago,/. linked to a guy on a DSL modem - he had no chance! It's not "properly configure your web server" - the only way to get that server to server his family but not/. would be to accept the requests, check the HTTP_REFERER [sic], and allow/reject on that basis. It's actually more efficient, in most cases, just to serve the page, than to do all that extra testing.
I've got a site on a 128k link - it would stand no chance if/. linked to it! And if you think that "properly configuring" means "buying a fatter pipe", then why should I pay for one, just for/.'s benefit? It does me well enough for around 1000 visitors/day.
If/. wanted to link to my site, I'd be quite happy for them to mirror it for the day, with prior, explicit approval from myself.
If I was/.'ed, I would be unable to serve pages to innocent visitors coming in from Google or wherever, who quite possibly don't know or care about/. They just think "site's down" and go away. That gives them a bad impression of my site, which (if I was selling stuff) I could easily prove caused me financial loss. I'd just have to show that for every X visitors I get $Y in revenue, and that on the day I was/.'ed I only got $Z in revenue. I've lost $Y-$Z in that day, plus the customers who would have returned if they'd been able to get to the site when they first found it.
Though I only use Solaris and Linux, for those admins who use, say, Solaris, AIX, HP-UX, CDE is the ideal solution - it's the *Common* Desktop Environment. Maybe not the best, it's a compromise, but it's good enough for most uses, and works across *nix.
The question should not be, "Why should Sun use GNOME", but "Why hasn't CDE been ported to Linux?"
And when one small firm, uk.com, went down a few months ago, why did Nominet UK get all the flak?
I've not heard of Nominet being upset about uk.com - it confuses some users, but it's basically ripping off its own customers; no skin off Nominet.uk's nose.
F/OSS also has the benefit of not having to call itself "1.0" on release.
I am convinced that this helps coders and users - look at Phoenix 0.5 - it's the best browser I have ever seen, but I'll forgive it for crashing (twice since 0.2, IIRC) because it's not 1.0
Still the best browser on the planet, IMHO (and for what I want, of course).
It isn't uncommon for F/OSS to take the attitude "You can read the source, so why make a clean, documented API?" - at least with closed-source stuff (and well-written F/OSS) you get a clean, stable, and well-documented API.
This must be the 3rd Pro-Closed post I've made in 2 days - waargh! - but let's face it, Closed Source programmers *have* to do it, F/OSS programmers *can* do it, but it may be a case of "when I get around to it." Try hacking GnuCash if you don't believe me (failing that, try *installing* GnuCash on LFS - the dependancies just to install it are bad enough, let alone the haiku you've got to crack to get to actually edit the code.
I can't name names, but one proprietary UNIX source is beautiful - in terms of code quality, down to annotation and indentation, and that is only for internal / priveleged partner use only.
Personally, I'm on more of a shell-script level, myself, but one lesson I have learned, is that if you write it well, you can fob it off onto someone else when you're sick of it. Write it badly, and you're stuck with it for the rest of your life!
Re:I learned my lesson
on
Immortal Code
·
· Score: 1
My attitude is:
If I wrote it, I should take a copy of it.
If I'm very proud of it, I should document it well, and take copies of the code and the documentation. That is my work. The code itself belongs to my employer, but while I work for them, I have no problem with having a local copy on my PC (hey, I'm giving the company a free backup).
Should the terrible happen (I leave, or the company dies), I've got a record of what I've done. If the company dies, I'd feel bad just throwing the existing code into a new company, but re-using the ideas is not a bad thing.
A trivial example:
At one firm, I cut my Unix teeth. I wrote a bunch of shell scripts to make my life easier. I still have some of them (unfortunately, not all). They technically belong to that company, but if I hadn't felt the need to write them (for myself) the would not have existed, and would not have belonged to them (I'm talking about v. small scripts in ~/bin here, stuff like re-titling an Xterm window, though I've also got some copies of a script for rebuilding a [snip]UX kernel to various parameters (back in the day when [snip]UX kernels needed rebuilding, not just tuning).
[snip]UX does not even exist any more (at least, it was never Y2K certified, and you can't buy it), but if someone has managed to keep it going, and needs a kernel rebuild, I can do it for them. All I am using the hard disk for, is as an archive for my brain. I can't remember all the details involved (why should I bother?) but the script does remember it all. I'm just making allowances for my disability (I can't remember scripts I wrote 10 years ago) here, and using my own PC to assist me to overcome that disability.
If [snip]NX was still maintained and on sale, with a PS staff to support it, maybe I'd be taking money from those people, but it isn't.
Nobody has asked me for the service, anyway. One person did ask me if I had a tape of a slightly older OS, which I'm sure I did have, once, but couldn't find the goods when asked.
Re:Unbelievably depressing?
on
Immortal Code
·
· Score: 1
I write code for my company; if I leave the firm, or if they go down, the code is gone.
I would have no qualms in using the code as a demo of my ability (I wrote it) but couldn't use that code again.
Coding is just typing, though. My current project has taken about 2 years (of my time - it's a 1-man project, really) of which maybe 2 weeks is coding. The skill is in the research, knowing what to do, and how to do it. That knowledge is mine, it lives in my head. The code is trivia. Coding is typing. Knowing what to type is understanding.
OMG, this is starting to sound like Zen, but there's something in that, I suppose. The loop doesn't matter, it's why it's a loop that matters.
Doesn't the autoconf/automake stuff diminish this rule? In that Makefile is generated from Makefile.in? I must admit, I tend to build Makefile by hand.
I stopped editing other peoples' Makefiles once I spotted the importance of <TAB> (ie, within a day of encountering a Makefile)
I stand by your professional conduct.
If the "very large company" was a dot-bomb, though, would you have the same qualms?
I've been developing some software (only a few thousand lines, but contains a *ton* of work) for the past few years. If my employer went bust, I'd have no problem with showing that code to somebody else in this (interview) scenario. I wouldn't rush off and decide to GPL it just because it happens that I wrote it and I have a copy, but I'd certainly put it, its feature list and sample output in my portfolio if an interviewer asked what I'd done before.
What am I supposed to do? Say "I've done lots of cool stuff, but I can't tell you about it. Do I get the job?"
If I'd worked on, say, MS Works, I could say so, and depend upon the fact that people remember it. If I've worked for a company who never made it, I couldn't make those assumptions - I'd have to do the whole "the code is good; the business model was bad" thing. The only way to prove that (without refering to a product they know) is to show what I've done.
Of course, if I'd worked for IBM on some proprietary stuff and wanted to move to HP, there would be no excuse for showing HP any IBM code (whether released or not).
Good, Patriotic American
on
Superbowl XXXVII
·
· Score: -1, Troll
Fortunately, I'm none of the above.
Does this excuse me from avoiding a stupid game where men in silly suits chase each other around a field? Or does it mean I'm a terrorist?
Still looking for anything in this story which might count as "News for Nerds" or as "Stuff that matters", though.
Is this from the GwB school of "it's important in my country, it must be important to everyone" importance?
Think of it as a distraction from the pending war, which is a distraction from the fact that he has no idea how to run foreign policy.
No it's a distraction from the fact that he has no idea how to run domestic policy.
No US President has had any idea about foreign policy. Bush just has this special quality of "Oil, Oil, Voters (specially the Oil-buying ones), and Oil"
This capability, of course, implies the capability of Weapons of Mass Destruction (as opposed to the Weapons of Mass Peace which the USA currently holds).
I wouldn't choose lindows for myself, but it seems like a damn good way to get started to me.
But, of course, you are a person who does not percieve any problems with running as root.
It's not about the services you provide (many of them run as root on any *nix system, though this is changing)
It's about the user-level apps you use ("Download this file to/etc/passwd? [ YES | NO ]") - there tend to be more of these, and less well-scrutinised (how many people are concerned about Apache running as root, compared to how many are concerned about mIRC running as root?)
If I download and use some stupid app which tries to rewrite/etc/passwd, it will fail unless I'm logged in as root. The root account should *only* be used to run totally trusted applications. I am happy using Phoenix as a user - it's development, may crash, could even wipe my files (I download the binary, not the source) but it can't kill my machine, let alone make it a part of a DDoS attack.
Installing software as root should be done by a qualified system administrator only. Installing software in/home/luser/ is done at luser's own risk.
I do the speedtouchconf.sf.net script for configuring the Alcatel USB Modem under Linux. I've had about 20 different people contact me so far about it - mainly problems they've had with it, many of which are my mistakes.
The modem in question is obscure enough to get little enough interest, but of thost 20 (ish), two were LindowsOS users, who were missing the gcc and make utilites required. Being Debian-based, they could "apt-get install gcc" and were happy enough.
I don't like Lindows' attitude, but the OS seems - from this brief acquantance-by-proxy - enough like Debian not to matter, from a developer's point of view. I can give the same advice to a LindowsOS user as to a Debian user. At that level, it suits me fine.
One interpretation of these figures would be to say that 10% of Linux / Alcatel USB Modem users are on LindowsOS; I suspect this would be inaccurate, and that the LindowsOS users are maybe rather less used to Linux than users of other distro's, so the 10% is disproportionate because of that fact.
1. Install fancy moving pictures in a small box in peoples' houses
2. Control these boxes to show various sitcoms, films, cartoons, etc.
3. Show animated ads with this technology
4. Profit!
I don't enjoy giving telemarketers a hard time - much less technical support staff. Okay, telemarketers know what they're getting for their $4/hr, but tech support staff, the guys you have to phone to get a genuine problem dealt with, do not have the power to deal with the problem you have, so I can't make them take responsibility.
A current example (email, rather than phone) - my ISP allows SSI, including #exec. I had a little play around, found that "#exec niscat passwd.org_dir" and other useful things worked, on Friday night, so I sent them an email. On Tuesday morning (1am) I got a response, saying "Not being a server engineer myself, none of this happens to mean very much to me but I will certainly ensure this matter is passed on to our product team for investigation."
Being 4am Wednesday morning, I've had no response. I really don't want to lay in to poor old A Jupp - he's done what he can, but I can't lay in to anyone else - s/he's my interface to the people who *can* do something.
The people put on the phones are given all the responsibility with none of the power.
This is a really cheap way for companies to rip off their customers - "Here's our helpline!" [it's useless]. We've done our corporate responsibility; the frustrated customers speak to the frustrated support staff, and there is no mechanism for all this frustration to pass upwards to the decision makers, so a vicious circle is created, none of which affects the salary bonuses of the directors.
Slashdot Mirror
I wear this tshirt: http://www.thinkgeek.com/tshirts/frustrations/388b /
Saves a *lot* of time :)
Looks like sf.net are safe!
For example, Solaris comes with a Companion CD containing lots of Free Software.
Actually, the Solaris example is a good one - the kernel is SunOS, the whole operating system is Solaris. So SunOS <==> Linux, Solaris <==> GNU/Linux
They can also expect to link to amazon.com, microsoft.com, sun.com, etc, and be confident it can handle the /.'ing. If it's "Jimmy Bob Joe's Page about My New PC" then it's pretty certain to be unable to cope with the load.
It's not rocket science, and the /. editors claim to be techies - they can make some basic assumptions. Maybe even apologise to people when they inadvertently post a link to some poor sap's DSL line.
Apache does that; I assume other decent web servers do, too.
That is, of course, exactly the problem - I remember a month or two ago,
I've got a site on a 128k link - it would stand no chance if /. linked to it! And if you think that "properly configuring" means "buying a fatter pipe", then why should I pay for one, just for /.'s benefit? It does me well enough for around 1000 visitors/day. /. wanted to link to my site, I'd be quite happy for them to mirror it for the day, with prior, explicit approval from myself.
If
If I was /.'ed, I would be unable to serve pages to innocent visitors coming in from Google or wherever, who quite possibly don't know or care about /. They just think "site's down" and go away. That gives them a bad impression of my site, which (if I was selling stuff) I could easily prove caused me financial loss. I'd just have to show that for every X visitors I get $Y in revenue, and that on the day I was /.'ed I only got $Z in revenue. I've lost $Y-$Z in that day, plus the customers who would have returned if they'd been able to get to the site when they first found it.
http://www.faqs.org/faqs/unix-faq/shell/csh-whynot /
Wrong. OR is a boolean operator to Google. Check the "Advanced Search" link.
It's not perfect, but it's Good Enough.
I don't want bells and whistles (I'm used to IceWM, clean, simple, stays out of my way) just access to the machine.
The question should not be, "Why should Sun use GNOME", but "Why hasn't CDE been ported to Linux?"
Facts. Optional, on slashdot.
And when one small firm, uk.com, went down a few months ago, why did Nominet UK get all the flak?
I've not heard of Nominet being upset about uk.com - it confuses some users, but it's basically ripping off its own customers; no skin off Nominet.uk's nose.
I am convinced that this helps coders and users - look at Phoenix 0.5 - it's the best browser I have ever seen, but I'll forgive it for crashing (twice since 0.2, IIRC) because it's not 1.0
Still the best browser on the planet, IMHO (and for what I want, of course).
This must be the 3rd Pro-Closed post I've made in 2 days - waargh! - but let's face it, Closed Source programmers *have* to do it, F/OSS programmers *can* do it, but it may be a case of "when I get around to it." Try hacking GnuCash if you don't believe me (failing that, try *installing* GnuCash on LFS - the dependancies just to install it are bad enough, let alone the haiku you've got to crack to get to actually edit the code.
I can't name names, but one proprietary UNIX source is beautiful - in terms of code quality, down to annotation and indentation, and that is only for internal / priveleged partner use only.
Personally, I'm on more of a shell-script level, myself, but one lesson I have learned, is that if you write it well, you can fob it off onto someone else when you're sick of it. Write it badly, and you're stuck with it for the rest of your life!
If I wrote it, I should take a copy of it.
If I'm very proud of it, I should document it well, and take copies of the code and the documentation. That is my work. The code itself belongs to my employer, but while I work for them, I have no problem with having a local copy on my PC (hey, I'm giving the company a free backup).
Should the terrible happen (I leave, or the company dies), I've got a record of what I've done. If the company dies, I'd feel bad just throwing the existing code into a new company, but re-using the ideas is not a bad thing.
A trivial example:
At one firm, I cut my Unix teeth. I wrote a bunch of shell scripts to make my life easier. I still have some of them (unfortunately, not all). They technically belong to that company, but if I hadn't felt the need to write them (for myself) the would not have existed, and would not have belonged to them (I'm talking about v. small scripts in ~/bin here, stuff like re-titling an Xterm window, though I've also got some copies of a script for rebuilding a [snip]UX kernel to various parameters (back in the day when [snip]UX kernels needed rebuilding, not just tuning).
[snip]UX does not even exist any more (at least, it was never Y2K certified, and you can't buy it), but if someone has managed to keep it going, and needs a kernel rebuild, I can do it for them. All I am using the hard disk for, is as an archive for my brain. I can't remember all the details involved (why should I bother?) but the script does remember it all. I'm just making allowances for my disability (I can't remember scripts I wrote 10 years ago) here, and using my own PC to assist me to overcome that disability.
If [snip]NX was still maintained and on sale, with a PS staff to support it, maybe I'd be taking money from those people, but it isn't.
Nobody has asked me for the service, anyway. One person did ask me if I had a tape of a slightly older OS, which I'm sure I did have, once, but couldn't find the goods when asked.
I would have no qualms in using the code as a demo of my ability (I wrote it) but couldn't use that code again.
Coding is just typing, though. My current project has taken about 2 years (of my time - it's a 1-man project, really) of which maybe 2 weeks is coding. The skill is in the research, knowing what to do, and how to do it. That knowledge is mine, it lives in my head. The code is trivia. Coding is typing. Knowing what to type is understanding.
OMG, this is starting to sound like Zen, but there's something in that, I suppose. The loop doesn't matter, it's why it's a loop that matters.
I stopped editing other peoples' Makefiles once I spotted the importance of <TAB> (ie, within a day of encountering a Makefile)
Is it re-use? Is it theft? How much can you do in C? Honestly? Without dragging in libc and a bunch of other libraries, not a lot.
Ideas matter; coding is just typing. * it's 2am, I can't spell.
If the "very large company" was a dot-bomb, though, would you have the same qualms?
I've been developing some software (only a few thousand lines, but contains a *ton* of work) for the past few years. If my employer went bust, I'd have no problem with showing that code to somebody else in this (interview) scenario. I wouldn't rush off and decide to GPL it just because it happens that I wrote it and I have a copy, but I'd certainly put it, its feature list and sample output in my portfolio if an interviewer asked what I'd done before.
What am I supposed to do? Say "I've done lots of cool stuff, but I can't tell you about it. Do I get the job?"
If I'd worked on, say, MS Works, I could say so, and depend upon the fact that people remember it. If I've worked for a company who never made it, I couldn't make those assumptions - I'd have to do the whole "the code is good; the business model was bad" thing. The only way to prove that (without refering to a product they know) is to show what I've done.
Of course, if I'd worked for IBM on some proprietary stuff and wanted to move to HP, there would be no excuse for showing HP any IBM code (whether released or not).
Does this excuse me from avoiding a stupid game where men in silly suits chase each other around a field? Or does it mean I'm a terrorist?
Still looking for anything in this story which might count as "News for Nerds" or as "Stuff that matters", though.
Is this from the GwB school of "it's important in my country, it must be important to everyone" importance?
No it's a distraction from the fact that he has no idea how to run domestic policy.
No US President has had any idea about foreign policy. Bush just has this special quality of "Oil, Oil, Voters (specially the Oil-buying ones), and Oil"
This capability, of course, implies the capability of Weapons of Mass Destruction (as opposed to the Weapons of Mass Peace which the USA currently holds).
But, of course, you are a person who does not percieve any problems with running as root. /etc/passwd? [ YES | NO ]") - there tend to be more of these, and less well-scrutinised (how many people are concerned about Apache running as root, compared to how many are concerned about mIRC running as root?) /etc/passwd, it will fail unless I'm logged in as root. The root account should *only* be used to run totally trusted applications. I am happy using Phoenix as a user - it's development, may crash, could even wipe my files (I download the binary, not the source) but it can't kill my machine, let alone make it a part of a DDoS attack. /home/luser/ is done at luser's own risk.
It's not about the services you provide (many of them run as root on any *nix system, though this is changing)
It's about the user-level apps you use ("Download this file to
If I download and use some stupid app which tries to rewrite
Installing software as root should be done by a qualified system administrator only. Installing software in
I do the speedtouchconf.sf.net script for configuring the Alcatel USB Modem under Linux. I've had about 20 different people contact me so far about it - mainly problems they've had with it, many of which are my mistakes. The modem in question is obscure enough to get little enough interest, but of thost 20 (ish), two were LindowsOS users, who were missing the gcc and make utilites required. Being Debian-based, they could "apt-get install gcc" and were happy enough. I don't like Lindows' attitude, but the OS seems - from this brief acquantance-by-proxy - enough like Debian not to matter, from a developer's point of view. I can give the same advice to a LindowsOS user as to a Debian user. At that level, it suits me fine. One interpretation of these figures would be to say that 10% of Linux / Alcatel USB Modem users are on LindowsOS; I suspect this would be inaccurate, and that the LindowsOS users are maybe rather less used to Linux than users of other distro's, so the 10% is disproportionate because of that fact.
2. Control these boxes to show various sitcoms, films, cartoons, etc.
3. Show animated ads with this technology
4. Profit!
A current example (email, rather than phone) - my ISP allows SSI, including #exec. I had a little play around, found that "#exec niscat passwd.org_dir" and other useful things worked, on Friday night, so I sent them an email. On Tuesday morning (1am) I got a response, saying "Not being a server engineer myself, none of this happens to mean very much to me but I will certainly ensure this matter is passed on to our product team for investigation."
Being 4am Wednesday morning, I've had no response. I really don't want to lay in to poor old A Jupp - he's done what he can, but I can't lay in to anyone else - s/he's my interface to the people who *can* do something.
The people put on the phones are given all the responsibility with none of the power.
This is a really cheap way for companies to rip off their customers - "Here's our helpline!" [it's useless]. We've done our corporate responsibility; the frustrated customers speak to the frustrated support staff, and there is no mechanism for all this frustration to pass upwards to the decision makers, so a vicious circle is created, none of which affects the salary bonuses of the directors.