Just a few years back I helped one of my friends put a bigger hard drive into a TiVo that was running an incredibly old kernel... I think it was in the 2.0 era, or very close to it (there's a possibility I'm wrong here, and the information was written to a corrupted part of my brain).
I believe (though I have no empirical evidence at hand to prove it) the reason swap should be 2x the size of your total amount of RAM:
1.) single program chews up all of your available memory
2.) kernel swaps all of that out for more important processes
3.) the new processes chew up all of the available system memory
4.) the original process(es) become important: kernel swaps all RAM out (making swap now 2x the amount of available RAM), pulling the original swapped memory back into physical memory
I'm pretty sure it's something like that. If there's not enough swap to handle this process (assuming it can't just create swap files on-the-fly), the kernel goes into Rambo mode and starts killing processes that are taking up too many resources. Trust me, nobody wants to see the kernel become Rambo (I know there's a pun in there somewhere about the "colonel" in the Rambo flicks).
It seems more efficient. Except trees don't use electricity, and therefore don't contribute to the original problem, pollution, etc. Oh, and did anybody remember that trees produce oxygen, whereas these machines most decidedly do not? That seems like a rather important detail.
...riddle me this: when you get up tomorrow and go to sign in to GMail and all you get is a single page saying "Due to financial constraints and an inability to derive revenue from GMail to pay our bills, we have regretfully been forced to terminate the service.", where are your game serial numbers now and how do you plan on getting at them?
Keep a copy of all your email through a mail client. Seems pretty simple...?
That reminds me of a silly exploit on one of the hosted servers that I managed at an ISP (when reading, keep in mind I was a noob sysadmin). Through some pretty simple port scanning, I noticed a hosted server had a very odd port number open. I did a simple telnet to that port on the server, and was shocked to find I got a bash prompt with root-level permissions... keep in mind this server was publicly accessible (public IP).
I immediately notified the company that owned the server and told them to close the hole (and find how the attacker got in) immediately so other servers didn't get attacked by their rather beefy server. Hours went by with no response to my insistent calls and emails. I logged into their server and echoed stuff like "FIX YOUR HOLE" to their/dev/tty's (really annoying, BTW), but they did nothing.
I finally got tired of their inaction. I logged into their server through the hole (since I didn't have root/sudo on their server) and poked around for a few minutes to find where the hole had been created. It took a few minutes--I had no path set, so I had to use absolute paths to programs/files that I was unfamiliar with--but I found the problem was a silly entry in their inetd.conf file (or something like that). I removed it, restarted inetd, and tested that the hole was indeed plugged.
I found it absolutely amazing that they failed to act at all. On the rare occasions that an exploit was found on my server, it was fixed within a few minutes/hours of finding out (depending on how hard it was to fix and my accessibility to said server). I guess some people just don't care as long as their server is still running.
I was going to post this as a response to TheLink's post. I worked on the administrative section of a website for many years. It was built to have a single user account selected at all times, and assumed that they would only have a single window open and would only use my links (not the back button).
After a very short period of time, I found they completely ignored my requests. They would POST a search query, go to a result, then go back (re-POSTing the query). It was fairly intensive on the server, so I learned the trick of caching the results & redirecting after a POST (without printing anything to the screen). It's amazing how many intelligent web developers told me it was a stupid idea, even after explaining it... since I was in a position of authority, I eventually just forced them to do it my way, even though they somehow didn't understand the security (and user-friendliness) behind the concept.
I would guess the reason behind tying the tech support experience to not hiring him is because either
a.) he didn't seem to have potential for anything else, or
b.) something didn't strike 'em right in the interview, so they had to come up with an excuse.
I'd say the problem is mostly with the interviewing skills. Personally, I've had about an 80% success rate with interviews as long as I actually got a face-to-face. My first computer experience was working for a Gateway call center, which got me a position as tech support for an ISP: they'd lost their sysadmins prior to me coming in, so I showed interest in doing some of those duties and read all the appropriate books... in a few months I was managing the webserver (80 hosted websites). by the end of a year, I was formally the system administrator. By the end of another year, I added web development to that when the need arose. I was the lead programmer and eventually the project manager.
The above experience is great talking material at interviews. Being able to explain a journey from a wannabe geek in school ("programming" in BASIC) to tech support to sysadmin to web developer to lead programmer/project manager shows prospective employers that I have ambition and initiative.
Moral of the story: aim high. Even tech support positions can lead to much larger opportunities if only you look for them and grab on when the opportunity presents itself. Don't assume that you'll be handed anything on a silver platter (when promotions are available, usually the people that get said promotions are the one that apply for them).
A small company that I was employed with for about 8 years doing web & software development, we made it a point to find students that were taking any sort of "geek" class, especially those in the helpdesk. When we (I) interviewed them, I looked for a couple of important qualities:
* Real world experience (building perl scripts to handle problems/automate stuff, etc)
* Actually wrote a script outside of school (see above)
* Good grasp of the Internet and its fundamentals (i.e. they know what a form is & the difference between frontend vs. backend systems)
* Did NOT LIE about their experience or lack thereof
* Weren't afraid to say "I don't know"
That last one is very important. Whenever I encountered somebody that fluffed their resume by saying things like "CMS Experience" which was directly tied into a terribly laid-out web page on MySpace, I'd pretty quickly red-flag that person. Now that I'm a contractor, I've found that people actually respect me more when I admit to not knowing something. Nobody likes to work with somebody that knows everything--especially when they don't know what they're talking about.
Building tests for your code proves the code "is correct" (read: "works properly") and bug-free to the extent the testing covers. When new bugs are discovered, tests can be added/modified to check for that, proving your code is still good.
The process of "proving your code is correct," to me, involves lots of face-to-face see-I-told-ya-so's, and displaying how it works. Tests allow you to say with authority, "I wrote 4,322 tests to make sure the system works the way it's supposed to. If you got an error message without the system crashing, you did something wrong."
AMEN! Code that means exactly what it says is key to sanity. Consider:
$UI->update_section('message_box', $myError);
versus:
## Update the "message_box" part of the user interface with the error... $obj->$callMethod($tName, $data);
The first one is pretty straight forward. The second one is so irritating to read that one is more likely to lead the reader into assuming that reading the comments is more important than reading the code (remember: assuming makes an @ss out of "u" and "ming"... and that's one pissed-of Chinaman, I'll tell ya).
Commenting code is good, as long as the comments are useful. If your comments are misleading, then their presence is much more damaging than if they hadn't been there at all. I used to comment all my code, even if the code's purpose was very obvious, then I got bitten by it:
I coded something a long time ago that needed to be fixed. I looked at the comments around where the problem occurred, and made a minor fix. Later, it seemed the bug had mutated into a different form, and it all went back to the afore-mentioned block of code I'd fixed. I read the comments and fixed accordingly... which completely broke the system. When I had made the first fix, I didn't update the comment to explain the updated functionality, so the comment was completely misleading.
This goes along with "naming conventions." I put this in quotes because it can be more than just the code. For instance, if the front end changed "todos" into "tasks", make sure the code IMMEDIATELY uses the new term and drops the old one altogether. Seems simple, but this seemingly simple thing can avoid potential psychological breaks later.
Funny, I can run a full LAMP stack with X, and vi on a P3 with 256meg of ram. Let me know how all that rich multimedia content that's 'only internet and eamil' does on a similar spec, and then tell me again that what I do needs more grunt./rant over
Here, here! My server is living proof of that concept. It literally does just email and internet (yes, I have X and vim on it)... 800mhz, 256M of RAM, and some crappy video card that's at least a decade older than the server itself (can anyone say "trio 64"?).
In Evolution I got notifications of new emails. I was able to quickly and easily see my calendar. Notifications of new events happened without me checking for them.
Using OWA, I have to compulsively check/reload the page to check for new messages. I have to keep my appointments for the day in my mind constantly, or check the calendar, lest I miss them--not usually a problem, but it can become hectic on days where I have a lot of meetings and such.
Before this even posted, I was told by one of the Slashdot Editors (?) that I could use Zarafa; I installed and quickly realized it was a replacement for the Exchange server. This may fix the problem of needing Exchange at all, but my problem is that I've been forced to use Exchange; I need a client that will talk to that server from my Linux box.
Also: I have a VM, and it works fine... it's on another box at my desk. My problem is that I want to use something that runs on Linux. Even if it's run through WINE... I haven't looked into using Outlook '03 under WINE yet, I'll admit.
I'd only be slightly irritated. The people that do the course for free aren't getting all the benefits of the students that actually paid for it... you know, like college credit?
When I first read the parent post, I laughed pretty hard (on the inside). After thinking about it for a few moments, I actually began to worry that the underlying sarcasm that had provided the initial humor might not really be there... it's scary to think that the parent post might actually be completely sarcasm-free.
Reisinger alleges her First Amendment rights were violated by the city. She seeks $250,000 in compensatory damages, unspecified punitive damages and unspecified declaratory relief.
Keeping the unwashed masses away from Linux may mean less dumbing-down and all that, but it also means something else: less users. If there's not a large enough user base for a distro of Linux, it will eventually wither and die; who wants to work on software that only a handful of people are still using?
Sorry, but your analogy is completely fucking wrong. The base gripe here is apparently that people are complaining because Pidgin, an instant messaging program, doesn't have voice/video capabilities. So let's make a different analogy, going back to the original one about building a free house:
If someone says, "this house doesn't have a garage" that's fine.
But if someone says, "it is completely unacceptable that you built this house without a garage," then the only acceptable response from the builder who built the house for FREE is "go fuck yourself whiner."
Relating Pidgin's lack of voice/video support to having some nail sticking up in a house is just wrong. That implies that there's something wrong. Saying it is missing something is another thing altogether... as long as those requesting the feature understand that the developers are working on it FOR FREE.
Three arms and a leg? Try an order of magnitude higher... according to this, it would cost about $24.5M to send just the 16GB of data via text messaging.
The problem here is more likely the American "innocent until proven guilty" paradigm: the police normally wouldn't be able to prove that the offender was doing anything wrong, as proving it would require having it all on video or witnesses (that are willing to testify). And all it takes to get the charges dropped is probably just fighting it; the cop would rarely have enough solid evidence, so it comes down to the afore-mentioned paradigm (besides, arresting for these offenses, unless they were very grave, could potentially put a large and unnecessary burden on the courts).
Slashdot Mirror
Just a few years back I helped one of my friends put a bigger hard drive into a TiVo that was running an incredibly old kernel... I think it was in the 2.0 era, or very close to it (there's a possibility I'm wrong here, and the information was written to a corrupted part of my brain).
I believe (though I have no empirical evidence at hand to prove it) the reason swap should be 2x the size of your total amount of RAM:
1.) single program chews up all of your available memory
2.) kernel swaps all of that out for more important processes
3.) the new processes chew up all of the available system memory
4.) the original process(es) become important: kernel swaps all RAM out (making swap now 2x the amount of available RAM), pulling the original swapped memory back into physical memory
I'm pretty sure it's something like that. If there's not enough swap to handle this process (assuming it can't just create swap files on-the-fly), the kernel goes into Rambo mode and starts killing processes that are taking up too many resources. Trust me, nobody wants to see the kernel become Rambo (I know there's a pun in there somewhere about the "colonel" in the Rambo flicks).
It seems to be nothing more than a monitor with Unable to connect to database on it. Big B&W monitors make your thumbs hurt?
It seems more efficient. Except trees don't use electricity, and therefore don't contribute to the original problem, pollution, etc. Oh, and did anybody remember that trees produce oxygen, whereas these machines most decidedly do not? That seems like a rather important detail.
...riddle me this: when you get up tomorrow and go to sign in to GMail and all you get is a single page saying "Due to financial constraints and an inability to derive revenue from GMail to pay our bills, we have regretfully been forced to terminate the service.", where are your game serial numbers now and how do you plan on getting at them?
Keep a copy of all your email through a mail client. Seems pretty simple...?
That reminds me of a silly exploit on one of the hosted servers that I managed at an ISP (when reading, keep in mind I was a noob sysadmin). Through some pretty simple port scanning, I noticed a hosted server had a very odd port number open. I did a simple telnet to that port on the server, and was shocked to find I got a bash prompt with root-level permissions... keep in mind this server was publicly accessible (public IP).
I immediately notified the company that owned the server and told them to close the hole (and find how the attacker got in) immediately so other servers didn't get attacked by their rather beefy server. Hours went by with no response to my insistent calls and emails. I logged into their server and echoed stuff like "FIX YOUR HOLE" to their /dev/tty's (really annoying, BTW), but they did nothing.
I finally got tired of their inaction. I logged into their server through the hole (since I didn't have root/sudo on their server) and poked around for a few minutes to find where the hole had been created. It took a few minutes--I had no path set, so I had to use absolute paths to programs/files that I was unfamiliar with--but I found the problem was a silly entry in their inetd.conf file (or something like that). I removed it, restarted inetd, and tested that the hole was indeed plugged.
I found it absolutely amazing that they failed to act at all. On the rare occasions that an exploit was found on my server, it was fixed within a few minutes/hours of finding out (depending on how hard it was to fix and my accessibility to said server). I guess some people just don't care as long as their server is still running.
AMEN!
I was going to post this as a response to TheLink's post. I worked on the administrative section of a website for many years. It was built to have a single user account selected at all times, and assumed that they would only have a single window open and would only use my links (not the back button).
After a very short period of time, I found they completely ignored my requests. They would POST a search query, go to a result, then go back (re-POSTing the query). It was fairly intensive on the server, so I learned the trick of caching the results & redirecting after a POST (without printing anything to the screen). It's amazing how many intelligent web developers told me it was a stupid idea, even after explaining it... since I was in a position of authority, I eventually just forced them to do it my way, even though they somehow didn't understand the security (and user-friendliness) behind the concept.
What's even funnier (or more distressing) is when you look at CERN International on Google Maps... is that crumbled building?
No doubt. Of course, the laughing happened only when I saw the "click to reset." That was after my heart leapt into my throat.
I would guess the reason behind tying the tech support experience to not hiring him is because either
a.) he didn't seem to have potential for anything else, or
b.) something didn't strike 'em right in the interview, so they had to come up with an excuse.
I'd say the problem is mostly with the interviewing skills. Personally, I've had about an 80% success rate with interviews as long as I actually got a face-to-face. My first computer experience was working for a Gateway call center, which got me a position as tech support for an ISP: they'd lost their sysadmins prior to me coming in, so I showed interest in doing some of those duties and read all the appropriate books... in a few months I was managing the webserver (80 hosted websites). by the end of a year, I was formally the system administrator. By the end of another year, I added web development to that when the need arose. I was the lead programmer and eventually the project manager.
The above experience is great talking material at interviews. Being able to explain a journey from a wannabe geek in school ("programming" in BASIC) to tech support to sysadmin to web developer to lead programmer/project manager shows prospective employers that I have ambition and initiative.
Moral of the story: aim high. Even tech support positions can lead to much larger opportunities if only you look for them and grab on when the opportunity presents itself. Don't assume that you'll be handed anything on a silver platter (when promotions are available, usually the people that get said promotions are the one that apply for them).
I completely agree.
A small company that I was employed with for about 8 years doing web & software development, we made it a point to find students that were taking any sort of "geek" class, especially those in the helpdesk. When we (I) interviewed them, I looked for a couple of important qualities:
* Real world experience (building perl scripts to handle problems/automate stuff, etc)
* Actually wrote a script outside of school (see above)
* Good grasp of the Internet and its fundamentals (i.e. they know what a form is & the difference between frontend vs. backend systems)
* Did NOT LIE about their experience or lack thereof
* Weren't afraid to say "I don't know"
That last one is very important. Whenever I encountered somebody that fluffed their resume by saying things like "CMS Experience" which was directly tied into a terribly laid-out web page on MySpace, I'd pretty quickly red-flag that person. Now that I'm a contractor, I've found that people actually respect me more when I admit to not knowing something. Nobody likes to work with somebody that knows everything--especially when they don't know what they're talking about.
Building tests for your code proves the code "is correct" (read: "works properly") and bug-free to the extent the testing covers. When new bugs are discovered, tests can be added/modified to check for that, proving your code is still good.
The process of "proving your code is correct," to me, involves lots of face-to-face see-I-told-ya-so's, and displaying how it works. Tests allow you to say with authority, "I wrote 4,322 tests to make sure the system works the way it's supposed to. If you got an error message without the system crashing, you did something wrong."
AMEN! Code that means exactly what it says is key to sanity. Consider:
$UI->update_section('message_box', $myError);
versus:
## Update the "message_box" part of the user interface with the error...
$obj->$callMethod($tName, $data);
The first one is pretty straight forward. The second one is so irritating to read that one is more likely to lead the reader into assuming that reading the comments is more important than reading the code (remember: assuming makes an @ss out of "u" and "ming"... and that's one pissed-of Chinaman, I'll tell ya).
Commenting code is good, as long as the comments are useful. If your comments are misleading, then their presence is much more damaging than if they hadn't been there at all. I used to comment all my code, even if the code's purpose was very obvious, then I got bitten by it:
I coded something a long time ago that needed to be fixed. I looked at the comments around where the problem occurred, and made a minor fix. Later, it seemed the bug had mutated into a different form, and it all went back to the afore-mentioned block of code I'd fixed. I read the comments and fixed accordingly... which completely broke the system. When I had made the first fix, I didn't update the comment to explain the updated functionality, so the comment was completely misleading.
This goes along with "naming conventions." I put this in quotes because it can be more than just the code. For instance, if the front end changed "todos" into "tasks", make sure the code IMMEDIATELY uses the new term and drops the old one altogether. Seems simple, but this seemingly simple thing can avoid potential psychological breaks later.
Funny, I can run a full LAMP stack with X, and vi on a P3 with 256meg of ram. Let me know how all that rich multimedia content that's 'only internet and eamil' does on a similar spec, and then tell me again that what I do needs more grunt. /rant over
Here, here! My server is living proof of that concept. It literally does just email and internet (yes, I have X and vim on it)... 800mhz, 256M of RAM, and some crappy video card that's at least a decade older than the server itself (can anyone say "trio 64"?).
In Evolution I got notifications of new emails. I was able to quickly and easily see my calendar. Notifications of new events happened without me checking for them.
Using OWA, I have to compulsively check/reload the page to check for new messages. I have to keep my appointments for the day in my mind constantly, or check the calendar, lest I miss them--not usually a problem, but it can become hectic on days where I have a lot of meetings and such.
Before this even posted, I was told by one of the Slashdot Editors (?) that I could use Zarafa; I installed and quickly realized it was a replacement for the Exchange server. This may fix the problem of needing Exchange at all, but my problem is that I've been forced to use Exchange; I need a client that will talk to that server from my Linux box.
Also: I have a VM, and it works fine... it's on another box at my desk. My problem is that I want to use something that runs on Linux. Even if it's run through WINE... I haven't looked into using Outlook '03 under WINE yet, I'll admit.
I'd only be slightly irritated. The people that do the course for free aren't getting all the benefits of the students that actually paid for it... you know, like college credit?
When I first read the parent post, I laughed pretty hard (on the inside). After thinking about it for a few moments, I actually began to worry that the underlying sarcasm that had provided the initial humor might not really be there... it's scary to think that the parent post might actually be completely sarcasm-free.
The city was suing the woman. It also states:
Reisinger alleges her First Amendment rights were violated by the city. She seeks $250,000 in compensatory damages, unspecified punitive damages and unspecified declaratory relief.
RTFA?
Keeping the unwashed masses away from Linux may mean less dumbing-down and all that, but it also means something else: less users. If there's not a large enough user base for a distro of Linux, it will eventually wither and die; who wants to work on software that only a handful of people are still using?
Sorry, but your analogy is completely fucking wrong. The base gripe here is apparently that people are complaining because Pidgin, an instant messaging program, doesn't have voice/video capabilities. So let's make a different analogy, going back to the original one about building a free house:
If someone says, "this house doesn't have a garage" that's fine.
But if someone says, "it is completely unacceptable that you built this house without a garage," then the only acceptable response from the builder who built the house for FREE is "go fuck yourself whiner."
Relating Pidgin's lack of voice/video support to having some nail sticking up in a house is just wrong. That implies that there's something wrong. Saying it is missing something is another thing altogether... as long as those requesting the feature understand that the developers are working on it FOR FREE.
Three arms and a leg? Try an order of magnitude higher... according to this, it would cost about $24.5M to send just the 16GB of data via text messaging.
It just goes to show, if ya can't fork it, spoon^H^H^H^H^Hthread it!
The problem here is more likely the American "innocent until proven guilty" paradigm: the police normally wouldn't be able to prove that the offender was doing anything wrong, as proving it would require having it all on video or witnesses (that are willing to testify). And all it takes to get the charges dropped is probably just fighting it; the cop would rarely have enough solid evidence, so it comes down to the afore-mentioned paradigm (besides, arresting for these offenses, unless they were very grave, could potentially put a large and unnecessary burden on the courts).