As a software development major, I'm starting to get really sick of constantly having three or four projects on my plate at any given moment. I love writing software, but if I have three projects in different languages, all due the same week (and I'll be getting three more when I hand in the current ones), I can't really put that much time in to any of the projects. However, this isn't so bad when you consider I've been writing the same crap for about three years now. Granted, in a multitude of different languages - but a linked list is a linked list, and a queue is a queue... I get it, will samples in five languages really prove it?
How many times do I have to jump through hoops before I'm allowed to actually get back to learning about what I love? Currently, I have a project due tomorrow (well, midnight by email) that I was allowed to use any language for (the second program due tomorrow must be done in LISP), so I decided to use it as an excuse to pick up Perl, which I've wanted to do for a long time now. However, under the time constraints of a paper, two programs and an exam this week, and a make up day taken from Easter vacation, I ended up learning quite a bit of Perl, but not enough to finish the project. So, I sighed and wrote the thing in Java.
Normally I'd take a low grade in order to keep working with Perl, but doing this so many times means my QPA can't afford to absorb another failed project because I wanted to go out of the way to learn something (I've failed for going off on tangents for concurrency, alpha blending, creating dynamic thread priorities, and quite a bit of kernel hacking, etc...). How many times am I going to have to do the same thing over, and over and over before I'm allowed to go off and learn something? How many times am I going to have to "study the test" instead of getting in another chapter from "Code Complete" or "The Mythical Man Month"? Why should I spend nights trying to figure out what curve balls the prof is going to throw at me on the exam, instead of discussing top down versus bottom up design (oh, failed a project doing frameworks bottom up...), or non deterministic garbage collection? Am I the only one who thinks the greater majority of undergrad work is busy work?
If you're using Linux, which TCP Congestion algorithm are you using? Reno isn't very fair; if a single connection is congested beyond the first hop, you'll slow down the rest of your connections when the window slides to smaller units. Have you tried Bicubic, Veno, or any of the other 9 or 10 congestion algorithms?
You can change them on the fly by echoing the name into your procfs, IIRC. Also, if you have the stomache for it, and two connections to the internet, you can load balance and/or stripe them using Linux advanced Routing & Traffic Control (mostly the ip(1) command). Very cool stuff if you want to route around a slow node or two (check out the multiple path stuff) at your ISP(s).
FWIW, I've heard the rule of thumb on the government (The US... I assume European governments are roughly in the same ballpark where economy allows) is about 5 years ahead of the private sector when it comes to any given technology.
That is, so far as research goes. I think implementation lags the private sector by several years due to bureaucracy.
Damn it. There are times when you just look at the article title and you know that a long, delicious, juicy flamewar is coming up...
And I just lost my mod points, too.:( If it makes you feel any better, you don't have the dilemma of trying to decide whether to mod a fanboy into the ground or light him up like a Christmas tree via posting a reply. Ironically, this pushes that very problem up stream to current mods. Right as they started doing 10 mod points, no less (when did that happen anyway?).
Happy hunting! First one to get forcibly rejected from Slashdot gets a bottle opener key chain! Double points for a k-line, and Quad for a z-line.
"we can do whatever we like with it, like install Linux on it."
24 Carat Pure Slashdot Gold.
We have a winner. I call for a slashdot version of the Godwin; any technical thread on the viability of any technology is over the moment anyone claims something to the effect of "... We could install Linux on it!"
However, asking "... does it run on Linux?" is still fair game.
Funny you say that; I'm from the other direction! I'm a programmer, and I started riding a motorcycle; I'm still a software development major in college, but I find that I work on my bike in much the same way I work on my computer. Although, I don't love working on my bike as much as I love working on my computer or software, either.
Yeah, but I, for one, am stuck with a quarter million SLOC VBA(which is really a subset of VB 6) application that has no upgrade path to.NET. Furthermore, even if I could get out of VB 6, I'd have to refactor all the DAO calls to ADO, and that to ADO.NET or wrap them in an ODBC wrapper class. Sorry, I'm a little bitter about supporting someone else's custom Access application without any way out. The least they could have done is given me a way to get all the binary forms and reports out; sure, I can get the modules and class code, but I've no recourse other than remaking about a hundred forms. I understand that they moved on to.NET and there is no real way to port the code accurately, but they could have at least given me a way to export the design.
All other things being equal, after trying to find anything in the MSDN, let alone the vaporware API I inherited (DAO Jet library, try to find it some time - they didn't actually document it, and what they did document, was neither complete nor useful. That's being charitable.), seeing a Javadoc is like stumbling out of the desert and in to an Evian bottling factory.
Sure, the J2SE API as HTML files is 60 MB or something, but at the very least, it's consistent, well laid out, the cross references make sense, and I can drill down to the objects that are probably what I need in about 3 or 4 'hops'. Also, how large is the MSDN now? I think I've got about 4 versions (VS 6.0 - VS2k8) each weighing in at something like 700 MB. Back to my desert analogy, it's like "water, water everywhere; but not a drop to drink" with the MSDN.
I think that there are two different kinds of emotions here; with a movie, you're being 'forced' to feel the way you do, so it doesn't seem to be a 'real' emotion although it feels the same. In real life, if you will, these events cause really deep, pure emotions that are... well, difficult to handle at times.
I salute you! That, sir, is brilliant! Although I'm not sure the legal status of fake browsing... I'd say though that it's probably fairly safe; you aren't targeting specific sites or anything. Otherwise, spidering the web would fall under this arena and everyone would be suing everyone else who owns a search engine. Although, obviously, IANAL.
From their website
What does it keep?
At first, Phorm's technology collects information on browser type, response to advertising, the URLs of some of the web pages viewed, and search terms entered. Neither URLs nor search terms are stored - they are discarded immediately. The matching information that's left is assigned to an anonymous, randomly-generated ID number. The random ID marks an anonymous list of the categories of products or services in which a user appears to be interested. I think they're sniffing on the wire passively instead of using cookies. Although, it's hard to tell from their blurb that doesn't contain a single element of useful or technical data. Please correct me if I'm wrong; I'm just assuming from what I can gather.
Good point; make sure script doesn't request any page content other than the index/plain text. Like elinks, I guess. That and a little bit of common sense dictionary filtering and/or metadata tags. Although I see where you were going on the whole with it... I haven't the foggiest idea how to make sure I don't land on a page that puts me on a government list somewhere:). Any ideas?
This is one of the best, honest, pieces of advice I've read on slashdot in a good long while.
For the record, I'm 23 and somewhere between your attitude (read: good days) and the OP's (read: bad days, putting out fires that I didn't create, which were the results of fires I did create, so I can get to putting out my own fires).
Lately I've started to take on the mentality that I'm going to do my best in spite of, and sometimes to spite, myself. At the end of the day, I just want to own my technical domain; I want to beat the living crap out of it and own it in every regard, with or without recognition. I've found myself so much more 'free' when I don't expect recognition and I'm only doing it for myself, my principles, and love of the field. Except when I'm stuck programming VBA. I hate VBA and Access databases. Someone else can own them.
You're kidding, right? Were you going to school on grants? I'm at a state school at about $14,000/year, books not included (try $500/semester from the bookstore, or $100/semester from online bookstores). Oh, yeah, meal plan is extra. It costs the same per month to live off campus as it does to share a glorified closet with a roommate... really a no brainer (especially for a software development major). I'll rough the five minute drive so I can: play my stereo as loud as I want when I want (studio apartment, no neighbors), keep my gun (no weapons on campus), run 7 computers and have 2 desks, hang two white boards behind my desks (I do the 'L' configuration with the wall to my back, I can swivel my chair to either desk or whiteboard and my bookshelf is within arms length - I recommend it to everyone!) not to mention my own bathroom. Oh, and I can write code in my boxers or naked if I'd like; you can't put a price tag on that!
I guess they (Phorm) just track web URLs; I was thinking just a simple dictionary attack with a bit of depth to it should take care of this. I just pulled this from my butt at this moment, but I think it would work if you created a shell script or even batch file to do the following...
Get your favorite tar balled dictionary, pull a random word from it, google the random word with elinks or something, and follow a random link with wget. From that site, pull 3 unique links and visit them, from those sites pull 2 unique links and follow them, from those sites pick a single unique link and follow that. Rinse, lather, repeat.
This should give a deep enough tree with a large enough fanout over enough topics to mask your normal usage patterns. Bonus points for switching up protocols and ports every now and then.
Perhaps the old hacker trick of lowering your signal/noise ratio via injecting bad/misleading data (somewhere in the flow)? If you can't be very quiet, you can usually benefit from being very loud.
Well, I think that the noop scheduler/elevator should yield the best performance. It has the least overhead, and the scheduling overhead for deadline and cfq are trade offs for optimizing a read head. The anticipatory scheduler adds a 1ms delay, but very little in the way of processing overhead. I'm just wondering how much performance could be yielded if all the various disk scheduling (read ahead, write behind, delays for queues, read and write starving, etc.) code is going to go the way of the... erm, really dead thing...
Just wondering, are you using the SSD for swap space?
I imagine that's where it really starts to shine; when you're already in trouble and start paging (or swapping - I never know which is correct anymore. I always thought you paged out to swap, but everyone seems to be calling it swapping now) out to swap, you don't incur the extra several ms delay for rotational latency and perhaps a seek, as well. My only concern would be the number of write cycles if you start thrashing fairly hard.
I just realized you said eepc, do you know which disk scheduler you're using with it? I'd be curious to see the difference in CFQ and noop when the mechanical disk is out of the equation.
Ummm, I don't know where to start with this post... I'll just sum it up in a thought; you need to run Slackware once or twice to understand why *nix is *nix. I'm not just saying that because I'm a Slackware fanboy, but rather, because you seem to miss the elegance and simplicity of text files for configs.
There are a million reasons why a single text file in/etc/ that can be edited over SSH, and has a man 5 page, is superior to any other kind of scheme. There are a million reasons why a GUI interface for maintenance is a nightmare (and how would you like to set it up without a command line?) as compared to SSH. If you want to know why I say this, you'll first have to understand why *nix is *nix.
There have been a few times that I've used 'links' when I'm ssh'ed into a headless box and I can't wget a file I need because it requires me to hit an 'I agree' radio button. For instance, when getting a new java JDK.
I use elinks for the same reason as the guy in the summary; it just works when I need it. Sure, I could download to my desktop and scp it over, but why go through the trouble when elinks just works and firewalls aren't an issue when connecting to a web server? The box may not be on a network where I can get a cvs/git/svn connection, but I've always got a hole in the firewall for connecting to web servers.
Slashdot Mirror
How many times do I have to jump through hoops before I'm allowed to actually get back to learning about what I love? Currently, I have a project due tomorrow (well, midnight by email) that I was allowed to use any language for (the second program due tomorrow must be done in LISP), so I decided to use it as an excuse to pick up Perl, which I've wanted to do for a long time now. However, under the time constraints of a paper, two programs and an exam this week, and a make up day taken from Easter vacation, I ended up learning quite a bit of Perl, but not enough to finish the project. So, I sighed and wrote the thing in Java.
Normally I'd take a low grade in order to keep working with Perl, but doing this so many times means my QPA can't afford to absorb another failed project because I wanted to go out of the way to learn something (I've failed for going off on tangents for concurrency, alpha blending, creating dynamic thread priorities, and quite a bit of kernel hacking, etc...). How many times am I going to have to do the same thing over, and over and over before I'm allowed to go off and learn something? How many times am I going to have to "study the test" instead of getting in another chapter from "Code Complete" or "The Mythical Man Month"? Why should I spend nights trying to figure out what curve balls the prof is going to throw at me on the exam, instead of discussing top down versus bottom up design (oh, failed a project doing frameworks bottom up...), or non deterministic garbage collection? Am I the only one who thinks the greater majority of undergrad work is busy work?
If you're using Linux, which TCP Congestion algorithm are you using? Reno isn't very fair; if a single connection is congested beyond the first hop, you'll slow down the rest of your connections when the window slides to smaller units. Have you tried Bicubic, Veno, or any of the other 9 or 10 congestion algorithms?
You can change them on the fly by echoing the name into your procfs, IIRC. Also, if you have the stomache for it, and two connections to the internet, you can load balance and/or stripe them using Linux advanced Routing & Traffic Control (mostly the ip(1) command). Very cool stuff if you want to route around a slow node or two (check out the multiple path stuff) at your ISP(s).
FWIW, I've heard the rule of thumb on the government (The US... I assume European governments are roughly in the same ballpark where economy allows) is about 5 years ahead of the private sector when it comes to any given technology.
That is, so far as research goes. I think implementation lags the private sector by several years due to bureaucracy.
And I just lost my mod points, too.
Happy hunting! First one to get forcibly rejected from Slashdot gets a bottle opener key chain! Double points for a k-line, and Quad for a z-line.
24 Carat Pure Slashdot Gold.
We have a winner. I call for a slashdot version of the Godwin; any technical thread on the viability of any technology is over the moment anyone claims something to the effect of "... We could install Linux on it!"
However, asking "... does it run on Linux?" is still fair game.
Funny you say that; I'm from the other direction! I'm a programmer, and I started riding a motorcycle; I'm still a software development major in college, but I find that I work on my bike in much the same way I work on my computer. Although, I don't love working on my bike as much as I love working on my computer or software, either.
Yeah, but I, for one, am stuck with a quarter million SLOC VBA(which is really a subset of VB 6) application that has no upgrade path to .NET. Furthermore, even if I could get out of VB 6, I'd have to refactor all the DAO calls to ADO, and that to ADO.NET or wrap them in an ODBC wrapper class. Sorry, I'm a little bitter about supporting someone else's custom Access application without any way out. The least they could have done is given me a way to get all the binary forms and reports out; sure, I can get the modules and class code, but I've no recourse other than remaking about a hundred forms. I understand that they moved on to .NET and there is no real way to port the code accurately, but they could have at least given me a way to export the design.
All other things being equal, after trying to find anything in the MSDN, let alone the vaporware API I inherited (DAO Jet library, try to find it some time - they didn't actually document it, and what they did document, was neither complete nor useful. That's being charitable.), seeing a Javadoc is like stumbling out of the desert and in to an Evian bottling factory.
Sure, the J2SE API as HTML files is 60 MB or something, but at the very least, it's consistent, well laid out, the cross references make sense, and I can drill down to the objects that are probably what I need in about 3 or 4 'hops'. Also, how large is the MSDN now? I think I've got about 4 versions (VS 6.0 - VS2k8) each weighing in at something like 700 MB. Back to my desert analogy, it's like "water, water everywhere; but not a drop to drink" with the MSDN.
They're 'inter-operable' now that the UK is cracking down on them. It's merely a good faith gesture on Microsoft's part.
I think that there are two different kinds of emotions here; with a movie, you're being 'forced' to feel the way you do, so it doesn't seem to be a 'real' emotion although it feels the same.
In real life, if you will, these events cause really deep, pure emotions that are... well, difficult to handle at times.
I salute you! That, sir, is brilliant! Although I'm not sure the legal status of fake browsing... I'd say though that it's probably fairly safe; you aren't targeting specific sites or anything. Otherwise, spidering the web would fall under this arena and everyone would be suing everyone else who owns a search engine. Although, obviously, IANAL.
Good point; make sure script doesn't request any page content other than the index/plain text. Like elinks, I guess. That and a little bit of common sense dictionary filtering and/or metadata tags. Although I see where you were going on the whole with it... I haven't the foggiest idea how to make sure I don't land on a page that puts me on a government list somewhere :). Any ideas?
For the record, I'm 23 and somewhere between your attitude (read: good days) and the OP's (read: bad days, putting out fires that I didn't create, which were the results of fires I did create, so I can get to putting out my own fires).
Lately I've started to take on the mentality that I'm going to do my best in spite of, and sometimes to spite, myself. At the end of the day, I just want to own my technical domain; I want to beat the living crap out of it and own it in every regard, with or without recognition. I've found myself so much more 'free' when I don't expect recognition and I'm only doing it for myself, my principles, and love of the field.
Except when I'm stuck programming VBA. I hate VBA and Access databases. Someone else can own them.
You're kidding, right? Were you going to school on grants? I'm at a state school at about $14,000/year, books not included (try $500/semester from the bookstore, or $100/semester from online bookstores). Oh, yeah, meal plan is extra. It costs the same per month to live off campus as it does to share a glorified closet with a roommate... really a no brainer (especially for a software development major). I'll rough the five minute drive so I can: play my stereo as loud as I want when I want (studio apartment, no neighbors), keep my gun (no weapons on campus), run 7 computers and have 2 desks, hang two white boards behind my desks (I do the 'L' configuration with the wall to my back, I can swivel my chair to either desk or whiteboard and my bookshelf is within arms length - I recommend it to everyone!) not to mention my own bathroom. Oh, and I can write code in my boxers or naked if I'd like; you can't put a price tag on that!
I guess they (Phorm) just track web URLs; I was thinking just a simple dictionary attack with a bit of depth to it should take care of this. I just pulled this from my butt at this moment, but I think it would work if you created a shell script or even batch file to do the following...
Get your favorite tar balled dictionary, pull a random word from it, google the random word with elinks or something, and follow a random link with wget. From that site, pull 3 unique links and visit them, from those sites pull 2 unique links and follow them, from those sites pick a single unique link and follow that.
Rinse, lather, repeat.
This should give a deep enough tree with a large enough fanout over enough topics to mask your normal usage patterns. Bonus points for switching up protocols and ports every now and then.
Perhaps the old hacker trick of lowering your signal/noise ratio via injecting bad/misleading data (somewhere in the flow)? If you can't be very quiet, you can usually benefit from being very loud.
Well, I think that the noop scheduler/elevator should yield the best performance. It has the least overhead, and the scheduling overhead for deadline and cfq are trade offs for optimizing a read head. The anticipatory scheduler adds a 1ms delay, but very little in the way of processing overhead. I'm just wondering how much performance could be yielded if all the various disk scheduling (read ahead, write behind, delays for queues, read and write starving, etc.) code is going to go the way of the... erm, really dead thing...
I imagine that's where it really starts to shine; when you're already in trouble and start paging (or swapping - I never know which is correct anymore. I always thought you paged out to swap, but everyone seems to be calling it swapping now) out to swap, you don't incur the extra several ms delay for rotational latency and perhaps a seek, as well. My only concern would be the number of write cycles if you start thrashing fairly hard.
I just realized you said eepc, do you know which disk scheduler you're using with it? I'd be curious to see the difference in CFQ and noop when the mechanical disk is out of the equation.
Ummm, I don't know where to start with this post... I'll just sum it up in a thought; you need to run Slackware once or twice to understand why *nix is *nix. I'm not just saying that because I'm a Slackware fanboy, but rather, because you seem to miss the elegance and simplicity of text files for configs.
/etc/ that can be edited over SSH, and has a man 5 page, is superior to any other kind of scheme. There are a million reasons why a GUI interface for maintenance is a nightmare (and how would you like to set it up without a command line?) as compared to SSH. If you want to know why I say this, you'll first have to understand why *nix is *nix.
There are a million reasons why a single text file in
Ah, yes, good ol' C-x M-c M-comcast. Friggin' emacs...
I use elinks for the same reason as the guy in the summary; it just works when I need it. Sure, I could download to my desktop and scp it over, but why go through the trouble when elinks just works and firewalls aren't an issue when connecting to a web server? The box may not be on a network where I can get a cvs/git/svn connection, but I've always got a hole in the firewall for connecting to web servers.