neko,

g_m is correct. makes it a hell of a lot easier when %s-ing paths!


g_m,

i'm one of those "universities must teach theory" people so i'm a bit sensitive about this topic .

neko,

tdd:

with practice, it will start to feel natural. i'm now at the point where i feel uncomfortable when writing code without having tests. this is a good thing!

it does mean that i do a lot of refactoring. this is also a good thing, but at the same time it can be...

lol @ gif!

yeah...I hate to think of all the \/'s I've written over the years

Yeah...it's one of those things where I know that once it sticks I'll wonder how I lived without it (like emacs!). Just need to make the initial time investment and effort to force myself to stop when I find myself banging out code without writing tests first.

i think i agree, actually. again, wasn't even making a judgement one way or the other, or saying that a CS degree is not a worthwhile activity, only that it's not an efficient path to employment. that doesn't mean you shouldn't do it....

+1 to your comments on TDD

A CS degree is closer to a math degree than to "programming" imo. Helpfull for lots of non-programming jobs.

yes. There is a huge demand for programming skills which universities ignore. Programming should be taught by the engineering department.

Our university has an entire department called Software/Systems Engineering

Re TDD: Do you guys follow the rules religiously? Things like never write production code that doesn't have a test already failing?

I got my undergrad in Computer Engineering and am getting my masters in Computer Science. While I do like the balance I got from all of this, I think to true Software Engineering program would do a much better job of making GOOD programmers that can work well on larger code bases and manage all the other pieces that come with it.

i mean, if i'm in eclipse and i've written a test for a class that doesn't exist, i'm not going to actually run the test and see the red bar and then make the class -- i'll just make the class and try an implementation and then run test. but i guess the test is already written, so yeah. it ends up becoming a source of comfort and you want to do it that way.

i will say this: like many things in programming, the up-front time of getting it all setup right so it's easy to do makes all the difference. if, eg, you were constantly jumping back to a console, and hitting up arrow, return to run your test, or if the test runner was too slow, etc, it's going to be much harder to form a habit around it.

ideally, yes, this is how i do things.

i'm finding in my new job that a couple things prevent me from doing this really truly 100% of the time:

- it can be difficult to write tests when you're still in the "wtf how does this work?" phase.

- especially when i'm also working on a system that has no existing testing infrastructure.

so in these instances, i compromise by hacking a bit, then circling back to the tests, but this does make me feel uncomfortable.

basically, i feel that i just need to get better at reading code and refactoring so that i can write stuff test-first more quickly and efficiently. i have zero doubt that test-first is the "best" way to write code.

So if I understand correctly, you guys are writing code for the purpose of testing classes that you haven't written yet?

just a warning that this is a bit of a slippery slope. imo you should always run the test and make sure it fails. i have written tests that cannot ever fail, which is obviously undesirable, and somehow only happens when i don't run the test each time through. red-green-refactor!

i don't mind bouncing to another terminal, mostly because i invariably want to do something slightly different with my test output (only run a certain test, grep for something, whatever). however, ":map T your-test-program<cr>" is a nice start. there are also tools for running your tests whenever you save or every N minutes or anything else you can think of.

generally, i agree with the sentiment that time spent getting your environment in order is always a wise investment.

oh and slow tests are murder. the codebase i work on now has a bunch of integration tests masquerading as unit tests and it is unbelievably painful to run your test suite, watch it sit there loading classes for 10 seconds, then spitting out output.

not sure where you're headed with this but yes. a better way to say it is we write code for the purpose of testing part of a class that we're going to write next.

Not really headed anywhere but just making sure I understood the conversation. I've never heard of doing testing that way.

None of my classes ever taught me anything about testing, and in my current workplace we don't have a formal testing program in place. I pretty much write my code and test it as I go. Then we usually release it in beta for feedback, and clean up any issues that arise from that.

Testing is one of the issues that makes me nervous about interviewing for a full time developers position outside my current company.

i think programming is important for biology too. pretty sure you can make a ton of money if you know statistics, programming, and biology

Not that there is 100% agreement about this, but I'd say TDD is widely accepted as good standard practice, and by many very smart programmers is touted as the only reasonable way to be coding nowadays. If you're a professional programmer you should google it and learn about it.

i have a hard time implementing test-first but here is a good start

http://blog.extracheese.org/2009/11/...arted_tdd.html

Nice link, greg. I like this line especially: "Building software without TDD was crushingly stressful, but I couldn't see it at the time."

LOL, that remind me of this one:

http://www.youtube.com/watch?v=j11QO5sc0Q8

Juk

question:

at this link

http://www.sbrforum.com/betting-odds...?date=20110813

see where it says Greenbay Clevland, then there is a image box which says "I" you can click to get the scores. where is that information coming from? i can't find the URL that is connected with the I link

http://www.sbrforum.com/ajax/?a=[SBR.Odds.Modules]OddsEvent_GetEvent&ConsensusBookId=19&DefaultBookI d=19&EventId=93212&Key=4263077f8f99858c5221f712fa1 3753d&League=NFL&MatchupLink=http%3A%2F%2Fwww.sbrf orum.com%2Fnfl-football%2Fmatchups%2F20110813-274%2F&Sport=football&SportsbookId=19&StartDate=20 11-08-13&UserId=0&View=ED&theme=default


EDIT: Just click F12 to see dev tools in ff or chrome, and it will show you the ajax post that's being made.

okay i dl the development tools i got the information now thanks!

June the 30th this year will be 24 hours and 1 second long, and will contain the time 23:59:60

[x-post from HN, this is due to a 'leap-second']

http://hpiers.obspm.fr/iers/bul/bulc/bulletinc.dat

Yeah. There's one every few years. It's because 1) the Earth spins very slightly slower than one spin per 24 hours and 2) the Earth's rotation actually slows down sporadically due to a whole variety of things.

I'm not really convinced that making sure that the sun is at its highest point at exactly noon (as opposed to like 30 seconds off) if you're right in the middle of a time zone is worth the confusion, but it's also not much confusion.

It causes a bit of a mess for Unix time.