XP - Extreme Programming

Simplicio · September 5th, 2002, 11:21 am

I may be joining a project where they are making extensive use of XP techniques; pair programming, rapid iterations, joint ownership etc. I have read a lot about it before and thought it made good sense, but never been given the opportunity/had to actually put it into practice. Does anyone have any experience or thoughts on XP?

lodette · September 6th, 2002, 6:13 pm

It can work pretty well in practice, and seems to be a more accurate formalization of the way people actually work. We've found the unit testing/"test first" discipline to be the most valuable part of XP for us (group of 18 developers).Still, there is no one-size-fits-all methodology, either for a group or a problem area.

jamesbattle · September 9th, 2002, 12:49 am

Totally agree. The "test first" is the key. Sadly few banks yet use any usefulmethodology and that's the reason they have hundreds and sometimesthousands of programmers.We try to spend as much time on testing as in development and sometimesend up with test programs that are more sophisticated than the code beingtested. We do this because: (a) it saves money, (b) it's much faster. Don Knuth had some nice stuff to say about he tested things like TEX.

Simplicio · September 10th, 2002, 8:04 am

I already kind of use a unit test first approach so this should suit me. What I am more curious about is pair programming; I never did anything like that as I generally work alone with people of other 'specializations', or occasionally on larger but more closed teams. Any anecdotes?JamesBattle << Do you have the link for this Knuth reference?

Simplicio · September 10th, 2002, 9:59 am

Googling for the Knuth reference I found these amusing articles:TeXBinary

lodette · September 10th, 2002, 11:49 am

I know it sounds like it would be counterintuitive/counterproductive at first, but pairs programming can be very effective during development. On the other hand, we've found it to be pretty useless for debugging. Then again, I'm not entirely sure that pairs debugging is meant to be part of XP.

jamesbattle · September 13th, 2002, 12:56 am

IMHO, withouth insult, some of this also makes me think that there's an ample helping of 'popular bullshit' in some of these books you might be reading...I'll dig up the Knuth references when I've got time to surface.

Simplicio · September 13th, 2002, 6:30 am

JamesBattle << IMHO, withouth insult, some of this also makes me think that there'san ample helping of 'popular bullshit' in some of these books you mightbe reading...Not insulted, but I didn't quite follow what you meant.

doofusmaximus · September 24th, 2002, 2:59 pm

I think XP pair programming can go both ways, pairing a timid worker and a loud mouth who wants their own way all the time may not be what paired programming is about. Or pairing novice and a guruthe novice will do/learn nothing and the guru will do all the workand increase their frustration I believe it can be beneficial when done right and both programmers learn and grow in the process. Approach with caution

mikewoodhouse · September 24th, 2002, 9:54 pm

A good set of unit tests developed using test-first (i.e. write a failing test, then write the simplest code that could possibly make the test succeed) is a marvellous discipline in itself but it's still going to lead (potentially) to spaghetti without refactoring, which is the driver for emergent design. A lot of the practices espoused by XP are just good programming: the originator(s) started off from the premise that they should take what worked and do it to extremes, hence the name.

DominicConnor · October 3rd, 2002, 10:00 am

In my misspent youth I got very into testing It is worth building a few stress tests using stuff like Rational's Visual Test. This works rather like themacro recorder in Excel, except it captures mousing,keyboard etc from various applications.Dominiconnor