Lines of code...
-
I heard, somewhere, that a REAL programmer is meant to only be able to generate 7, thats right Seven, line of fully tested, documented, (stamped, lost, signed and buried in peat for 3 months, then burnt as fire lights ^_^), per day... I am being stupid or does that number seem a tad _low_?
-
I heard, somewhere, that a REAL programmer is meant to only be able to generate 7, thats right Seven, line of fully tested, documented, (stamped, lost, signed and buried in peat for 3 months, then burnt as fire lights ^_^), per day... I am being stupid or does that number seem a tad _low_?
Only if it is a really, really mature product I can see this. This type of work usually involving sifting through lots of other code for fixes and tweaks. Otherwise I would say it should be more like 500-1000 lines of new code. I think this quote must have been taken out of context somehow.
-
I heard, somewhere, that a REAL programmer is meant to only be able to generate 7, thats right Seven, line of fully tested, documented, (stamped, lost, signed and buried in peat for 3 months, then burnt as fire lights ^_^), per day... I am being stupid or does that number seem a tad _low_?
Dunno where you read that. Dunno what a REAL programmer is, either. I do know that I did some research a while back, and (from what I remember) 50-100 per day is pretty good. However, lines of code is a bad measure of programmer productivity. It is useful for other things, such as measuring the number of bugs per new line of code/changed line of code, or general quantification of the overall complexity of an application. I also experimented with estimating software using a quantification of the overall complexity of the software, translated to a rough number of lines of code (10,000 = simple, 50,000 = medium, etc), and then dividing that by the individual programmer LOC figure to get the number of man-days of development. So a simple app would take 10,000 / 50 = for developer X = 200 man-days. For super-developer-Dave, it would take 10,000 / 200 = 50 man-days. At 7 lines per day, it would take 10,000 / 7 = 1428 man-days = about 4 years.
-
Dunno where you read that. Dunno what a REAL programmer is, either. I do know that I did some research a while back, and (from what I remember) 50-100 per day is pretty good. However, lines of code is a bad measure of programmer productivity. It is useful for other things, such as measuring the number of bugs per new line of code/changed line of code, or general quantification of the overall complexity of an application. I also experimented with estimating software using a quantification of the overall complexity of the software, translated to a rough number of lines of code (10,000 = simple, 50,000 = medium, etc), and then dividing that by the individual programmer LOC figure to get the number of man-days of development. So a simple app would take 10,000 / 50 = for developer X = 200 man-days. For super-developer-Dave, it would take 10,000 / 200 = 50 man-days. At 7 lines per day, it would take 10,000 / 7 = 1428 man-days = about 4 years.
Yes but would that code be 'bug' free? There are different criteria if one is producing safety critical software. These days, now I am showing my age, it seems you just hack a few thousand lines of code, and if there are bugs so what, just fix and post on the Net. The best example I could quote was a project, which I said would take 6 weeks and would involve a trip to Switzerland to test it, before delivering to the customer in Austria, I was UK based. I was made 'redundant' the compamny then took shortcuts, it finaly took 48 weeks, and the code still had bugs!
"Nothing is impossible for the man who doesn't have to do it himself." - A. H. Weiler
-
Only if it is a really, really mature product I can see this. This type of work usually involving sifting through lots of other code for fixes and tweaks. Otherwise I would say it should be more like 500-1000 lines of new code. I think this quote must have been taken out of context somehow.
Jeff Bogan wrote: I would say it should be more like 500-1000 lines of new code. Excluding generated code of course, but theres only a handful of code gods that can dish out 1000 lines of quality code. I prefer quality over quantity. top secret
Download xacc-ide 0.0.3 now!
See some screenshots -
Yes but would that code be 'bug' free? There are different criteria if one is producing safety critical software. These days, now I am showing my age, it seems you just hack a few thousand lines of code, and if there are bugs so what, just fix and post on the Net. The best example I could quote was a project, which I said would take 6 weeks and would involve a trip to Switzerland to test it, before delivering to the customer in Austria, I was UK based. I was made 'redundant' the compamny then took shortcuts, it finaly took 48 weeks, and the code still had bugs!
"Nothing is impossible for the man who doesn't have to do it himself." - A. H. Weiler
I worked on safety critical products for five years and testing could take 2-3 times a slong as develoment as you got further into a project due to regression requirements. Elaine :rose: The tigress is here :-D
-
I worked on safety critical products for five years and testing could take 2-3 times a slong as develoment as you got further into a project due to regression requirements. Elaine :rose: The tigress is here :-D
Not to mention wrting the specification and program design.
"Nothing is impossible for the man who doesn't have to do it himself." - A. H. Weiler
-
I heard, somewhere, that a REAL programmer is meant to only be able to generate 7, thats right Seven, line of fully tested, documented, (stamped, lost, signed and buried in peat for 3 months, then burnt as fire lights ^_^), per day... I am being stupid or does that number seem a tad _low_?
-
I heard, somewhere, that a REAL programmer is meant to only be able to generate 7, thats right Seven, line of fully tested, documented, (stamped, lost, signed and buried in peat for 3 months, then burnt as fire lights ^_^), per day... I am being stupid or does that number seem a tad _low_?
I've heard 10. However this includes days where you produce no code because you are doing design, documentation, in meetings, testing, etc. Some of that time can be better allocated to someone else (ie the test group, and the documentation group) who doesn't write code, but counts as a programer because the programer would do this work in other companies. Writing code is 10% of the total engineering. 50-60% is test, 10-20% is design. 10% meetings and other buisness functions, 10% documentation. I can write 500 lines of code on a good day. However I then need to set those lines of code. Testing is 5 times the effort already, so I've already down to 100 lines a day! Just to get the first version out I'm looking at an average of 50 lines per day! Once customers have a few versions out and it is stable I will be doing even worse because I need to understand the bug and get it released, which will really turn my average down.
-
I heard, somewhere, that a REAL programmer is meant to only be able to generate 7, thats right Seven, line of fully tested, documented, (stamped, lost, signed and buried in peat for 3 months, then burnt as fire lights ^_^), per day... I am being stupid or does that number seem a tad _low_?
I'm surprised people still talk about measuring output by lines of code. Does
if (x) blah();
count as 1, and
if (x)
blah();Count as 2? What about whitespace? Comments? Is a programmer more productive if we writes lots of code, or if he researches and uses 3rd party components and ends up writing a system that uses less lines of code? Is a programmer who writes lots of code and no documentation more of a REAL programmer than one that spends a lot of time writing up the docs? cheers, Chris Maunder