Author Topic: Suggestion: coding behavior statistics  (Read 3809 times)

Offline Steven 'lazalong'

  • Multiple posting newcomer
  • *
  • Posts: 13
Suggestion: coding behavior statistics
« on: September 01, 2007, 09:22:22 am »
Hey

As a freelance programmer I need to account for the time I charge.
To do that correctly I need to have some statistics on how I work.

Some statistics that could be stored:
  • How much time was passed coding on a project?
  • How much per file, per project, per workspace?
  • How much code was done per day?
  • Do I type or simply stare at my screen (idle time)?
  • When are the period during the day I code the most?
  • Do I use often undo/redo aka trial and error ;) ?
  • Etc.

All this will help me determine how many day-man I will use for a project.
The features that can be added are certainly endless.

Anyway I know this is a very low priority request.
But I wanted to throw it to you and see if you find other useful statistics to add.

If you find it worthwhile we could make a SF request.
Bye
« Last Edit: September 01, 2007, 09:48:50 am by Steven 'lazalong' »
-------------------------------
OGE - www.opengameengine.org

Offline wwolf

  • Multiple posting newcomer
  • *
  • Posts: 34
Re: Suggestion: coding behavior statistics
« Reply #1 on: September 01, 2007, 11:58:05 am »
Well, I get paid for just staring at my screen.  I stare, stare and THINK. :)  I really think that what you want to measure is the time you have spent working on specific tasks, not specific files.  So I would add timing possibility to a hierarchical TODO list feature, rather than code files.  Sort of SOA style. :)

I am writing this because I have been thinking a lot about a program that would have this as one of its features.  What I mean is that this is not just a random thought, I am planning a program and I have thought about this a lot.  BTW I have spent many man-hours on that, yet I do not have a single line of code. :)  But my looking at Code::Blocks is part of the process of understanding what people want from an IDE.  And of course I would like to have an IDE that supports wxWidgets, so... ;)

Offline thomas

  • Administrator
  • Lives here!
  • *****
  • Posts: 3979
Re: Suggestion: coding behavior statistics
« Reply #2 on: September 01, 2007, 02:02:03 pm »
If you find it worthwhile we could make a SF request.
It sounds like an interesting idea, but if you make a request at Sourceforge, you're wasting your time. Nobody reads them, we have not used Sourceforge for well over a year (actually.... 2 years?).

Measuring actual productivity by what you type may be hard (as wwolf pointed out), but at least you could record what files you have open for how long, etc. to get a rough figure of what you did. This may help yourself optimize your work process (but it is not useable as a measure of payment).

There have been many other attempts at automated accounting of "work" in the past, for example people have been assessed on the number of code lines changed per week. However, 3 good and efficient lines of code may be a lot more work and are better than writing 1200 lines of bullshit code. Thinking for half an hour may be worth more than typing some rubbish that won't work. This is not taken into account by this form of assessment.
Also, studies have shown that if people are aware of this accounting measure, they will simply push their ranking by writing bloated (and inferior) code, which leads the assessment to absurdity.

In the end, the only effective measure for "work" is to define goals (milestones) that you must reach under the constraints of the design rationale. Whether by staring at the screen and finally writing 3 lines of code, or by constant trial and error, it is the outcome that is of interest.
"We should forget about small efficiencies, say about 97% of the time: Premature quotation is the root of public humiliation."

Offline Steven 'lazalong'

  • Multiple posting newcomer
  • *
  • Posts: 13
Re: Suggestion: coding behavior statistics
« Reply #3 on: September 01, 2007, 02:27:11 pm »
This may help yourself optimize your work process (but it is not useable as a measure of payment).
Yes - I agree with you it is the results that are a measure of payments.
But I explained it badly.
I want those statistics to know how much I demand at the signature of the contract.
"This project has x features which will need about y classes of that complexity and in my experience it will amount to z man-day. Hence I will ask this amount for the project."

Quote
However, 3 good and efficient lines of code may be a lot more work and are better than writing 1200 lines of bullshit code.
Yes again - this is to measure my performance!
Not to measure the work of others.

Anyway it was just an idea :)
-------------------------------
OGE - www.opengameengine.org

Offline Seronis

  • Almost regular
  • **
  • Posts: 197
Re: Suggestion: coding behavior statistics
« Reply #4 on: September 01, 2007, 06:04:50 pm »
Yes again - this is to measure my performance!
Not to measure the work of others.

That makes sense. First post it seemed as though you wanted to show the figures to someone as proof you have worked.  Using it for your _own_  record keeping process, rather than for someone _elses_ assessment of you, is a good use.