C programming productivity metrics

[menchar]

Does anyone know of a source for productivity metrics for C programming?  I don't need a tool per se (I like using the free version of the Campwood SourceMonitor tool), I'm looking for the empirical data itself.  I can imagine various entities might guard this information, if they have it at all.

How would someone know if you're a "good" C programmer when using CVI, or not?  I've always heard that in C the paradigm is "trust the programmer" to do the right things, since the language is so unconstrained.

I'm trying to characterize CVI development productivity for a software group and even if I measure it I'm wondering how I would go about determining if it's good or bad compared to other tools, other companies, etc.  I guess if I realy wanted to get carried away I'd compare it to other programming languages (e.g. C#, C++) used for scientific programming.

[ebalci]

Hi,

If you need a one-time assesment you may consider forcing the members of the group to take the CCVID exam.

But I do not know of an automated tool to "grade" a code.

However, there are some tools, like CrystalFlow, that can create a flowchart representation of the C code.

I never used it. I recently downloaded the demo version but did not install and tried it yet.

Maybe examining such a chart is easier than examining various C files in a project to evaluate certain criteria.

[menchar]

Thanks for the reply.  Well, knowing how to use CVI is certainly part of it, but I would think proficiency with CVI would be reflected in work / time measurements.

I've used SourceMonitor to generate metrics.  What I am looking for is data that would help me know if  our C programming productivit is good, or not, when using CVI.  It would be interesting to compare CVI development projects to each other, within and without the company, as well as compare CVI development using C Vs. C development using some other IDE (Visual Studio?).  I just don't know where I would get that type of info.

[SLocke]

The only source code metric that I've ever heard of to measure the "goodness" of code/programmers is defects (bugs) per kloc (kilo- lines of code).  Eg, 5 defects per kloc is 5 "bugs" per 1000 lines of code.....

It's a very subjective measurement (what's a defect?), and the obvious drawback is that you can't pre-evaluate a project because you have to know about the bugs to get the metric.

I've heard of companies computing this for each programmer (based on checking repositories for the lines of code generated, and checking against bug reports/defects in those specific lines).....This would give you a historical trend for an ongoing project or programmer....

FYI - the space shuttle has a rating of 0 bugs per 500,000 lines of code....most contract programming is around 5-15, and most well tested windows applications are below 1...