[time-nuts] Re: GPS D (i.e. gpsd) roll over bug..Oct 24th

[Hal Murray]

Magnus Danielson said:
> This is correct, but the bug re-introduces the 1024 wrapping issue,  which
> should have a basic safetynet after it. 

What do you mean by "safetynet"?

The gpsd project has a lot of checking.  It would be an interesting example 
for a software engineering course.  It's build time checking.  They have a 
large collection of output from various GPS devices and the corresponding 
output from gpsd.  They have a test harness to feed the samples to gpsd and 
compare the output with the expected output.  That catches a lot of 
fix-it-here and break-it-there type bugs.

In hindsight, it's easy to say "They shoulda...".  With a big budget you could 
do serious testing.  That's harder with volunteer projects.

-------

Occasionally, when I'm looking at a chunk of code, I get annoyed because there 
is so much extra code that it's hard to find and understand the mainline logic.

Sometimes that extra is cruft that didn't get cleaned up when things changed.  
Sometimes it's an obscure feature of little value.  Sometimes it is 
error/sanity checking.

It would be interesting to study how "extra" code influnces the introduction 
of bugs.


-- 
These are my opinions.  I hate spam.