Re: [gpsd-dev] wsg_separation Issue

[Dave Hart]

On Tue, May 8, 2012 at 2:33 PM, Bywater, Rick (SA-1)
<address@hidden> wrote:
> Eric,
>
> I looked into the clock_gettime issue a bit more.  I am not sure why, but
> scons did not configure correctly.  I re-ran it with --config=force and it
> built clean.  So, I re-ran Coverity with the corrected build.  You can find
> the new results attached.

The second logfile had a summary with counts of each class of putative
defect identified, but what's really needed is the blow-by-blow for
each defect giving the line(s) of code involved and the relevant
identifiers (such as variable names for uninitialized use).

Coverity analyzes ntp-dev at one of its highest (pickiest) levels as
part of the Scan initiative, but we don't get detailed reports in text
logs or emails -- we have to sign into a website and drill into each
defect in turn.  It's not as bad as it sounds -- they serve up the
source code as well as the defects in a useful way, and allow
colloboration in comments and tracking (such as marking a defect as a
false positive).

My advice is to try again/harder to work with Coverity to setup or
restore access for key gpsd developers to their Scan portal website,
and if needed, get them tracking your head or at least your release
tarballs -- that is, using automation to build a snapshot every
release or every few days, the results of each archived to dig into
on-demand using the portal.

The human I have dealt with regarding Scan in the past, David Maxwell,
is no longer working for Coverity, based on the bounces
address@hidden elicits.  http://scan.coverity.com/about.html
has several email aliases you might try instead.

Cheers,
Dave Hart