Re: testing status

[Greg Troxel]

"Gary E. Miller" <address@hidden> writes:

>> gpsd:ERROR: shmget(0x47505344, 24024, 0666) for SHM export failed:
>> Invalid argument
>
> I only get that when I try to run gpsd as a normal user.  Did you
> forget to start as root?

maybe.  will have to try again.  But I'd expect EPERM, not EINVAL.

It would be nice (I realize it's a fair bit of work) to enhance the
error messages to relate them to what gpsd is doing semantically rather
than (only) tell the user about syscalls in the implementation.  I guess
also I can't tell (without reading the sources) if this resulted in
disabling something or if the code side some plan B.

>> gpsprof (100 samples) works fine.  A nit is placement of the RCC
>> 266-01 labels, but it seems improved from 3.19.
>
> Moving labels arounnd in gnuplot is a PITA.  Also very
> "terminal" and font dependent.  Patches welcome.

I am not surprised by your comment...

>> So modulo gpsmss/qt building with libraries not in /usr,
>
> What is that?

I meant that on my system, as is normal on operating systems that have a
notion of base system and other things, things like gtk, gtkmm, qt, are
built with --prefix=/usr/pkg.  When those are all in /usr, which happens
on most Linux systems given the different approach to handling
base/extra (all in /usr, and labels on packages), then a class of bugs
about not passing -L/-R become invisible.

I do not want to dig into this pre-release...