Re: Recent 1024 bug

[Gary E. Miller]

Yo Hal!

On Tue, 26 Oct 2021 00:13:10 -0700
Hal Murray <halmurray@sonic.net> wrote:

> I just scanned
>   https://gitlab.com/gpsd/gpsd/-/issues/144#note_633612324

Yup, that just will not die.

> I don't fully understand what went wrong or why, but it got me
> thinking about the testing setup.

Dyou you care to understand?  It would shed light on testing.

> Do we need to update some of the expected results occasionally?

We do.

> (rather than hack the code to reproduce the old results)

Wo do that to.

> Suppose there is a 1024 time warp between when the test data was
> captured and when the tests are run.

Yes, a continuing problem.

>  We would like gpsd to output
> new time (which will probably be in the future) rather than old time.

Why would we like that?  We tried that, it broke a lot of things.

>  That won't match the expected files which have old time.

Yup, which was part of the problem.

> So rather
> than hack the code,


Why not none of the above?  gpsd now uses the "# Date:" from the regression
ass the base date, so the regressions are always repeatable.

> why not update the expected file?

We tried that.  That is a lot of frequent updates, and sometimes past
2038 rollover which broke a ton of things.

GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        gem@rellim.com  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

pgpIPEFGpRHbT.pgp
Description: OpenPGP digital signature