[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[gpsd-dev] Major progress!

From: Eric S. Raymond
Subject: [gpsd-dev] Major progress!
Date: Mon, 9 Feb 2015 07:06:39 -0500 (EST)

Test termination is now deterministic and does not use a timeout -
CLOSE_DELAY is gone.  Building on this, blocking I/O has been
reinstated.  That should solve the select-buzzing problem naturally.

(The way this was done is that the test framework inserts a special
packet "# EOF" at the end of each test load which the daemon interprets as
a command to shut down the device. This cannot happen until the last
real data from the test load has been processed.)

WRITE_PAD still exists but on my system can be zero without problems.
It is quite possible that, with blocking I/O, it is no longer needed
anywhere. Please test this.

The port-zero allocation logic, inadvertently backed out, has been
restored. This should allow tests to run concurrently with a
production gpsd.

This is major progress. Remaining active issues:

1. Scrolling glitch in gpsmon direct

2. The ipcrm invocation in SConstruct needs to be rewitten portably.

3. Hal's gpsfake install glitch.

4. Still not known why tests fail on Frank Nicholas's Gentoo installation. 

5. Extra time-delta reporting for NTP. (Probably post-3.12.)

reply via email to

[Prev in Thread] Current Thread [Next in Thread]