Daqd exports all PVs as doubles, changing the channel list checksum and data checksum to be doubles to match.  If this is not done, then a negative value is shown if the high bit is set when viewing the cps_recv version of the checksum, where the daqd shows a large positive number.

Update the daqd to read a channel list file/master file when the last line is not newline terminated.

Data crc's are calculated on cycle 15.

The channel list is sorted and given new sequence numbers the same way the daqd is.

Data crc's are calculated on cycle 15.

The channel list is sorted and given new sequence numbers the same way the daqd is.

The sort predicate was wrong on the data block crc sort, it could reach a state where it never completed.

The blocks should be checksumed in dcuid order.  This tracks the dcuid + crc, sorts them by dcuid and then does the checksum.

Update the epics more frequently than 1s to allow caget to work w/o timeouts.

Update the data crc on data cycle 0.

It had counted chans + tp, now it only counts channels.

Fix a bad condition which had the logic inverted.

Adjust the alarm low values to make sure 0 is good.

* Linked the admin interface into the dc_stats to allow clearing the crc counters.
* Fixed a bug in the crc logic which would not allow it to always mark a crc (dropped data) on a transition between valid data and no data.
* Added some tests.
* Added a emplace_with_timeout to the message queue to allow implementing the DCStats stop method without blocking forever if a background thread is not running.

* PRDCR_NOT_STALLED
* PRDCR_UNIQUE_DCU_REPORTED_PER_S
* DATA_RATE
* TOTAL_CHANS
* PRDCR_OPEN_TP_COUNT
* PRDCR_TP_DATA_RATE_KB_PER_S
* PRDCR_MODEL_DATA_RATE_KB_PER_S
* PRDCR_TOTAL_DATA_RATE_KB_PER_S

Cps_recv had not been properly tracking the last viewed cycle, so every time it looked at the shared memory it saw what it thought was a new message and sent (and resent) the data that was present in the buffer was re-broadcasted (in this case written into the mbuf in same spot multiple times).

Added a test case to the message queue that was used in helping to narrow down issues.

Initial draft at checks for crc of the data, missing data, bad config.

Moved the epics commits to the diagnostic thread.  Still need to resolve the epics work done in the main thread.

Removing duplicate copies of catch.hpp.

Adding support for double values in simple_pv.

See merge request cds/advligorts!176

on ADC timeout if time signal from master is lost.

Added newline at end of daq master file.  Fixed tests for checkdaqconfig.py

See merge request cds/advligorts!175

Add a total crc pv to the daqd, this sums all the crcs across the system.

See merge request cds/advligorts!174

The idea is to allow trending one channel to see if there were any crcs, then followup investigation could be done on individual systems.

Adds '{prefix}TOTAL_CRC_SUM' as a pv to the daqd.

Updated the daqd shmem receiver to print a message when it finds and uses a skipped cycle.

See merge request cds/advligorts!173

Track the previous gps second in the daqd shmem_receiver

Closes #189

See merge request cds/advligorts!172

Note, this treats any 0 dcu count cycle as a skip.

Closes #189.

Updated daqd error message when it aborts due to a time skip to show the first 10 dcus received

See merge request cds/advligorts!171

Set a max depth for the cps_xmit memory pool to 2s + some change

See merge request cds/advligorts!169