Ensure that when the terrabyte option is passed in it is treated as a numeric (float) value instead of a string.  Multiplying by a string crashes the program.

- updated rcg release to 4.0.1

to same changes in map.c and mapApp.c after card search pattern was
changed to allow DAC models to appear in any order on the bus.

without an I/O chassis.  Not yet completed.

and time not yet being transmitted.

	- Switched module mapping search pattern in map.c mapApp.c.
	- Added card number info into mapping:
		- Change to ADC/DAC drivers
		- Change to cdsHardware structure
Fix for BIO count on user model STAT screen.
	- Send info after BURT to avoid race condition.
Fix for adding N as site/ifo designator for ANU.

Add build_edc_ini script

See merge request cds/advligorts!144

Still requires more testing.

Added enable_fir to Parameter block settings to allow models
other than sei to use FIR filters.

Rtcds enable option

Closes #122

See merge request cds/advligorts!143

Update rtcds to work with latest code

Closes #134 and #122

See merge request cds/advligorts!136

Erik von Reis authored 4 years ago and Jameson Rollins committed 4 years ago

rtcds lsmod now checks local_dc and cps_xmit processes. Also checks all dolphin modules. Closes #122

rtcds start and stop no longer require a model to be enabled, since we don't use "enabled" systemd units to start models at boot. Instead rtcds start and rtcds stop check to see if the model is active or not.

rtcds start now has a sleep between starts. This puts a gap between IOP starts and user model starts, sidestepping a race condition.

Also ignores errors when building every model, allowing for expected errors such as channels missing from the channels file. Closes #134

Standalone edc timing

See merge request cds/advligorts!141

Renaming delay_multiplier to delay_ms to make its purpose clearer.

The edc and local_dc operate off different (but close) start signals.  The edc starts based off of system time/gps, while the local_dc starts off of the iop model finishing.  These two clocks are different.  This adjusts the delay to start a little faster which has shown more stable data streams on the test stand.

This is a quick fix.  Better synchronization should be looked at after the release.

Allow a negative delay (defaults to -10ms).  This starts the data write early when negative.  This is done to mitigate the race between the streamers/local_dc and the edcu both
triggering off of the 16hz time cycles.  This makes the edcu work more like a fe in that it gets its data ready early.

with RCG V4.0 changes in kernel object generation
At this point, user models compile and run. Still need to work on
IOP.

Adding epics metrics to record buffer usage for trend buffers.

See merge request cds/advligorts!142

Shorten the standalone_edc's epics variable names

See merge request cds/advligorts!140

Adjusting the suffixes/names for the edcu epics variables to keep them managable.

Do not broadcast epics information if a prefix is not given.

Fix connection variable suffix in EzCaRead.pm

Closes #141

See merge request cds/advligorts!139

Close issue #141 - correct suffix of EPICS connection variable to _CONN in latter part of EzCaRead.pm to match value of header structure of same file

Standalone edc epic cas issues

Closes #139

See merge request cds/advligorts!138

 * The old pv code was not replaced when it was converted from a model.
 * Put the epics cas thread loop in with the diagnostics thread to not delay the edc
 * Updates ~8Hz
 * Consolidated all the hard coded channel names (and parts of names) into a class so
 that there is only once source of truth.