I have now fixed:
# caget from empty arrays no longer returns nonense in first element but 0.
Warning: This will break calling dbr_size_n(TYPE,COUNT) with COUNT=0 when meaning 1. Base does not do this but 3rd pary software may.
# Array records reading empty arrays now report 0 bytes read (NORD or similar)
Fixed for aai, wavevorm, subArray, aSub. There may be more (compress?) but I have not checked yet.
# dbpf can now put arrays, including empty arrays
# caput can now put empty arrays
# scalar records reading from empty arrays via db link now get INVALID/LINK alarm.
Does not work yet for CA links.
I will be on vacation next week (from now on).
I am not creating a mrege request yet as this is not complete.
I have now fixed: n(TYPE, COUNT) with COUNT=0 when meaning 1. Base does not do this but 3rd pary software may.
# caget from empty arrays no longer returns nonense in first element but 0.
Warning: This will break calling dbr_size_
# Array records reading empty arrays now report 0 bytes read (NORD or similar)
Fixed for aai, wavevorm, subArray, aSub. There may be more (compress?) but I have not checked yet.
# dbpf can now put arrays, including empty arrays
# caput can now put empty arrays
# scalar records reading from empty arrays via db link now get INVALID/LINK alarm.
Does not work yet for CA links.
I will be on vacation next week (from now on).
I am not creating a mrege request yet as this is not complete.
I took the change for 1-element-constant input links out of this branch https:/ /code.launchpad .net/~dirk. zimoch/ epics-base/ +git/epics- base/+ref/ fix_zero_ size_arrays