* Fix typo in epicsEventWaitWithTimeout.
* Reword epicsEvent timeout documentation to reflect the actual behavior.
Negative values of timeout are handled the same way as zero values.
Change the documentation to reflect this behavior.
* fix(aao, aai): change VAL field type from DOUBLE[] to FTVL[]
* fix(aSub): change 'Input Value Fields' type from FT[A–U] to FT[A–U][NOA–NOU]
* fix(aSub): change 'Output Value Fields' type from FTV[A–U] to FTV[A–U][NOVA–NOVU]
* fix(waveform): change VAL field type from FTVL to FTVL[NELM]
* fix(aai): remove 'analog' from the record name
* fix(aao): remove 'analog' from the record name
* fix(aai,aao): remove 'Analog' from the record names
* Update normativeTypes to fix NTNDArray::getValueSize
* Update normativeTypes for Release Notes
---------
Co-authored-by: Andrew Johnson <anj@anl.gov>
This reverts commit e4fbd83c93.
It introduced a build problem when using INSTALL_LOCATION:
The system yacc and lex are used instead of the EPICS antelope and e_flex.
Tests killed by a signal should cause the .t script to exit with an
error code, otherwise its exit code will be 0, and the test harness
won't be aware of the unsuccessful exit.
This change also makes the test runner more robust, so attaching to a
running test (e.g. with gdb or strace) won't cause the .t script to
exit.
* Allow CA clients to determine the server protocol version
Adds a call to the CA client API that allows a client to determine the server's
protocol minor version number. This is needed to allow the ca-nameserver to
report a server's protocol version correctly to a client.
* ca_host_minor_protocol return for disconnected channels
ca_host_minor_protocol now explicitly returns CA_UKN_MINOR_VERSION for a disconnected channel.
OP_SYS_LDLIBS ends up on the GeSys object link command line, which ends
up breaking things. When linking GeSys objects, a symbol root is not
defined using -u or -e, which is required for --gc-sections to work.
These were defined as macros in osdMessageQueue.h, but declared as
prototypes in epcisMessageQueue.h public header. We need to implement
them as C functions for these to actually be usable.
* warn to stderr when un-setting inapplicable CPP flag on output links
* Warn and unset CP flag for output links
* use errlogprintf instead of fprintf for dbParseLink CP/CPP warning
* Pass source record name to dbParseLink for debug information when modifier is discarded
* make CP/CPP discard warning less verbose
* Include field name of affected record when warning of CP/CPP discard
If you overwrite INSTALL_LOCATION in CONFIG_SITE, then it would be the case that
TOP_CFG_CONFIGS would point to those from EPICS base and not from the module
that you are loading. Fetching these earlier fixes that issue.
* Allow testdbGetFieldEqual to check for NAN double and float values
* Print DBR_xxx names instead of numbers
* use %g instead of %e to format floating point values to improve readability
gcc knows __gnu_printf__ since version 4.4.0, not only for MinGW but for
all targets. As we would need to check the gcc version anyway to avoid
"unrecognized format function type" errors with older MinGW versions,
we can simply use it for recent gcc versions and not depend on MinGW at all.
