- Allowed write buffers to become very large in asynnet. Thus in order to allow for the transfer
of large images. tested ub to 2kx2k, 16MB
- Handle lack of space in the write buffer more gracefully: just skip the image
- Fixed a bug in processnode which hangs SICS when processnode is called from multiple
clients on the same node
- Fixed a bug in multicountsersec which caused thrashing of the Tcl results which caused ugly
error messages when calling the transfer script. Now a copy of the connection is made
before calling the transfer script
Status codes are now determined by a special task which inspects SICS for what
is going on. Before this was a global variable which caused conflicts when
multiple instances in the code tried to set it.
- In network.c and asynnet.c code was added to close the socket when a
connect failed. Otherwise the socket would leak and the process
runs out of file descriptors.
- Multicountserc now forcefully handles the counting status. Before I had
the status flickering between counting and nobeam because the HM has no way to
diagnose nobeam
- In motorsec.c the pDrivInt->iErrorCount was not reset to 0 properly. Leading to death.
- Removed the stack smashing test flag for the time being.
- Arrayutil was not summing correctly
- Added interest to motorsec in order to support old status clients
- core dump because of bad free in devser
- fixed transferring of count time in multicounter
- added stack debugging to sllinux_def
- Modifications to tasdrive to wait for finish of monochromator properly
There is a rare condition when SICS has frehsly stared up and on the first call
to a coordinated driving, like in TAS, something fails. Then the group has ID 0,
the default and that one is always active. This has been fixed by starting valid
groups at 7, defining 0 ans IDUNDEFINED and checking in isTaskGroupRunning for
IDUNDEFINED
Then there is an issue in tasdrive. I added a call to CheckStatus when starting the
mono. This is necessary for the eiger monochromator as with that one starting is deferred
to the CheckSttaus function. The reason is that in other use cases, the looser might want
to drive a2 and a2w at the same time. This can only correctly be accomodated by deferring
the caclulations to the CheckStatus phase.
- Added a processnode methos to scriptcontext. processnode waits for the
scriptchain of a node to finish.
- Fixed a bug in sicsget
- Made histmemsec dim and rank manager privilege. To allow chnage at runtime.
Is required for SANS
- Fixed some issues with multicountersec, mostly relating to passing things through
in a sensible way.
- Updated motorsec.c to work with a client based driver
SKIPPED:
psi/polterwrite.c
psi/tabledrive.c
psi/tabledrive.h
a server hangs up diretcly after a message.
- Some output was suppressed while tracing, fixed
- Make ready to compile both with Ubuntu 32 and 64 or /usr/local
based ON in general. Just in psi/makefile_linux the top include has
to be adapted
- Removed epics drivers from SICS. Still in source to be reenabled when
needed
- Added FindMotor
- Added sput, sappend and slab to nxscript using the new sicsget module
- Fix to sget to fix the "mot softzero" problem
SKIPPED:
psi/make_gen
psi/psi.c
psi/sinqhttpopt.c
psi/spss7.c
- Added initial version of sicsget. This is a more generalised way of
reading and writing SICS data wherever it is. The thing is extendable if
reading something the current way is to slow. This has both a C interface
and an interpreter interface.
- Added a multicountsersec to teplace hmcontrol and multicounter
- Fixed a case sensitivity bug in haddcheck
- Made oscillate work with second generation motors for POLDI
- Added a time stamper to trace. Now there will be time stamps in trace files which allow
to correlate things from the master log with the trace.
- Updated polterwrite.
- Updated testprot to work with the behave test
SKIPPED:
psi/make_gen
psi/polterwrite.c
psi/psi.c
psi/sputterprot.c
- In support of HMsec, allow the count command in countsec to run without a
preset parameter. If this is missing the old, stored preset is used.
- Removed a check which causes initialisation of the HM from a batch file fail
for POLDI
- fixed @nl@ encoding of \n problem with certain kinds of batch files generated by GTSE
- Fixed a bug with summing on hitsmemsec. nxdataset now uses 64 bit int for dimensions.
- Fixed motorsec to remove itself properly from a parent when linked. Please note
that there is still the limitation that a second generation motor can only be linked
once in a Hipadaba tree
- Extended subsampling a HM for second generation HM
- Added a hscriptnotify. This calls a script on the update of a node
- POLDI has been rebuilt, chnages to data file writing
- For testing, sinqhttopt now understands a special command, byteswapoff,
which switches off byte swapping. Required when the HM ius run on an Intel box.
SKIPPED:
psi/polterwrite.c
psi/sinqhttpopt.c
- Added error fields to hwardware objects: motor, counter, histmem
- Optimised sinqhttpopt
- Added haddcheck which adds checks to hipadaba nodes. The current implementation
only checks for selctions agaisnt the values property. Expand when more checks are
required.
SKIPPED:
psi/polterwrite.c
psi/sinqhttpopt.c
psi/tasinit.c
