CVS-commit comments since synApps 5.0
=============================================================================

support/sscan/configure/CONFIG_APP

Changes for 3.14.7
=============================================================================

support/sscan/configure/RULES.iocBoot

deleted; Obsolete in 3.14.7
=============================================================================

support/sscan/documentation/Scans.ppt

New file
=============================================================================

support/sscan/documentation/busyRecord.html

New file
=============================================================================

support/sscan/documentation/sscanRecord.html

Update for new data-storage-client handshake (AWAIT, AAWAIT).  Lots of
minor improvements.
=============================================================================

support/sscan/documentation/sscanReleaseNotes.html

New scanparm stuff

Updated for synApps 5.1
=============================================================================

support/sscan/sscanApp/Db/scan.db

don't specify TnCD=1, since this is already in the .dbd file
=============================================================================

support/sscan/sscanApp/Db/alignParms.db

AR field is now all uppercase
=============================================================================

support/sscan/sscanApp/Db/scanParms.db

AR field is now all uppercase
=============================================================================

support/sscan/sscanApp/Db/scanParms2D.db

AR field is now all uppercase
=============================================================================

support/sscan/sscanApp/Db/scanParms2Pos.db

AR field is now all uppercase.  Allow positioners and sscan record to all be in
different iocs.  Don't FLNK to get second scanparm record processed, because
then it can't know whether it's supposed to LOAD, or GO, or what.  Now have
OLOAD and OGO output links.  Deleted unused scanParms_np record.  Now first
scanparm record writes NP to second record, which writes to the sscan record.

scan parms for one-dimensional, two-positioner scan
=============================================================================

support/sscan/sscanApp/Db/scanParmsRemote.db

AR field is now all uppercase
=============================================================================

support/sscan/sscanApp/Db/scan_settings.req

added BSWAIT, ASWAIT, AWCT, AAWAIT, ACQT, ACQM, A1PB, A1CD
=============================================================================

support/sscan/sscanApp/op/adl/display_fields.adl

documentation displays
=============================================================================

support/sscan/sscanApp/op/adl/scan.adl

Show dstste (data-state) field

mostly cosmetic changes
=============================================================================

support/sscan/sscanApp/op/adl/scanParms2Pos.adl

Allow positioners and sscan record to all be in different iocs.
Use <sscan>.BUSY instead of xxx:scanActive

scan parms for one-dimensional, two-positioner scan
=============================================================================

support/sscan/sscanApp/op/adl/scan_detector_help.adl

documentation displays
=============================================================================

support/sscan/sscanApp/op/adl/scan_full.adl

Show dstste (data-state) field

mostly cosmetic; added doc displays; await
=============================================================================

support/sscan/sscanApp/op/adl/scan_help.adl

Add dstste (data-state) field

documentation displays
=============================================================================

support/sscan/sscanApp/op/adl/scan_more.adl

Show dstste (data-state) field

mostly cosmetic; added doc displays; await
=============================================================================

support/sscan/sscanApp/op/adl/scan_positioner_help.adl

documentation displays
=============================================================================

support/sscan/sscanApp/op/adl/scan_positioners.adl

mostly cosmetic; added doc displays; await
=============================================================================

support/sscan/sscanApp/op/adl/scan_triggers.adl

mostly cosmetic; added doc displays; await
=============================================================================

support/sscan/sscanApp/src/Makefile

saveData now runs on solaris and linux

publish sscanMenu.dbd
=============================================================================

support/sscan/sscanApp/src/recDynLink.c

Removed DEBUG macro; fixed format warning

recDynLinkAddInput was failing to decrement precDynLink->onQueue if
messageQueueSend() failed

Use epicsThreadGetStackSize(epicsThreadStackBig) instead of hard-coded (20000)
stack size

taskDelay(1) -> epicsThreadSleep(epicsThreadSleepQuantum))

Check CA-context pointer.

Added a bunch of debug statements

Input and output tasks are no longer independent; they share one CA context. 
recDynLinkGetCallback() needs the behavior of the recDynOut task (ca_flush, no
ca_pend_event), applied to a chid owned by the recDynInp task, so the simplest
thing was to join them.

Use ca_get_callback() to acquire data.  Previously, the sscan record
relied on monitors coming in before the callback resulting from a
ca_put_callback -- whose processing triggered the monitors.  This was
almost always ok, but CA buffer overruns and TCP retries could result
in the sscan record acquiring stale data.

checks for null pointers

Don't clear while an action message is on queue

Added diagnostics for uninitialized pointer problem

Changes for Windows compatibility
=============================================================================

support/sscan/sscanApp/src/recDynLink.h

Use ca_get_callback() to acquire data.  Previously, the sscan record
relied on monitors coming in before the callback resulting from a
ca_put_callback -- whose processing triggered the monitors.  This was
almost always ok, but CA buffer overruns and TCP retries could result
in the sscan record acquiring stale data.

Don't clear while an action message is on queue

Changes for Windows compatibility
=============================================================================

support/sscan/sscanApp/src/req_file.c

long int printf
=============================================================================

support/sscan/sscanApp/src/req_file.h

define our own max FILENAME_LENGTH
=============================================================================

support/sscan/sscanApp/src/saveData.c

Many changes; saveData now runs on solaris and linux

Use epicsThreadGetStackSize(epicsThreadStackMedium) instead of hard-coded
(10000) stack size

merged 3.13 AWAIT handshake into 3.14 version
Now scan record uses its double buffered data arrays,
and doesn't wait for saveData unless both buffers are
in use.
=============================================================================

support/sscan/sscanApp/src/scanparmRecord.c

New fields OLOAD and OGO -- links that write LOAD and
GO fields, so multiple scanparm records can be combined
into a single scan specification.  Check links before
using them.  Fields NP & MP changed from DBF_SHORT to
DBF_LONG
=============================================================================

support/sscan/sscanApp/src/scanparmRecord.dbd

Notes for last commit didn't mention new fields
OLOAD and OGO -- links that write LOAD and GO fields,
so multiple scanparm records can be combined into a
single scan specification.
Also, NP & MP changed from DBF_SHORT to DBF_LONG

Now include sscanMenu.dbd, instead of having separate definitions of
the menu fields
=============================================================================

support/sscan/sscanApp/src/sscanMenu.dbd

Lots of fixes to (new) recDynLinkGetCallback()-based data acquisition,
mostly affecting array-valued PV's.

Use ca_get_callback() to acquire data.  Previously, the sscan record
relied on monitors coming in before the callback resulting from a
ca_put_callback -- whose processing triggered the monitors.  This was
almost always ok, but CA buffer overruns and TCP retries could result
in the sscan record acquiring stale data.

New file, Tim forgot to commit
=============================================================================

support/sscan/sscanApp/src/sscanRecord.c

v5.26 Allow before-scan and after-scan links to write to their own sscan record.

ExportAddress for volatile variables.

Check for npts<=0 in init_record.

Reject illegal npts in special().

ajdLinParms() now checks that step-increment agrees with overall scan direction.
(Freeze flags for start, end, and width are ignored in this check.)

Abort and killing scans reworked to allow sscan record to wait for data-storage
client even if it doesn't wait for anything else, to abandon even that caution
if user insists, and to show user where it is in the process.

scan abort no longer clears .PAUS

Lots of fixes to (new) recDynLinkGetCallback()-based data acquisition,
mostly affecting array-valued PV's.

Updated the internal version number to 5.22

Use ca_get_callback() to acquire data.  Previously, the sscan record
relied on monitors coming in before the callback resulting from a
ca_put_callback -- whose processing triggered the monitors.  This was
almost always ok, but CA buffer overruns and TCP retries could result
in the sscan record acquiring stale data.

get enum strings; FLNK if scan fails, so ca_put_callback() completes;
call scanOnce() from special if we're waiting for saveData and AWAIT==0;
if limit trouble stops scan, hang instead of declaring done.

merged 3.13 AWAIT handshake, and array-trigger/array-read-at-end-of-scan
capabilities into 3.14 version
=============================================================================

support/sscan/sscanApp/src/sscanRecord.dbd

Include sscanMenu.dbd, so scanparm record can have identical menus.

AWAIT is no longer a PP field (now special() will call scanOnce() if
processing is needed)

merged 3.13 AWAIT handshake, and array-trigger/array-read-at-end-of-scan
capabilities into 3.14 version
=============================================================================

support/sscan/sscanApp/src/sscanSupport.dbd

Added recDynLinkDebug

added variable() calls for sscan-record volatiles

added registrar and variable statements for saveData

=============================================================================

support/sscan/sscanApp/src/xdr_lib.c
support/sscan/sscanApp/src/xdr_lib.h

ifdef vxWorks (first time running on Solaris and Linux ioc's)