setTimeStep is now compatible with a 'writeControl adjustableRunTime;'
setting in the systemControlDict. If 'adjustableRunTime' is selected
then the time-step values set by this function object will not be
exactly as specified, but write intervals will be matched exactly.
All function object time adjustment is now done during the execute
methods, so the specific setTimeStep hooks have been removed.
All function objects now re-read as a result of run-time modifications
to the system/controlDict.
Function objects that write log files (via the logFiles class) will now
generate a new postProcessing/<funcName>/<time> directory as a result of
either restart or run-time modification. Log files will therefore never
be overwritten by restart or run-time modification, except for when a
case is restarted at the same time as a previous execution (e.g.,
repeated runs at the start time).
Given that logFiles now writes the log files into postProcessing/<func
name>/<time> it is no longer useful to add '_<time>' to the log file name in the
case that the file already exists without this extension.
This is used to set the directory name for the results of the functionObject, if
not specified a unique name is generated automatically from the function type
and argument list, e.g.
#includeFunc patchAverage(name=inlet, fields=(p U))
writes surfaceFieldValue.dat in postProcessing/patchAverage(name=inlet,fields=(pU))/0 and
#includeFunc patchAverage(funcName=inlet, name=inlet, fields=(p U))
writes surfaceFieldValue.dat in postProcessing/inlet/0.
This avoids attempting to write temporary fields before they have been created.
The executeAtStart can be set to 'yes' in the writeObjects dictionary or the
functions
{
#includeFunc writeObjects(executeAtStart = yes, <field1>, <field2>...)
}
Additional flexibility for handling of field arguments has been extended
to dictionary lists of field settings, as well as word lists of field
names. This means that the following syntax is now supported:
postProcess -func "fieldAverage(p, U { prime2Mean on; }, T)"
postProcess -func "fieldAverage(fields=(p U { prime2Mean on; } T))"
Function object argument parsing now takes all "field", "fields" and
"objects" arguments and combines them into a single list of
fields/objects that the function should operate on. This means that the
following postProcess executions are now all equivalent and function as
expected:
postProcess -func "
flowRatePatch
(
name=outlet,
phi,
alphaRhoPhi.air,
alphaRhoPhi.particles
)"
postProcess -func "
flowRatePatch
(
name=outlet,
fields=(phi alphaRhoPhi.air alphaRhoPhi.particles)
)"
postProcess -func "
flowRatePatch
(
name=outlet,
objects=(phi),
alphaRhoPhi.air,
field=alphaRhoPhi.particles
)"
As are the following:
postProcess -func "mag(U.air)"
postProcess -func "mag(field=U.air)"
Rather than being tied to the Time class the dlLibraryTable libs is now a global
variable in the Foam namespace which is accessable by any class needing to load
dynamic libraries, in particular argList, Time and codeStream.
When selecting fields with complex automatically generated names it is
cumbersome to escape all the regular expression meta characters and more
convenient to specify the name as a literal string, the new regExp option
provides the ability to switch the regular expression support and defaults to
true for backward compatibility.
For example to cache and write out an intermediate temporary field generated by
the kOmegaSST model:
cacheTemporaryObjects
(
"((interpolate(((1|((1|(1|A(U)))-H(1)))-(1|A(U))))*snGrad(p))*magSf)"
);
functions
{
#includeFunc writeObjects(regExp=off, "((interpolate(((1|((1|(1|A(U)))-H(1)))-(1|A(U))))*snGrad(p))*magSf)")
}
The regExp option can also be specified in a writeObjects dict as
regExp off;
Additionally regular expression support has been added to the includeFunc
argument parsing.
When operating on multi-region cases, function files are now searched
for in the following order:
- First the $FOAM_CASE/system/<region> directory is searched
- Second the $FOAM_CASE/system directory is searched
- Finally a search is done through the preconfigured functions in
$WM_PROJECT_DIR/etc
Previously, in a multi-region case, preconfigured objects were
considered before those in the case's (non-region) system directory.
This was considered counter-intuitive. Functions residing in the system
directory should always take precedence, whether they are in the region
subdirectory or not.
This change formalises the usage of the "log" keyword in function
objects. By default, logging to stdout is activated when running
"postProcess" or "<solver> -postProcess", and deactivated when a
function is being executed as part of a run.
This behaviour can now be overridden in the function object dictionary
when operating in either mode.
As of 2ebed5ec, functonObjectList::start contains run commands that
trigger if some function objects are executable at time-zero. A call to
this method in functionObjectList::New has been removed (and replaced
with functonObjectList::read) as it was causing postProcess and <solver>
-postProcess to execute twice; once on construction, and once at the
selected initial time.
This change also fixes lookup errors generated by postProcess which
occured because fields (specified with -field or -fields arguments) are
not read and registered prior to function object construction.
Usage of functonObjectList::start within Time::run has not been altered.
By default most functionObjects now execute and write at the start-time except
those that perform averaging (fieldAverage, dsmcFields) which cannot be executed
until the end of the first time-step. There is also a new optional
functionObject dictionary entry "executeAtStart" which defaults to true but can
be set false if the execution and results of the functionObject are not required
or appropriate at the start-time.
A result of this change is that time logs of forces, sampling etc. now include a
values for time 0.
If the functionObject requires an object list rather than a field list the
non-named arguments are now inserted into the object list, for example
functions
{
#includeFunc writeObjects(kEpsilon:G)
}
which is equivalent to
functions
{
#includeFunc writeObjects(objects = (kEpsilon:G))
}
Both the functionObject call context (the command line for postProcess, and the
controlDict path for run-time post-precessing) and the configuration file
context where the arguments are substituted are now printed in the error
message, e.g.
postProcess -func 'patchAverage(name=inlet, ields=(p U))'
generates the message
--> FOAM FATAL IO ERROR:
Essential value for keyword 'fields' not set in function entry
patchAverage(name=inlet, ields=(p U))
in command line postProcess -func patchAverage(name=inlet, ields=(p U))
Placeholder value is <field_names>
file: /home/dm2/henry/OpenFOAM/OpenFOAM-dev/etc/caseDicts/postProcessing/surfaceFieldValue/patchAverage from line 13 to line 17.
and with the following in controlDict
functions
{
#includeFunc patchAverage(name=inlet, ields=(p U))
}
generates the message
--> FOAM FATAL IO ERROR:
Essential value for keyword 'fields' not set in function entry
patchAverage(name=inlet, ields=(p U))
in file /home/dm2/henry/OpenFOAM/OpenFOAM-dev/tutorials/incompressible/pimpleFoam/RAS/pitzDaily/system/controlDict at line 55
Placeholder value is <field_names>
file: /home/dm2/henry/OpenFOAM/OpenFOAM-dev/etc/caseDicts/postProcessing/surfaceFieldValue/patchAverage from line 13 to line 17.
adding support for argument substitution into sub-dictionaries for
e.g. pressureDifferencePatch, white space before, in and after the argument list
and continuation lines, for example:
functions
{
#includeFunc flowRatePatch(name=inlet)
#includeFunc flowRatePatch ( name = outlet )
#includeFunc pressureDifferencePatch \
( \
patch1 = inlet, \
patch2 = outlet \
)
#includeFunc yPlus
#includeFunc residuals
}
Currently these deleted function declarations are still in the private section
of the class declarations but will be moved by hand to the public section over
time as this is too complex to automate reliably.
Replaced all uses of complex Xfer class with C++11 "move" constructors and
assignment operators. Removed the now redundant Xfer class.
This substantial changes improves consistency between OpenFOAM and the C++11 STL
containers and algorithms, reduces memory allocation and copy overhead when
returning containers from functions and simplifies maintenance of the core
libraries significantly.
This clarifies the purpose which is to indicate that the object should be read
or written on this particular processor rather than it is or is not valid.
for consistency with WM_PROJECT. Now "etc" files are assumed to be in etc
sub-directories of WM_PROJECT_SITE and WM_PROJECT_INST_DIR allowing other files
to be stored in those directories. The search order is now:
Search for files from user/group/shipped directories.
The search scheme allows for version-specific and
version-independent files using the following hierarchy:
- \b user settings:
- ~/.OpenFOAM/\<VERSION\>/
- ~/.OpenFOAM/
- \b group (site) settings (when $WM_PROJECT_SITE is set):
- $WM_PROJECT_SITE/\<VERSION\>/etc/
- $WM_PROJECT_SITE/etc/
- \b group (site) settings (when $WM_PROJECT_SITE is not set):
- $WM_PROJECT_INST_DIR/site/\<VERSION\>/etc/
- $WM_PROJECT_INST_DIR/site/etc/
- \b other (shipped) settings:
- $WM_PROJECT_DIR/etc/
\return The list of full paths of all the matching files or
an empty list if the name cannot be found.
Optionally abort if the file cannot be found.
Optionally stop search after the first file has been found.
This change was proposed and agreed by the sponsors of the OpenFOAM project on
the OpenFOAM Hub, see https://openfoam.org/maintenance/
for
db/functionObjects/timeControl/timeControl.H: timeControls
primitives/bools/Switch/Switch.H: class switchType
primitives/strings/fileName/fileName.H: fileType
primitives/strings/wordRe/wordRe.H: compOption
