Determining the Application State to Transfer¶
In many cases, it is not necessary to transfer the entire state of your client application to initiate the server-side session appropriately. In some cases, you may not need to transfer any application state at all, because the arguments of the procedure call initiating the server-side session already provide all the information necessary to retrieve the data that is needed to successfully execute the request.
Specifying Case Content¶
To reduce the amount of data that is sent between the server and the client, or to prevent that data is transferred at all, you can provide the AIMMS PRO library with information about what data should be stored in the case files sent back and forth. The mechanisms are slightly different for projects using the data management style storing all cases in a single
.dat file, as opposed to the data management style where all individual cases are directly stored in separate files on disk. The latter option is only available in AIMMS 3.12 and later.
Specifying Case Types¶
For the single
.dat file data management style, you can instruct the AIMMS PRO library to use a specific case type by setting the following element parameters:
pro::ManagedSessionInputCaseType, denoting which case type should be used for data transfer from the client to the server
pro::ManagedSessionOutputCaseType, denoting which case type should be used for data transfer from the server to the client
If you do not set these two element parameters explicitly, the AIMMS PRO library will use the predefined case type
AllIdentifiers by default. This means that the values of all identifiers that do not have the
NoSave property set will be transferred back and forth. If you set any of the above element parameters explicitly to the empty element, AIMMS PRO will not create and transfer an input, and/or output case file respectively.
Specifying Case Content¶
For the data management style using separate files on disk for each case, case content is specified through a subset of the predefined set AllIdentifiers. You can specify the contents sent in input and output cases by filling the following sets:
pro::ManagedSessionInputCaseIdentifierSet, for input cases
pro::ManagedSessionOutputCaseIdentifierSet, for output cases
If not specified explicitly, both sets will default to AllIdentifiers. If you assign the empty set to any of these sets, AIMMS PRO will not create and transfer an input, and/or output case file respectively. Starting with AIMMS 4.50.1, AIMMS PRO will always remove the set AllDefinedParameters from the above sets before saving the case for performance reasons because of smaller files and lower transfer times. As of AIMMS 4.59.2 it is possible to change this behavior by indicating which identifiers will be removed by specifying that in the set:
which has as default the value ‘AllDefinedParameters’. Sometimes, e.g. when doing case comparisons, you do want to store the defined parameters in the case in order to have case comparison work correctly. In that case you can empty the above identifier set, resulting into actually saving AllIdentifiers into the case.
pro::ManagedSessionOutputCaseIdentifierSetneeds to be assigned during the solver session.
Make sure that the data of sets and parameters referenced in the index domain attribute of the variables computed server side and communicated via the output case to the client side, are also available client side or passed in the output case.
ReferencedIdentifierscan be used to find all referenced identifiers in a given subset of
AllIdentifiersand thus extend the identifier subsets pointed to by
We advise you to use those 2 sets, if you would like to optimize your data transfer time.
Optional action to be taken after case load¶
You can specify the action to be taken after a case is loaded client or server side. In the client session, the action to be taken after a case load is defined by the element parameter
pro::session::PostLoadResultCaseHook. For example, by setting:
pro::session::PostLoadResultCaseHook := 'postProcessComputedResults' ;
postProcessComputedResults will be executed by the
pro::delegateToServer completion callback procedures that load the result case, and by the load results button in the “Managed Requests” dialog. The element assigned to
pro::session::PostLoadResultCaseHook should reference a procedure without arguments.
pro::session::PostLoadInputCaseHook can reference a procedure to be executed in a server session after loading the input case.
Saving Log Files¶
By default, at the end of a server-side session, the messages.log file created during that session will be stored for inspection within the client session. If you do not want to have this log file saved, you can prevent this by setting the parameter
pro::session::SaveSessionMessages to 0.