Variant Size for Arrays and Matrices

The size of an array or matrix is determined when the element is created. In addition, ASCET allows the determination of variants in array/matrix sizes via system constants (including user-defined system constants).

The following rules apply:

Only system constants in the same component as the array/matrix can be used for variant size determination.
The system constants used for variant size determination must be of type limitInt, wrapInt, sdisc, udisc, or enumeration.
For an array/matrix of scope exported, system constants of scope exported or imported can be used for variant size determination.
For an array/matrix of scope local, system constants of scope local or exported or imported can be used for variant size determination.

If you use an array or matrix that has at least one dimension with a system constant size, keep the following in mind:

Index protection (see Code Generation Options, Protected Vector Indices) uses the value of the system constant instead of the max. size.
If the Resolve System Constants option in the ASCET options window, "Targets\<target>\Build" node of the selected target, is set to Generation Time, the initial value of the system constant is used to transform the variant-size array/matrix into a fixed-size array/matrix.
If the Resolve System Constants option in the ASCET options window, "Targets\<target>\Build" node of the selected target, is set to Compile Time, the following constraints are checked and the following happens.
- At least one value of the system constant must be included in the interval [1..maxSize].
- If the array/matrix scope is exported or imported, the scope of the system constant must be exported or imported, too.
- If the value range of the system constant includes values < 1 or > maxSize, a check is generated that issues an error message if the system constant value is outside the interval [1..maxSize]:
  #if (SC_name < 1 || SC_name > maxSize)
  #error The system constant SC_name must be between 1 and maxSize, because it is used as an array/matrix size of ArrayMatrix_name.
  #endif
  In addition, a warning of type WMdl651 is issued during code generation:
  Possible type mismatch in array max size: <name> and <[1,maxSize]> – will be checked at compile time
- During initialization, the max. size is set to the max. size defined in the model, the current size is set to the value of the system constant.
- The access methods for array/matrix lengths will return the value of the system constant.
If the Resolve System Constants option in the ASCET options window, "Targets\<target>\Build" node of the selected target, is set to Run Time, the following happens.
- During initialization, the max. size and the current size of the array/matrix are set to the max. size defined in the model.
- The access methods for array/matrix lengths will return the value of the system constant.
- The compatibility check for arrays/matrices changes: arrays/matrixes are only compatible if they use the same system constant.
  This is because it is not possible to ensure that different system constants always have the same value if they can be changed at run time.
- For experimental targets, the index protection generates a check against the lower and upper bound, because a change of the value of the system constant does not automatically change the current size of the array/matrix.
- For ASCET‑SE targets, a warning (WMdl28) is issued during code generation. By default, this warning is promoted to an error.
  If you ignore the global promotion, the index protection generates a check against 0 and min(<system_costant>, <max.size>), because the changed value of the system constant may be larger than the max. size.
Models that use normal arrays/matrices with variant size can only be exported to ASCET V6.2.0 or newer.
Models that use arrays/matrices with variant size as messages can only be exported to ASCET V6.4.0 or newer.
Models that use arrays larger than 2048 elements or matrices with one or both axes larger than 63 can only be exported to ASCET V6.4.3 or newer.
The generated initialization code for arrays/matrices with variant size is affected by the setting of the Generating Sparse Arrays target option; see Initialization of Composite Elements.