The standard suggests that it should be possible to propagate errors between a subprogram caller and a called subprogram via the property Deployment_Properties::Actual_Subprogram_Call. In section E.7.3, Semantics (2), there is a bulleted list indicating the possible propagation paths. The 15th item reads, "a subprogram caller to every called subprogram (expressed by subprogram access connections or call binding (and the opposite direction)". As a side note, this line has two opening parentheses, but only one closing parenthesis.
First of all, I am assuming that the term "call binding" implies the use of the property Actual_Subprogram_Call. In such a case, it is not clear what the syntax of the propagation on the subprogram should be. Should this be an access propagation, a bindings propagation, or something else?
I would suggest that we remove this line from the standard and choose not to support propagations across call bindings.
brlarson
commented
1 year ago
The standard suggests that it should be possible to propagate errors between a subprogram caller and a called subprogram via the property
Deployment_Properties::Actual_Subprogram_Call. In section E.7.3, Semantics (2), there is a bulleted list indicating the possible propagation paths. The 15th item reads, "a subprogram caller to every called subprogram (expressed by subprogram access connections or call binding (and the opposite direction)". As a side note, this line has two opening parentheses, but only one closing parenthesis.First of all, I am assuming that the term "call binding" implies the use of the property
Actual_Subprogram_Call. In such a case, it is not clear what the syntax of the propagation on the subprogram should be. Should this be an access propagation, a bindings propagation, or something else?I would suggest that we remove this line from the standard and choose not to support propagations across call bindings.