Clarify impure in systems of equations. #3571
Conversation
chapters/functions.tex
Outdated
| Comment: The semantics are undefined if the function call of an impure function is part of an algebraic loop. | ||
| This can occur for the deprecated case of external functions assumed to be impure, or when using \lstinline!pure(expression)! to call an impure function in a pure function. |
There was a problem hiding this comment.
| Comment: The semantics are undefined if the function call of an impure function is part of an algebraic loop. | |
| This can occur for the deprecated case of external functions assumed to be impure, or when using \lstinline!pure(expression)! to call an impure function in a pure function. | |
| Comment: The exact semantics are undefined if the function call of an impure function is part of an algebraic loop. | |
| Specifically the number of calls with external side-effects is unspecified. | |
| However, for impure functions where the outputs only depend on the inputs the algebraic loop should be solved correctly. | |
| This can occur for the deprecated case of external functions assumed to be impure, or when using \lstinline!pure(expression)! to call an impure function in a pure function. |
I believe this was the intended idea.
Not a general undefined-behavior in C-terms (with nasal demons etc), but just the pragmatic point that you cannot solve an algebraic loop with an impure function that e.g., returns the current time in nano-seconds, and that you have no control of the number of external side-effects generated.
That's why it is non-normative.
There was a problem hiding this comment.
How about speaking of side-effect semantics? Wouldn't that then be nicely elaborated by the outputs-only-depending-on-inputs rule?
There was a problem hiding this comment.
Ok, makes sense.
There was a problem hiding this comment.
Have now done that.
chapters/functions.tex
Outdated
| @@ -232,6 +232,7 @@ \section{Pure Modelica Functions}\label{pure-modelica-functions} | |||
| If a function is declared as \lstinline!impure! any function extending from it shall be declared as \lstinline!impure!. | |||
| \item | |||
| A deprecated semantics is that external functions (and functions defined in Modelica directly or indirectly calling them) without \lstinline!pure! or \lstinline!impure! keyword are assumed to be impure, but without any restriction on calling them. | |||
| Similarly they can be part of systems of equations without restrictions. | |||
There was a problem hiding this comment.
If we manage to write the text clearly, there shouldn't be a need for this sentence, as the systems-of-equations rule should be perceived as one of the calling restrictions.
There was a problem hiding this comment.
The main issue which I still find unresolved is that we both have
- It is an error if an impure function call is part of a systems of equations (including linear systems), even if called in agreement with the restrictions above.
and
- …if the function call of an impure function is part of an algebraic loop. … This can occur…
Could we avoid this contradiction by a more careful formulation of 1, something like this:
Inside systems of equations (including linear systems), an impure function call is only allowed in the deprecated case of an external function neither marked
pureorimpure, or whenpure(…)is used to bypass purity checking. The system of equations shall then be solved as if all occurring functions were pure, but the side-effect semantics of the impure function calls are undefined. Specifically, the number of times an impure function is called when solving the equations is undefined.
Co-authored-by: Henrik Tidefelt <[email protected]>
|
@henrikt-ma this should now handle the comments. |
|
@henrikt-ma is it ok now? |
No description provided.