Start clearing up annotation syntax for functions. #3617
Conversation
|
I updated the description to make it a link to the issue #3495 for an easy lookup. |
There was a problem hiding this comment.
Suggesting minor change that will make this more compatible with #3621.
Co-authored-by: Henrik Tidefelt <[email protected]>
|
Language group: Describe these in text (not as default), and then later revisit all. |
Also provide smarter default for LateInline.
This has now been implemented. @henrikt-ma |
chapters/functions.tex
Outdated
| @@ -1719,6 +1722,8 @@ \section{Function Inlining and Event Generation}\label{function-inlining-and-eve | |||
| Late inlining is especially useful for differentiation and inversion of functions; for efficiency reasons it is then useful to replace all function calls with identical input arguments by one function call, before the inlining. | |||
| \end{nonnormative} | |||
|
|
|||
| The default for late inlining is tool-specific, in particular tools may automatically delay inlining to differentiate or possible invert the function before inlining it. | |||
|
|
|||
There was a problem hiding this comment.
Above, the default was described after giving the meaning of Inline = false. It would make sense to be consistent and describe the default after the meaning of LateInline = false.
chapters/functions.tex
Outdated
| @@ -1719,6 +1722,8 @@ \section{Function Inlining and Event Generation}\label{function-inlining-and-eve | |||
| Late inlining is especially useful for differentiation and inversion of functions; for efficiency reasons it is then useful to replace all function calls with identical input arguments by one function call, before the inlining. | |||
| \end{nonnormative} | |||
|
|
|||
| The default for late inlining is tool-specific, in particular tools may automatically delay inlining to differentiate or possible invert the function before inlining it. | |||
There was a problem hiding this comment.
Trying to fix some language:
| The default for late inlining is tool-specific, in particular tools may automatically delay inlining to differentiate or possible invert the function before inlining it. | |
| The default for late inlining is tool-specific. | |
| In particular, tools may automatically delay inlining in order to take advantage of function annotations for derivatives and inverses. |
There was a problem hiding this comment.
ok, handled both of the ones above.
| \end{lstlisting}\end{synopsis} | ||
| \begin{semantics} | ||
| Has only an effect within a function declaration. | ||
|
|
||
| If {\lstinline!true!}, the model developer proposes to inline the function after the function is differentiated for index reduction, and before any other symbolic transformations are performed. | ||
| This annotation cannot be combined with annotations {\lstinline!Inline!} and {\lstinline!LateInline!}. | ||
| The default is to not perform this specific inlining. |
There was a problem hiding this comment.
How about swapping order with the previous sentence, and then have a paragraph break before the rule about combining annotations?
There was a problem hiding this comment.
Should be done now.
It is intended to work towards close #3495 - but it doesn't handle all of the annotations as some are problematic.
Note that I deliberately added default values in some cases, I believe they are the correct ones.
It might be that we should add more, or view this as enough to close it - and later think about the rest.
The problem with the other ones are:
smoothOrder(normallyConstant=x)=2(I have might have missed some that could be handled.)