Support after delays on broadcast connections in C++ #553
Conversation
Codecov Report
@@ Coverage Diff @@
## master #553 +/- ##
============================================
+ Coverage 66.39% 66.41% +0.01%
- Complexity 3470 3474 +4
============================================
Files 132 132
Lines 22310 22311 +1
Branches 2885 2885
============================================
+ Hits 14812 14817 +5
Misses 6351 6351
+ Partials 1147 1143 -4
Continue to review full report at Codecov.
|
| @@ -177,7 +177,7 @@ class ASTUtils { | |||
| * if the port on the left or right of the connection involves a bank of reactors or a multiport. | |||
| * @param connection The connection. | |||
| */ | |||
| private static def boolean isWide(Connection connection) { | |||
| static def boolean isMultiportConnection(Connection connection) { | |||
There was a problem hiding this comment.
The name isMultiportConnection for this method seems misleading because the connection may not involve any multiports and the method could still return true.
There was a problem hiding this comment.
I found isWide also not very clear, because it could be multiple individual connections in case of o1, o2 -> i1, i2. Do you have a suggestion?
There was a problem hiding this comment.
How about isMultipleConnection?
There was a problem hiding this comment.
I renamed it to hasMultipleConnections, which seems better from a grammatical point of view.
| @@ -0,0 +1,27 @@ | |||
| target Cpp; | |||
|
|
|||
| reactor Source { | |||
There was a problem hiding this comment.
I think at least the Source reactor (but also a parametrized version of Sink) can be imported for a bit of a more minimalistic design. This is by no means a critical issue though.
There was a problem hiding this comment.
Yep, having better guidelines for writing tests and establishing a set of basic library reactors would be nice.
Currently, after delays do not work on broadcast connections (using the
(...)+syntax). In part, this is because theiteratedproperty of the connection is not preserved in our AST transformation. I fixed this, so that the newly generated upstream connection is also iterated if the original connection was iterated. This solves the issue for C++. However, a similar test written in C still fails, and I haven't found out why. I will push the test that I have tobroadcast-after-cbased on this branch, and will open a separate issue.This PR also does a few cleanups and adds C++ tests.