Fix grammar around variadic function calls
Reporter: JoshDreamland | Status: closed | Last Modified: December 22, 2017, 05:41:08 pm Because 'V' is used to denote variadic functions, but then 'n' is substituted in for later code simplicity, 'V' needs to be treated like 'n' when determining token separation. This change lets
cprime
donate an alternate interpretation of c
, the token marker character, so this can be detected. RobertBColton
This resolves at least the issue that crept up in #1019 so far.
if z>=other.z3+other.z
ds_list_add(list,other.z3+other.z)
Now correctly parses to:
if(enigma::varaccess_z(int(self))>= enigma::varaccess_z3(int(other))+ enigma::varaccess_z(int(other)))
ds_list_add(enigma::varaccess_list(int(self)), (enigma::varargs(), enigma::varaccess_z3(int(other))+ enigma::varaccess_z(int(other))));
However choose(1,2,3);
in a script in an empty game throws an exception up to LGM but does not crash.
Copying constant names [0]
SYNTAX CHECKING AND PRIMARY PARSING:
1 Scripts:
choose ( 1 , 2 , 3 ) Second pass...
Thread-14:
java.lang.Error: Invalid memory access
at com.sun.jna.Native.invokeInt(Native Method)
at com.sun.jna.Function.invoke(Function.java:383)
at com.sun.jna.Function.invoke(Function.java:315)
at com.sun.jna.Library$Handler.invoke(Library.java:212)
at com.sun.proxy.$Proxy0.compileEGMf(Unknown Source)
at org.enigma.EnigmaRunner$CompilerThread.run(EnigmaRunner.java:663)
Also #214 is still an issue:
show_message(choose(a,b,c));
C:/ProgramData/ENIGMA/Preprocessor_Environment_Editable/IDE_EDIT_objectfunctionality.h:39:43: error: 'a' was not declared in this scope
show_message(choose((enigma::varargs(), a, b, c)));
^
C:/ProgramData/ENIGMA/Preprocessor_Environment_Editable/IDE_EDIT_objectfunctionality.h:39:46: error: 'b' was not declared in this scope
show_message(choose((enigma::varargs(), a, b, c)));
^
C:/ProgramData/ENIGMA/Preprocessor_Environment_Editable/IDE_EDIT_objectfunctionality.h:39:49: error: 'c' was not declared in this scope
show_message(choose((enigma::varargs(), a, b, c)));
^
``` gml
show_message(string(choose(1,2,3)));
show_message(string(choose(1,2,3)));
Becomes;
``` cpp
show_message(toString(choose((enigma::varargs(), 1, 2, 3))));
show_message(string(choose(a,b,c)));
Becomes:
show_message(toString(choose((enigma::varargs(), enigma::varaccess_a(int(self)), enigma::varaccess_b(int(self)), enigma::varaccess_c(int(self))))));
show_message(string(show_message(string(a))));
Becomes:
show_message(toString(show_message(toString(enigma::varaccess_a(int(self))))));
So.... LOOKS LIKE YOU FIXED IT!!!! But choose(1,2,3);
is still crashing all by its little self. I also need to test other games.
- FPS Example: Compiles and runs fine as always.
- Animation Platform Example (DarkAceZ gave it to me): Compiles and runs great, also lacking the graphics problems it had before oddly enough.
- Project Mario: Working great!
- AI Tutorial: Cool beans!
- Game of Life: Pi in the sky!
- Destructible blocks: Good to go!
- Platform engine from EDC: now compiles again so what crept up in #1019 is definitely gone.
No one has merged this, yet. Is there a reason not to?
I think we just wanted to fix it as best we could, as in, this is fine to pull, but
choose(...)
still doesn't work with numeric literals. All of the games I tested continue to work, and this PR fixes one game that broke because of #1019.
This is problematic because in #1019 I made all the data structure functions variadic, as they are in GMS.
ds_list_add(items, 1, 1, 2, 3, 4, 4, 4);
Meaning we should probably fix the last issue before we make a new release. If you want to do it now and then merge or merge and then do it later, it doesn't matter, just as long as it's fixed before a new release.
Leave a commentView this issue on GitHub
Please sign in to post comments, or you can view this issue on GitHub.