Pretty constructors in java parse tree #2152
Conversation
kroening
requested review from
cesaro,
mgudemann and
peterschrammel
as code owners
|
|
||
| virtual void output(std::ostream &out) const = 0; | ||
| virtual void output( | ||
| const irep_idt &class_name, |
There was a problem hiding this comment.
This seems like something the methodt should store rather than something every user has to pass to output()?
There was a problem hiding this comment.
No, this would be an antipattern; its not unusual to require context for output, say think of indentation.
There was a problem hiding this comment.
I think it is reasonable to pass this in, I also think it is reasonable for the method to know what class it belongs to FWIW.
|
|
||
| bool is_constructor() const | ||
| { | ||
| return base_name=="<init>"; |
There was a problem hiding this comment.
Nooo - this re-duplicates the check that <init> is the name of constructors that I've painstakingly removed once! Either leave as is of remove the is_constructor in java_bytecode_convert_method
There was a problem hiding this comment.
I think that is_constructor in java_bytecode_convert_method can be entirely removed. I'd replace the call in java_bytecode_convert_method.cpp:381 to use this new method. The other two uses java_bytecode_convert_method.cpp:575 and 1327 can be replaced by type.get_is_constructor().
thk123
left a comment
thk123
left a comment
There was a problem hiding this comment.
Aside from previous review this looks OK. It would be good to have a test demonstrating this new behaviour.
|
|
||
| virtual void output(std::ostream &out) const = 0; | ||
| virtual void output( | ||
| const irep_idt &class_name, |
There was a problem hiding this comment.
I think it is reasonable to pass this in, I also think it is reasonable for the method to know what class it belongs to FWIW.
|
Closing as this appears to be superseded by #2160. |
No description provided.