www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Post-mixin-expansion source output

reply Mithun Hunsur <me philpax.me> writes:
Hi all,

I was discussing this with Stefan Koch and a few other people at 
DConf - would it be possible to have a compiler switch for 
outputting the complete source file after the mixins have been 
expanded, similar to gcc -E?

I assume that mixin expansion occurs in the semantic pass, which 
might make this difficult; however, it'd be pretty useful, 
especially when working on heavily CTFE-dependent code where 
almost everything is generated at compile-time; while you can 
splatter pragma(msg) around, it can get rather tedious, 
especially when debugging.

A quick look at CompileStatement.semantic indicates that the 
string is parsed, with the resulting statements made the subject 
of a CompoundStatement - at this point, I'm assuming the AST has 
already been processed (lowering, etc), making it difficult to 
recover the original source code from the AST. Perhaps it would 
be possible to splice in the mixin string into the original 
source?
May 08 2016
next sibling parent Iakh <iaktakh gmail.com> writes:
On Sunday, 8 May 2016 at 10:24:12 UTC, Mithun Hunsur wrote:

 Hi all,

 I was discussing this with Stefan Koch and a few other people 
 at DConf - would it be possible to have a compiler switch for 
 outputting the complete source file after the mixins have been 
 expanded, similar to gcc -E?


I think it would be better to have mixin-dump file with all mixins
printed into it and references set to this file. So debug info
could point to some "real" source. But I didn't do any 
investigation
how to implement it.
May 08 2016
prev sibling parent Stefan Koch <uplink.coder googlemail.com> writes:
On Sunday, 8 May 2016 at 10:24:12 UTC, Mithun Hunsur wrote:

 Hi all,

 I was discussing this with Stefan Koch and a few other people 
 at DConf - would it be possible to have a compiler switch for 
 outputting the complete source file after the mixins have been 
 expanded, similar to gcc -E?

 [...]


Yes exactly, splicing it into the original source is precisely 
what I want to do.
You are correct the constructs are already lowered.
May 08 2016