Commit Graph

407 Commits

Author SHA1 Message Date
Chris Lattner
1fde9f69fd remove the diag::tuple_global_missing_type diagnostic, which printed as
"a tuple type specifying the type of a global must explicitly state the type of each tuple element".

This diagnostic doesn't make any sense, and is apparently only generated by the old
type checker.  It is also the only thing keeping the "isFirstPass" argument to
TC.validateType() alive, so remove it.

The new type checker does have issues with default values, I filed 
rdar://13946171 to track it, but this old approach (which I wrote, *shame*) 
seems way too broken to salvage.



Swift SVN r5244
2013-05-21 04:08:32 +00:00
Chris Lattner
b4e7ab4247 use TypeLowering to produce SILTypes that make sense now that aggregates
have been simplified.  This still isn't ideal, but is a lot better than
what we had, and is waiting for SILFunctionTypeInfo to stop containing
SILTypes.


Swift SVN r5232
2013-05-20 19:36:57 +00:00
Chris Lattner
a38747b6a2 remove parsing logic for SILType calling convention attribute, this
is hoisted up to Type now.



Swift SVN r5230
2013-05-20 19:29:44 +00:00
Chris Lattner
087d4c8430 Implement the parsing logic for SIL types (the SILType specific attributes).
All of it is still dropped on the floor.


Swift SVN r5195
2013-05-17 01:02:02 +00:00
Chris Lattner
611a3a97db Parse sil linkage types, and actually create the SILFunction parsed.
This gets us prototypes parsing and printing, though types aren't
correct at all yet.



Swift SVN r5192
2013-05-16 22:48:07 +00:00
Chris Lattner
bde8f753da wire up minimal sil function parsing (still not validating or creating
a SIL function).  This introduces contextual lexing state for SIL function 
bodies for SIL-specific lexing rules.


Swift SVN r5179
2013-05-16 18:46:46 +00:00
Chris Lattner
192475ad88 carve out a new file for the sil parser logic.
Swift SVN r5177
2013-05-16 17:59:30 +00:00