On Wednesday, 30 November 2016 at 13:49:01 UTC, Stefan Koch wrote:
On Tuesday, 29 November 2016 at 23:09:08 UTC, Stefan Koch wrote:
On Tuesday, 29 November 2016 at 23:00:08 UTC, Stefan Koch
wrote:
I fixed a bug in continue break handling.
For the record it was an off by one error.
The value for unresolvedGotos would add one referencing jump.
But the count of referencing would be initialized to zero
instead of one.
I just added support for assert in newCTFE.
Also I noticed a performance regression in the bytecode
generator.
(compared to an earlier version that handled much less)
Nontheless I get a solid 4x performance improvement over the
old evaluator when compiled with ldc.
Furthermore struct support regressed.
It has to be rewritten anyway in order to support method-calls
properly.
Every now and then it would be interesting to have a list of
things that work and a list of things still to do.
Keep up the good work!
