Le 21/06/2012 16:01, Marcus Denker a écrit :
On Jun 21, 2012, at 3:36 PM, Goubier Thierry wrote:
It seem the final objective is to send something to Esug ?
If I make it in time, yes. Ok, it seems I shouldn't try the RecompileTests then? No, I don't use them right now...
So far, I have only the IRDecompilerTests which are failing.
That is the one that depends on PetitParser (experiment, not needed will be done after the rest works).
State of things:
1) Backend.
BC->IR->BC seem ok.
- IRComparer new compareImage shows 16 differences, mostly pop-combining and unused code after return skipped, seems good. (this compares #abstractSymbolic on the bytecode level)
- IRComparer new compareImageFrameSize 10 methods have smaller framesize , 2 a larger one. This needs to be checked.
2) AST to IR AST-IR-BC - OCRecompiler new inspect; recompileImage shows 10 problems. 2 are undeclareds, one is related to primitive errors (which is not yet supported). Rest are wrong stack depth on jump targets in the backend. - CompilerComparer new compare: Morph (this compares #abstractSymbolic on the bytecode level) ==> 39 differences, need to be analyzed (same for the whole image).
Ok, this is where I may see things. To insert things in the AST, I can subclass the RBProgramNodeVisitor ?
3) Missing: IR->AST, mapping BC to text offset needs to be checked, and then integration into the system. Error handling, things like that.
On the good side,
OCRecompiler new doInstallMethods; recompileImage
seems to work. (did not yet run too many tests, first the rest of the problems need to be fixed).
Marcus
-- Marcus Denker -- http://marcusdenker.de
Thanks! Thierry -- Thierry Goubier CEA list Laboratoire des Fondations des Systèmes Temps Réel Embarqués 91191 Gif sur Yvette Cedex France Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95