On Tue, 23 Aug 2011, SF Markus Elfring wrote: > > OK, if you like you can call it that, because there are pointers from > > local variable references back to thir declaration points. > > Thanks for your acknowledgement. > > Would you like to clarify any documentation?
There is no documentation of the AST. > > But the distinction doesn't seem very important. > > I disagree here. I prefer to avoid confusion with such fundamental terms from > computer science. An important point of an abstract semantic graph seems to be the notion of abstraction. But there is no such notion in coccinelle. All of the tokens in the source code are still present. > Examples: > - Clang developers seems to talk even about abstract semantic trees. > http://clang.llvm.org/docs/InternalsManual.html#libast I believe that Clang does more reorganization of the code in its internal representation. > - How do you think about the information from a document like "Keystone: A > Parser Front-End for the C++ Language"? > http://www.cs.clemson.edu/~malloy/projects/keystone/doc.html#asg > > > > In particular, the code is not abstracted in any way. > > Do you plan any improvements in this direction? In the context of Coccinelle, this would not be considered an improvement. Coccinelle needs to reconstruct the source code as is, so it can't abstract. Some of the effect of abstraction is provided by the isomorphisms, which in some sense introduce abstraction at the transformation specification level. julia _______________________________________________ Cocci mailing list [email protected] http://lists.diku.dk/mailman/listinfo/cocci (Web access from inside DIKUs LAN only)
