[cfe-dev] Getting Stmt's parent, striping a QualType from CVR Qualifiers, eclipse & llvm

Ilya Mirsky ilya.mirsky at gmail.com
Thu Aug 26 06:53:54 PDT 2010


Hi clang developers,

First, I would like to thank Douglas Gregor, who consistently helps and
answers questions. I learn a lot from his replies, so Doug, I really
appreciate it!

I'm working on a program that does a src-2-src transformations, and I have
the following issues:

   1.  For some Stmt, how can I get its parent Stmt. i.e. when Replacing a
   Stmt which forms an IF body, sometimes it requires placing the new code
   inside brackets (replacement of a single Stmt with a block of Stmts), or if
   I'm replacing a Stmt which is part of a compound expression I want to get
   the location of the semicolon, in order to insert some code right afterward.
   How can I achieve this?

   2. Calling ParmVarDecl::getType::getAsString on the src code "const char
   *str" returns "TODO".
   How can I get just the "char*", without the const?
   I've tried to apply QualType::removeConst, QualType::removeCVRQualifiers
   but none seems to do any change.
   What is the right way to get the stripped type, without the CVR
   Qualifiers?

   3. Currently I'm working on VS, but I want to switch to eclipse (minGW),
   unfortunately I'm having problems with llvm in eclipse. I run cmake with
   eclipse CDT configuration, but when I try to compile in eclipse I get
   errors. When I try to link an llvm client to existing llvm libraries (built
   with VS) I also get multiple compilation errors (the include lib is
   specified).
   My assumption is that minGW is more strict than VS. Is this really the
   problem, or am I missing something?
   Can anybody help me work it out?


Thanks in advance,
Ilya


-- 
Best Regards,
Ilya Mirsky
www.cs.bgu.ac.il/~mirskyil <http://www.cs.bgu.ac.il/%7Emirskyil>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20100826/3ef8c718/attachment.html>


More information about the cfe-dev mailing list