It may be possible to do this more simply by looking at just RecursiveASTVisitor and properly understanding the couple patterns it uses to traverse the whole AST.
--Sean Silva On Wed, Jul 18, 2012 at 11:38 AM, Joshua Cranmer <[email protected]> wrote: > On 7/17/2012 11:51 PM, Sean Silva wrote: >>> >>> On the contrary, I actually like having line/column numbers so I can >> >> match things up with where they are in the source tree. >> >> This is usually unnecessary in my experience (at least while learning) >> because usually the stuff you feed it to be dumped will only contain >> one use of a give identifier or number, which uniquely tracks the node >> back to the source. >> >>> My attempt to do a similar thing for clang ended up failing extremely >> >> miserably. >> >> any diagnosis as to why it failed? > > <https://github.com/jcranmer/viewsource/tree/master/native-tools> has my > attempt to make this work. The clang tree didn't follow rigorous enough > design standards to make scraping names possible without more than a few > hacks, and the fact that half the methods start by declaring assertions > means I need to have very long guard lists or actually scrape assertion code > from building clang. > > > -- > Joshua Cranmer > News submodule owner > DXR coauthor > _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
