On Mon, Sep 29, 2008 at 4:12 PM, Chris Lattner <[EMAIL PROTECTED]> wrote: > Author: lattner > Date: Mon Sep 29 18:12:31 2008 > New Revision: 56818 > > URL: http://llvm.org/viewvc/llvm-project?rev=56818&view=rev > Log: > Fix the root cause of PR2750 instead of the side effect. > > NumericLiteral parser is not careful about overrun because > it should never be possible.
I'm not very confident in the whole thing, but if you think it works, that's fine. It looks like the tokens in question always include some sort of terminator; the natural terminator when the token points to the source, a space for artificial numeric macros (with this patch), and null for tokens created through token pasting (which looks kind of suspicious...). Is there some sort of guarantee about the token including some sort of separator after the token? -Eli _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
