zturner added a comment. In http://reviews.llvm.org/D17363#366522, @clayborg wrote:
> You are right. I am slowly seeing that PDB was designed well to be consumed > as a user would want to, not compressed to the point of obfuscation and very > difficult to extract data from like DWARF. > > So then we really should leave things the way they are and let > "SymbolFilePDB::ResolveSymbolContext(const lldb_private::FileSpec &, > uint32_t, bool, uint32_t, lldb_private::SymbolContextList &)" do the search > because it can do so more efficiently than the manual search we do with DWARF. Yea, in fact I suspect that when we start having large modules with thousands of lines of pre-processed source things will perform really poorly with PDB under the current design. Because it already exposes an API to return stuff in a manner convenient for being consumed, but then we have to convert it to LLDB's format for being consumed. So the extra overhead done by the underlying API will have been wasted. I still need to fix the issue regarding termination entries before I can submit, but that sounds like the last issue. I'll try to get that worked out and then submit. Thanks! http://reviews.llvm.org/D17363 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
