Re: [rfc 2/2] Follow DW_AT_linkage_name for methods [Re: The future of dwarf2_physname]

[Keith Seitz <keiths at redhat dot com>]

I don't know if you are waiting on me for a reply on this or not, but I 
thought I would throw out my fractional pennies.

On 05/24/2011 02:21 PM, Jan Kratochvil wrote:
> Also after a chat with Tom Tromey I no longer use DMGL_RET_POSTFIX.  That is
> for ELF symbol `long m<char>()' GDB with proposed patch uses `m<char>()'
> (and not `m<char>()long').  This is compatible with gdb-7.2 (physname)
> template functions.

I am relieved to hear this (mis?)feature is staying put. I'm a practical 
kind of person, and I know I'd really *hate* to have to type "break 'int 
blah<int>(blah)'" every time. Yuck.

As for how to address the (remote?) possibility that we will need a 
return type, I'd like to offer the idea that we might introduce some 
sort of qualifier flag, for example, "break blah<int>/int" or "break/int 
blah<int>" or something along those lines. Of course, this is well 
beyond the scope of 7.3. Just thought I'd toss this out there, though, 
as a starting point should this actually become a problem.

> Anyway DW_AT_linkage_name following patch with no regressions on
> {x86_64,x86_64-m32,i686}-fedora15-linux-gnu is attached.  It would need GCC
> push of the libiberty/ change.

I'll reiterate what I wrote to Tom yesterday, "I'm all for it."

Keith