🚀 go-pugleaf

In article <NIERMANN.90Mar8143924@dwarfs.csg.uiuc.edu> niermann@dwarfs.csg.uiuc.edu (Tom Niermann) writes:
>
>I am using gdb3.5 and g++ 1.36.4 on a sun3os4.
>
>The funtion (break function) does not work for functions.
>
>However (break function__Fi) does work for a function with one int parameter.
>
>Is gdb going to changed to use the real names of functions instead of their
>g++ mutations? Or is there a way around this?
>
>niermann@csg.uiuc.edu
>


In article <NIERMANN.90Mar8143924@dwarfs.csg.uiuc.edu> niermann@dwarfs.csg.uiuc.edu (Tom Niermann) writes:
>I am using gdb3.5 and g++ 1.36.4 on a sun3os4.
>The funtion (break function) does not work for functions.
>However (break function__Fi) does work for a function with one int parameter.
>Is gdb going to changed to use the real names of functions instead of their
>g++ mutations? Or is there a way around this?
>
>niermann@csg.uiuc.edu

     We at DG are currently making modifications to gdb 3.4 addressing 
this issue on COFF-based systems (specifically, our AViiON 88K boxes).  
Basically, when we read in the symbol table, we run all function 
names through the demangler (supplied with G++ 1.36.*).  
If it's an overloaded function candidate we place it in the 
misc_function_vector and reference that throughout gdb.

     Our changes are in conjunction with code we have developed to handle
class method resolution in COFF (gdb on systems with dbx-style debugging
info, such as your sun, already provides this functionality).  Our patches, 
when available, may require you to delete some unwanted logic.

     When we are reasonably comfortable with our modifications, we'll
send a note to this forum, informing interested parties where they
can get the patches via anonymous ftp.


Andy Sawczyn
Data General Corporation
sawczyn@dg-rtp.dg.com