This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Code Bloat g++


 In message <20000219111752.C516@wolery.cumb.org>you write:
  > On Thu, Feb 17, 2000 at 01:59:48PM -0700, Jeffrey A Law wrote:
  > >   In message <20000217115304.B13219@wolery.cumb.org>you write:
  > >   > On Thu, Feb 17, 2000 at 08:39:58PM +0100, Martin v. Loewis wrote:
  > >   > > That assumes that large debug information is redundant to a ridicul
  > ous
  > >   > > degree. I don't know whether this is the case, but until I'm proven
  > >   > > wrong, I'd always assume that every single bit in the debug
  > >   > > information serves a purpose.
  > >   > > 
  > >   > > With that assumption, it is ok for me if debug information is large
  >  -
  > >   > > because it could not be much smaller.
  > >   > 
  > >   > Stabs is in fact quite redundant, because everything is stored as
  > >   > ASCII strings.  I've seen the size cut in half by using -gdwarf-2
  > >   > instead.
  > > Note that for stabs we have the ability to remove duplicate information
  > > from the debugging sections.  We do not have that capability for dwarf2
  > > right now.
  > 
  > Is that something the assembler should be doing, or the linker?  Not
  > that I particularly want to go dig around in binutils, but...
It's mostly a linker issue, though the compiler & assembler usually have to
be modified to make this kind of thing work.

Jason has a couple proposals which make this kind of thing easier to do in
dwarf2, but I don't know their details.

jeff


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]