This is the mail archive of the
libstdc++@sourceware.cygnus.com
mailing list for the libstdc++ project.
Re: Binary size of the libstdc++.so library
- To: martin at loewis dot home dot cs dot tu-berlin dot de, oliva at lsd dot ic dot unicamp dot br
- Subject: Re: Binary size of the libstdc++.so library
- From: Phil Edwards <pedwards at jaj dot com>
- Date: Mon, 17 Jan 2000 19:40:25 -0500
- Cc: dg at px dot uk dot com, libstdc++ at sourceware dot cygnus dot com
Alexandre Oliva <oliva@lsd.ic.unicamp.br> wrote:
> On Jan 17, 2000, "Martin v. Loewis" <martin@loewis.home.cs.tu-berlin.de> wrote:
>
> > Are you sure? On a Linux system, and I guess on all ELF systems, strip
> > knows what the .dynsym section is, and won't remove it.
>
> Maybe GNU strip is smart enough to detect this. But I've just had
> trouble with an overly eager session of strip on HPsUX, using the
> native strip, that ended up corrupting all shared and static libraries
> of the GNU-software tree, and I ended up having to re-install
> everything. I recall having faced this kind of problem on SunOS4 and
> Solaris2 in the past too, and probably in not so many other
> architectures for the sole reason that my installation script would
> strip only programs.
Just FYI, the native strip(1) under IRIX 6.5 (and earlier?) has the lovable
habit of stripping out exception information. The resulting library can
be linked against and the executable run, but while the throw's are still
there, often the catch's won't be. Wonderful effect, that.
(To be fair, you can use command-line options to avoid that problem. /IF/
you know about them.)
Phil