This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: struct gmon_cg_arc_record
- To: jes at linuxcare dot com
- Subject: Re: struct gmon_cg_arc_record
- From: Ian Lance Taylor <ian at zembu dot com>
- Date: 15 Nov 2000 08:49:22 -0800
- CC: drepper at cygnus dot com, libc-hacker at sourceware dot cygnus dot com, binutils at sourceware dot cygnus dot com
- References: <200011141926.UAA20529@lxplus015.cern.ch> <m3ofzis52d.fsf@otr.mynet.cygnus.com> <d3r94eux9b.fsf@lxplus015.cern.ch> <m3aeb2s33g.fsf@otr.mynet.cygnus.com> <d3n1f2uvxu.fsf@lxplus015.cern.ch> <m34s1as27q.fsf@otr.mynet.cygnus.com> <d3itpquuhv.fsf@lxplus015.cern.ch> <m3n1f2qm6p.fsf@otr.mynet.cygnus.com> <d3em0eur8p.fsf@lxplus015.cern.ch> <20001115042349.15398.qmail@daffy.airs.com> <d3n1f16wy7.fsf@lxplus015.cern.ch>
From: Jes Sorensen <jes@linuxcare.com>
Date: 15 Nov 2000 17:07:12 +0100
>>>>> "Ian" == Ian Lance Taylor <ian@zembu.com> writes:
Ian> Yuck. If we're going to touch this, can't we avoid
Ian> sizeof(char*)? It prevents cross-configurations, and there is no
Ian> particular reason to believe that it is correct on any given
Ian> system. At the very least make it something which can be
Ian> configured for a particular system, rather than trusting that the
Ian> compiler and the C library agree on this size.
I wasn't aware of these cross compiler issues hence the reason why I
used the char * approach. How do you recommend we fix this then,
sizeof(unsigned long), new data type, use the glibc provided headers
if available?
I recommend making the size of the fields a value substituted by
configure. That will permit us to set the size based on the
configuration target. Use the current defaults.
If glibc has header files which define the appropriate size, or
something, then we can use them if available when configured for a
native system.
Ian