(N.B. The correct replay address to this account is:
ethersoft(a)ne.mediaone.net. If this message apparently comes from
acs(a)ne.mediaone.net, please tell me at acs(a)xemacs.org - I'm currently
struggling with a recalcitrant MTA.)
kyle_jones(a)wonderworks.com said:
(sigh) All right. I see two alternatives. We review our code base
and try to fix the aliasing problems. Or we write a autoconf test to
trap these compilers and pass the proper option to them to turn off
strict aliasing.
I think we should fix our code. We should take the effort to make our code
standards-compliant. Besides the optimization can apparently be a performance
win; therefore I'm not in favor of -fno-strict-aliasing.
- vin
PS. Kyle, will you forward the entirety of this message to acs(a)xemacs.org,
please? I'm still working on the above-mentioned MTA problem, and I'd like to
see what the headers look like when they get to you. Thanks.