Remove deprecated targets
Joseph S. Myers
joseph@codesourcery.com
Sun Jun 8 16:35:00 GMT 2008
On Sun, 8 Jun 2008, Bruce Korb wrote:
> Ten years is a long time to remember details. :) Anyway, way back when I
> wanted to provide a mechanism for determining when various fixes were
> no longer in use. Specifically, record the list of fixes that apply to each file
> and emit that list when "altered_ct" would be incremented inside of the
> test_for_changes function. If we could get the emitted list for every supported
> platform, then we'd know which fixes were being applied to files that were
> being altered. I was told way back when that that was not feasible.
One thing I hope we can achieve through the deprecation process is that
there is a list of supported targets and someone reading the lists with
access to each of them. (You don't actually need every combination of CPU
and OS; most OS headers should be pretty similar for different CPUs.)
I think you should post a list of fixes you think are obsolete and ask for
people finding those fixes are being applied in their builds to come
forward. (Make the build create a log file with the list of fixes, if it
doesn't already.) Then remove the fixes no-one reports are being used,
and put them back (with comments specifying what system they are used on)
if anyone reports a regression arising from their removal. That's similar
to what I'm doing for fixproto removals - invite people to test the
removal but presume it's OK for targets no-one tests.
--
Joseph S. Myers
joseph@codesourcery.com
More information about the Gcc-patches
mailing list