[PATCH 0/4] bitpos expansion summary reloaded
Mon Oct 22 20:45:00 GMT 2012
>>>>> "Siddhesh" == Siddhesh Poyarekar <firstname.lastname@example.org> writes:
Siddhesh> Here is a fix on top of the bitpos fixes based on the warnings
Siddhesh> generated from gcc -Wconversion. I have also attached the
Siddhesh> report for review; I have not rebased since the last
Siddhesh> submission to ensure that the line numbers don't go awry. Most
Siddhesh> of the extra warnings were either unrelated or were the length
Siddhesh> parameter to (store|extract)_(un)?signed_integer functions
Siddhesh> that are safe.
Siddhesh> I have also verified that this does not cause any regressions in the
Siddhesh> testsuite and that the gcc warnings generated after this were safe.
IIUC, this patch fixes some subset of -Wconversion warnings but leaves
the rest untouched.
Would it be very hard or ugly if we just tried to fix them all, and then
enabled -Wconversion in configure? Aside from maybe some code ugliness,
I wonder what the downsides would be.
The reason I ask is that I'm concerned about our ability to maintain
this change properly, and I wonder if this would be a cheap way to
handle the more mechanical aspects.
More information about the Gdb-patches