On Fri, Feb 02, 2024 at 11:43:21PM +0000, Jonathan Yong wrote:
> On 2/1/24 15:33, Jonathan Yong wrote:
> > On 2/1/24 15:25, Jakub Jelinek wrote:
> > > On Thu, Feb 01, 2024 at 03:55:51PM +0100, Jakub Jelinek wrote:
> > > > No, besides the formatting being incorrect both in ChangeLog and in the
> > > > patch, this pessimizes ILP32 hosts unnecessarily.
> > >
> > > So like this instead?
> > >
> > > 2024-02-01 Jakub Jelinek <ja...@redhat.com>
> > >
> > > * hwint.h (GCC_PRISZ, fmt_size_t, HOST_SIZE_T_PRINT_DEC,
> > > HOST_SIZE_T_PRINT_UNSIGNED, HOST_SIZE_T_PRINT_HEX,
> > > HOST_SIZE_T_PRINT_HEX_PURE): Define.
> > > * ira-conflicts.cc (build_conflict_bit_table): Use it. Formatting
> > > fixes.
> > >
> >
> > Looks good for ILP32/LLP64.
>
> Are you planning to push yet?
It needs to be reviewed first.
Passed successfully bootstrap/regtest on x86_64-linux and i686-linux.
Note, I think incrementally we should search for similar issues,
grep -C3 '%l[duxXi]' *.cc */*.cc | grep 'long)' | grep -v long.long | wc -l
101
is an upper bound on what should be looked at, I see at least a few dozens
from those (e.g. when the argument uses sizeof, or is say hash table
size () or elements () etc.).
Also, wonder if the pretty-printer.cc should get 'z' and 't' modifiers
added, for those we clearly can use 'z' or 't' when we implement it
ourselves. But at least that part is stage1 material I'd say.
In translated messages we certainly can't use macros like
HOST_SIZE_T_PRINT_*, because it is then untranslatable.
Jakub
