On Tue, Jul 28, 2020 at 9:28 AM Simon Marchi <sim...@simark.ca> wrote: > > On 2020-07-28 12:07 p.m., H.J. Lu via Gdb-patches wrote: > > What doesn't work with my pkg.m4 change? > > (1) It deviates from upstream. I don't think we should do this unless > absolutely needed. That's not the case here, the change is just there > because you don't want to set up pkg-config properly for cross-compiling.
Since when binutils can't fix issues in other packages? > (2) I don't think it's necessarily bad to try to do a link to confirm the lib > is indeed there, but as I said earlier I don't think that doing as if the > package was not there is the right response. This can happen if you have > a mis-configured pkg-config (like you have) or a broken installation (for > example, the .pc is there but the lib is not). In either case, there's > something wrong with the build environment and I think it's more useful > to abort and tell the user rather than silently failing. > > Point 2 should be discussed upstream anyway, there's no point making a > decision > local to binutils-gdb. > Unlike gdb, binutils should have as few external depecies as possible. debuginfod brings in some so many external depecies. -- H.J.