On Fri, Mar 20, 2015 at 2:19 PM, Paul Robinson < [email protected]> wrote:
> objcopy's default behavior on seeing an unrecognized switch is to dump its > entire help text. > That's not what I'd call user-friendly. > File a bug on objcopy? The best alternative I can imagine would be clang running objcopy to check if it supports the flag and then producing some other error message which seems like a workaround for an objcopy bug & a lot of work for a relatively infrequent case (& no doubt there'll be a stack overflow post with "why do I get this objcopy result when I enable split-dwarf" with the appropriate answer of "upgrade objcopy to version <blah>"). No-op by default seems misleading. Error when it's not actually problem seems misleading too. This produces errors when it's an error and makes it work when it works, with the drawback of having a somewhat confusing failure mode. - David > > > http://reviews.llvm.org/D8484 > > EMAIL PREFERENCES > http://reviews.llvm.org/settings/panel/emailpreferences/ > > >
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
