On Mon, Jul 27, 2026 at 3:24 PM Alexander Monakov <[email protected]> wrote: > > > On Sun, 26 Jul 2026, Richard Biener wrote: > > > > --- a/gcc/gcc.cc > > > +++ b/gcc/gcc.cc > > > @@ -3563,9 +3563,10 @@ execute (void) > > > thinking there's a compiler bug. Much more likely is > > > the user or OOM killer nuked it. */ > > > fatal_error (input_location, > > > - "%s signal terminated program %s", > > > - strsignal (WTERMSIG (status)), > > > - commands[i].prog); > > > + "%s terminated by signal %d (%s)", > > > + commands[i].prog, > > > + WTERMSIG (status), > > > + strsignal (WTERMSIG (status))); > > > > I think strsignal is usually translated in a way > > That matches „terminated by signal %d: %s.“, that is separated by a colon? > > This would match how psignal and psiginfo format their output. > > Regarding translations, on Linux strsignal produces a title-case string > (e.g. "Segmentation fault" with capital S), and translations follow that, > so it works equally poor in both contexts, either in parens or after a colon. > > I guess we can follow standard POSIX functions and use a colon here > (without the period in the end as common for error reporting). Personally, > I'd say parenthesis look a little bit nicer in this context.
Note I have no strong preference here, so I'll ack both variants here, letting Arsen be the tie-breaker. Richard. > > Alexander
