From: Jeff King <peff@peff.net>
Date: Sat, 26 Aug 2017 18:53:37 GMT
Subject: Re: cat-file timing window on Cygwin
Message-ID: <20170826185337.hrcswjuunon54kgj@sigill.intra.peff.net>
In-Reply-To: <253bf111-4e8a-54b7-6e39-2908270aa357@ramsayjones.plus.com>

On Sat, Aug 26, 2017 at 01:57:18AM +0100, Ramsay Jones wrote:

> > diff --git a/run-command.c b/run-command.c
> > index 98621faca8..064ebd1995 100644
> > --- a/run-command.c
> > +++ b/run-command.c
> > @@ -641,7 +641,6 @@ int start_command(struct child_process *cmd)
> >  	}
> >  
> >  	trace_argv_printf(cmd->argv, "trace: run_command:");
> > -	fflush(NULL);
> >  
> >  #ifndef GIT_WINDOWS_NATIVE
> >  {
> 
> I suspect not, but I can give it a try ...
> 
> ... oh, wow, that works! Ahem. (Hmm, so it's flushing stdin?!)

Interesting. I find it a little hard to believe there's so obvious a bug
as "fflush(NULL) flushes stdin", but well...that's what it seems like.

If that's truly what it is, this is the minimal reproduction I came up
with:

-- >8 --
#include <stdio.h>

int main(void)
{
	char buf[256];
	while (fgets(buf, sizeof(buf), stdin)) {
		fprintf(stdout, "got: %s", buf);
		fflush(NULL);
	}
	return 0;
}
-- 8< --

If this really is the bug, then doing something like "seq 10 | ./a.out"
would drop some of the input lines.

-Peff