Also, writefln or writef("\n") doesn't seems to suffer from this.
Setting a writefln("") just before the pause call fixes all.
Maybe it has something to do with the newline. Neglecting to flush until a
newline?


Tom

On Mon, 14 Nov 2005 03:09:30 +0000 (UTC), Tomás Rossi <Tomás_member@pathlink.com> wrote:
> Also, writefln or writef("\n") doesn't seems to suffer from this.
> Setting a writefln("") just before the pause call fixes all.
> Maybe it has something to do with the newline. Neglecting to flush until a newline?

I suspect the buffer flushes on a newline, so the writef goes into the buffer but isn't flushed, then, I suspect the pause text doesn't go into the buffer but straight to the console. So it gets there before the buffered text.

Regan

In article <opsz7n250r23k2f5@nrage.netwin.co.nz>, Regan Heath says...
>
>On Mon, 14 Nov 2005 03:09:30 +0000 (UTC), Tomás Rossi <Tomás_member@pathlink.com> wrote:
>> Also, writefln or writef("\n") doesn't seems to suffer from this.
>> Setting a writefln("") just before the pause call fixes all.
>> Maybe it has something to do with the newline. Neglecting to flush until
>> a newline?
>
>I suspect the buffer flushes on a newline, so the writef goes into the buffer but isn't flushed, then, I suspect the pause text doesn't go into the buffer but straight to the console. So it gets there before the buffered text.
>
>Regan

The 'terminal' IO library routines underlying phobos will flush on a newline or close by default for both Windows and Linux.

So it isn't strictly a bug unless it's defined otherwise. This default behaviour can be changed by using std.c.stdio.setvbuf(...);