Date:	Sat, 14 Oct 2000 20:37:50 +0200
From:	"Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
Sender:	halo1 AT zahav DOT net DOT il
To:	Martin Str|mberg <ams AT ludd DOT luth DOT se>
Message-Id:	<7263-Sat14Oct2000203750+0300-eliz@is.elta.co.il>
X-Mailer:	Emacs 20.6 (via feedmail 8.3.emacs20_6 I) and Blat ver 1.8.5h
CC:	djgpp-workers AT delorie DOT com
In-reply-to:	<200010132314.BAA29969@father.ludd.luth.se> (message from Martin
	Str|mberg on Sat, 14 Oct 2000 01:14:15 +0200 (MET DST))
Subject:	Re: Bug 000323 (LONG)
References:	<200010132314 DOT BAA29969 AT father DOT ludd DOT luth DOT se>
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

> From: Martin Str|mberg <ams AT ludd DOT luth DOT se>
> Date: Sat, 14 Oct 2000 01:14:15 +0200 (MET DST)
> 
> Ok, the good news is this: I've implemented the stuff for correcting
> this bug.
> 
> BUT I have not the time to continue working on it for the moment so
> I'd like to make my work available to you if any of you have time for
> testing and hacking it.

Thanks!

> The bad news is: COMPLETELY untested! And I really mean this. I think
> it compiled but I might have changed something after that. I don't
> think I've compiled a program using a libc containing these changes.

I have some comments on the implementation:

  - You seek to end of the file when it is open in the append mode.  I
    wonder whether this is what other implementations do.  My
    references don't seem to imply that opening the file already seeks
    to the end; they only say that before each write the file position
    is set to the end.

  - Instead of adding an fseek to each buffered stdio function (such
    as putc, fwrite, etc.), isn't it better to do that inside _flsbuf,
    which is the only function used by buffered functions to actually
    write to the file?

  - Please put a FIXME comment inside fcntl (where the F_GETFL flag
    should be handled) to say that O_APPEND flag needs to be looked up
    in __file_handle_modes[] and returned to the caller.

Thanks again for working on this.

- Raw text -