Mail Archives: cygwin-developers/1999/01/25/06:38:44

From: jeffdbREMOVETHIS AT goodnet DOT com (Mikey)
Subject: Re: Answer to your text == binary question
25 Jan 1999 06:38:44 -0800 :
Message-ID: <>
References: <19990124122533 DOT G9953 AT cygnus DOT com>
Reply-To: jeffdbREMOVETHIS AT goodnet DOT com
To: Christopher Faylor <cgf AT cygnus DOT com>

On Sun, 24 Jan 1999 12:25:33 -0500, you wrote:

>On Sat, Jan 23, 1999 at 07:08:20AM +0000, Mikey wrote:
>>Chris /dev/hd[a-z][1-8] IS a device ;-)
>I completely agree.  If you are referencing the disk as a raw device
>then it should follow CYGWIN=[no]binmode rules.

Under linux, and I would assume most variants of un*x there really isn't
a difference, whether you put a filesystem on it or not, and whether you
access it as a raw device or access a file on a mounted filesystem, you
get back out what you put in.

>>new automounts should default to whatever text/binary mode is set with
>>[no]binmode, if '/' isn't already mounted.
>It'll be really tough not to have '/' mounted if you are already at the
>point where you're referencing automounts.

Maybe I expressed myself poorly.
The first use of any cygwin process will still create a '/' mount,
AFAICS that should be considered an automount, since it is
created without user intervention via the mount command.

>>Obviously if someone sets nobinmode in their global configuration they
>>wouldn't want binary mounts.

By global configuration I meant the options
set in $CYGWIN before starting any cygnus

if [no]binmode is unset use the current default of text automounts and
binary pipe/device access if binmode is set use binary automounts and
binary pipe/device access if nobinmode is set use text automounts and
text pipe/device access

>I'm not sure what global configuration means in this context.
>>otherwise default to the '/' mode
>I don't think this is desireable.  It just adds another layer of obfuscation.

Why?  if I mount -b c:/ / obviously I have my tools setup to expect
binary data, if the header/source files that I access across the network
to setup a cross compiler come from a temp // mount or an automount I
still want them to be accessed as binary, unless I specifically override
that, via a mount/remount, and I think I should be able to expect that.

As far as I'm concerned all cygwin access should be via mounted dirs,
and all of mine is, but if you aren't going to enforce that, and
are going to provide another method of disk access you should
also provide a method to set the default access mode.

defaulting to '/' mode seems to me the most logical.


- Raw text -

  webmaster     delorie software   privacy  
  Copyright 2019   by DJ Delorie     Updated Jul 2019