Mail Archives: cygwin-developers/1999/01/22/07:51:43
At 08:26 AM 1/22/99 -0500, Christopher Faylor wrote:
>On Thu, Jan 21, 1999 at 11:54:33PM -0800, Geoffrey Noer wrote:
>>> I thought that since the automounts exist in the mount table, that
>>> the binmode status of the automount would be the logical way to
>>> control this.
>>
>>I meant using binmode status of the CYGWIN env variable to determine
>>what binmode to associate with the automounts. The status would be
>>stored in the same flags area as text or binmode for normal mounts.
>>The problem is that since the user isn't choosing whether or not to
>>add an automount, it's not possible for them to specify binmode in the
>>standard way (an argument to the mount command).
>>
>>(Maybe I'm not understanding you?)
>
>I thought that we agreed that the default would be text mode mounts. If
>the user wants to change that then they override it by specifying -b to
>the mount command along with whatever option is required to specify the
>automount directory. Then files opened using /mycygdrive will have
>whatever mode the user wants.
>
>That leaves CYGWIN=[no]binmode as the method for specifying the CRLF
>translation for devices and, possibly, std file handles inherited from
>the command shell. That was my intent when I implemented this feature.
>Although, now that I think of it, you can override devices with the
>mount command, too.
>
>cgf
>
I guess I'm not 100% sure what the definition of "automount" means. Assuming
we're talking about the set of "pre-defined" (like "/") directories that are
mounted, so long as they can be unmounted, remounted in the user selected mode, and this is persistent, I don't see a problem with choosing a "hard
coded" default (like "text" I guess) and leaving it as such. If "automounts"
include the mounts created "on the fly" (i.e. /cygdrive), I think there needs
to be some way to control the mount mode. I'm not sure I think pulling the
mode from "/" would be adequate, since this implies that the mode of "/" is
the preference, which may not always be true. I personally think that using
the "binmode" setting of the CYGWIN environment variable makes sense in this
regard (simply cause it already exists and I'd rather not see yet another
way to set this option in these cases). However, it seems Chris had a
different idea of how the "binmode" setting should fit into the whole "file
mode hierarchy" determination. I guess my points are:
1. If modes can be easily reset on automounts BEFORE the user actually
uses, any pre-defined default mode is adequate.
2. If modes cannot be set on some/any automounts before the user uses it,
we need a way to remove this limitation.
3. The mechanism to remove the limitation should be simple to use and
preferably not introduce new syntax if possible while not muddying
existing conventions.
Yeah, that seems like a reasonable "requirements list" to me. Feel free
to ignore them if they seem completely off-the-track!;-)
Larry Hall lhall AT rfk DOT com
RFK Partners, Inc. (781) 239-1053
8 Grove Street (781) 239-1655
Wellesley, MA, 02482-7797 http://www.rfk.com
- Raw text -