This is the mail archive of the
mailing list for the Cygwin project.
Re: setup.exe suggestion [email@example.com: Re: sshd dies]
- To: Robert Collins <robert dot collins at itdomain dot com dot au>
- Subject: Re: setup.exe suggestion [firstname.lastname@example.org: Re: sshd dies]
- From: Christopher Faylor <cgf at redhat dot com>
- Date: Fri, 9 Nov 2001 18:04:15 -0500
- Cc: cygwin-apps at cygwin dot com
- References: <20011109111335.E24506@redhat.com> <01b401c16972$94129ac0$0200a8c0@lifelesswks>
- Reply-To: cygwin-apps at cygwin dot com
On Sat, Nov 10, 2001 at 10:02:16AM +1100, Robert Collins wrote:
>----- Original Message -----
>From: "Christopher Faylor" <email@example.com>
>Sent: Saturday, November 10, 2001 3:13 AM
>Subject: setup.exe suggestion [firstname.lastname@example.org: Re: sshd dies]
>> It probably would be a good idea to add an "is cygwin running?" check
>> to setup.exe. It could be a little tricky to determine, though, since
>> the name of the shared regions that cygwin uses can change.
>Errgh, I really don't like this.
>Checking if package foo is running before upgrading it, yes.
>Checking if cygwin is running at all, no.
>> I'm sort of loath to add a standard named entity in cygwin just to
>> accomodate setup.exe.
>> Possibly, just walking through all of the processes and seeing if
>> loaded cygwin1.dll is a brute force approach.
>Which doesn't cover all cases :p. (Ie cygwin1.dll is only one file to be
>detected for write issues.)
>What we need is to be able to check for any of the .exe's or .dll's in a
>package, before uninstalling, and allow a [cancel][retry] option. You're
>right that the snapshot approach is probably best.
Maybe we can force this by changing how the files are written? If we use
the right sharing, maybe even Windows 9x will complain on overwriting a