public inbox for cygwin-apps@cygwin.com
 help / color / mirror / Atom feed
From: "Pierre A. Humblet" <Pierre.Humblet@ieee.org>
To: <cygwin-apps@cygwin.com>
Subject: RE: [ITP] Sendmail 8.14.9
Date: Wed, 27 Aug 2014 15:16:00 -0000	[thread overview]
Message-ID: <01d701cfc209$d9ed1040$8dc730c0$@ieee.org> (raw)
In-Reply-To: <20140826184940.GA11626@calimero.vinschen.de>

> -----Original Message-----
> From:  Corinna Vinschen
> Sent: Tuesday, August 26, 2014 14:50
> 
> On Aug 26 09:51, Pierre A. Humblet wrote:
> > > -----Original Message-----
> > > From:  Yaakov Selkowitz
> > > Sent: Friday, August 22, 2014 17:40
> > >
> > > On 2014-08-22 13:19, D. Boland wrote:
> > > >> On Aug 22 07:43, D. Boland wrote:
> > > >>> I re-packaged Sendmail with cygport. See:
> > > >>>
> > > >>> http://cygwin.boland.nl/x86/release/sendmail/
> > > >>
> > > >> Packaging looks good in theory.
> > > >>
> > > >> Unfortunately we have a problem.
> > > >>
> > > >> On inspection of your binary package I noticed that we have
> > > >> conflicts with exim and ssmtp packages:
> > > [snip]
> >
> > Whatever we do should be transparent to users who have already
> > installed ssmtp or exim and are updating to a newer version.
> 
> I guess that's exactly the problem.  Consider somebody has installed exim or
> ssmtp and then installs sendmail accidentally or out of curiosity.  This
> immediately overwrites the symlinks and even if the user never actually uses
> sendmail, the MTA installation is broken.
> 
> I don't have a clear concept yet if and how we should use alternatives to fix
> this problem.
> 
> Anyway, to have some working example, here's how Fedora does it:
> 
> The binaries, or symlinks to the actual binaries are called:
> 
>   /usr/sbin/sendmail.sendmail
>   /usr/bin/mailq.sendmail
>   /usr/bin/newaliases.sendmail
> 
>   /usr/sbin/sendmail.exim
>   /usr/bin/mailq.exim
>   /usr/bin/newaliases.exim
> 
>   /usr/sbin/sendmail.postfix
>   /usr/bin/mailq.postfix
>   /usr/bin/newaliases.postfix
> 
>   [analog for other MTAs]
> 
> Sendmail, mailq, newalias (etc.) are alternatives symlinks:
> 
>   /usr/sbin/sendmail -> /etc/alternatives/mta
>   /usr/bin/mailq -> /etc/alternatives/mta-mailq
>   /usr/bin/newaliases -> /etc/alternatives/mta-newaliases
> 
> And in /etc/alternatives are the symlinks to the actual binaries, as configured
> by postinstall or the user:
> 
>   /etc/alternatives/mta -> /usr/sbin/sendmail.exim
>   /etc/alternatives/mta-mailq -> /usr/bin/mailq.exim
>   /etc/alternatives/mta-newaliases -> /usr/bin/newaliases.exim
> 
> In fact, there are more than these three symlinks, mainly to rmail and to the
> man pages for the aforementioned binaries.
> 
> Maybe we can shamelessly borrow this scheme?!?
> 

I am fine with the idea, except that I don’t see the need for things such as /usr/sbin/sendmail.exim
The executables can be installed in the current locations (/usr/bin for exim) and symbolic links such as /etc/alternatives/mta can point directly to the executables.
Also it looks like we don't need to use the alternatives package itself, although we reuse the /etc/alternatives directory and naming scheme

Below is what I would do in the MTA postinstall and -config files, using mailq as an example
I am assuming that /usr/bin/mailq will not be created by setup.

In the postinstall:
If /usr/bin/mailq is a symbolic link NOT already pointing to /etc/alternatives/mta-mailq {
           create a symbolic link /etc/alternative/mta-mailq pointing to /usr/bin/mailq's target
           replace the target of /usr/bin/mailq to /etc/alternatives/mta-mailq
}
else if /usr/bin/mailq is a file {
           rename /usr/bin/mailq to /usr/bin/mailq.somemta
          create a symbolic link /etc/alternatives/mta-mailq pointing to /usr/bin/mailq.somemta
          create a symbolic link /usr/bin/mailq to /etc/alternatives/mta-mailq
}
else if /usr/bin/mailq does not exist {
          create a symbolic link /usr/bin/mailq pointing to /etc/alternatives/mta-mailq
          if the MTA has been installed previously (like, old configuration or log files already exists) {
              create a symbolic link  /etc/alternatives/mta-mailq pointing to a suitable target (if any, created by setup) for the MTA being installed
          }
}
The point of the above is not to break existing installations when updating.
Unexpected cases (like /usr/bin/mailq is a directory) are handled in the MTA-config file.

In the MTA-config file executed by the user
- verify that /usr/bin/mailq is a symbolic link to /etc/alternatives/mta-mailq, fix the situation if necessary
- present a dialog to the user with the current value of the target and ask if it should be changed to the current MTA

Pierre


  reply	other threads:[~2014-08-27 15:16 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-08-14 19:00 D. Boland
2014-08-14 20:23 ` Corinna Vinschen
2014-08-22  5:39   ` D. Boland
2014-08-22  8:41     ` Corinna Vinschen
2014-08-22 18:14       ` D. Boland
2014-08-22 21:39         ` Yaakov Selkowitz
2014-08-24  6:33           ` D. Boland
2014-08-25 23:16             ` Yaakov Selkowitz
2014-08-26  8:19               ` Corinna Vinschen
2014-08-26 13:51           ` Pierre A. Humblet
2014-08-26 18:49             ` Corinna Vinschen
2014-08-27 15:16               ` Pierre A. Humblet [this message]
2014-08-28 10:27                 ` Corinna Vinschen
2014-08-28 16:00                   ` Corinna Vinschen
2014-08-28 20:14                     ` Pierre A. Humblet
2015-02-17 11:57 ` D. Boland
2015-02-17 12:15   ` Corinna Vinschen
2015-02-17 15:06   ` Corinna Vinschen
2015-02-17 17:19     ` D. Boland
2015-02-17 22:51       ` Christian Franke
2015-02-17 23:04         ` Corinna Vinschen
2015-04-03 12:08           ` Corinna Vinschen
2015-04-04 11:37             ` Daniel
2015-04-07  8:58               ` Corinna Vinschen
2014-10-26 17:41 D. Boland
2014-10-27 17:07 ` Christian Franke
2014-11-02  6:58   ` D. Boland
2014-11-02 12:44     ` Christian Franke
2014-11-03  9:23       ` Corinna Vinschen
2014-11-03 17:16         ` Christian Franke
2014-11-03 17:24           ` Corinna Vinschen
2014-11-07  8:11             ` D. Boland
2014-11-15 18:34             ` D. Boland
2014-11-17  9:18               ` Corinna Vinschen
2014-11-19  4:45                 ` D. Boland
2014-11-19  5:21                   ` Yaakov Selkowitz
2014-11-20  8:23                     ` D. Boland
2014-11-20  8:53                       ` Yaakov Selkowitz
2014-10-30 17:35 D. Boland
2014-10-31  9:02 ` Corinna Vinschen
2014-11-01  5:38   ` D. Boland
2014-11-01 15:55     ` Corinna Vinschen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='01d701cfc209$d9ed1040$8dc730c0$@ieee.org' \
    --to=pierre.humblet@ieee.org \
    --cc=cygwin-apps@cygwin.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).