F20 System Wide Change: No Default Sendmail

[Adam Williamson]

On Sat, 2013-07-20 at 16:58 +0200, Nicolas Mailhot wrote:
> Le Sam 20 juillet 2013 00:46, Adam Williamson a écrit :
> > On Fri, 2013-07-19 at 14:23 -0500, Billy Crook wrote:
> >
> >> > It's not as if the default MTA configuration, as I understand it (no
> >> smart
> >> > host, attempt to send all mail directly with the FQDN) is generally
> >> useful
> >> > OOTB.
> >>
> >> It is to me, and I suspect I am not alone.
> >
> > I call bullshit. How is it useful to you? All it can possibly do is
> > deliver mail to root and users in /var/spool . We don't set up any mail
> > reader to read this mail out of the box. No app is very likely to know
> > your user name and send mail to you. The only practical thing you can do
> > with the OOTB sendmail configuration is manually read or configure an
> > agent to read the mail in root's /var/spool directory - and that is not
> > the recommended way to read root's mail _anyway_, the recommended way is
> > to alias root to a user account.
> 
> And the recommended way is trivial (even though anaconda should really set
> up the alias at install time)
> 
> What won't be trivial *at* *all* is if you encourage app authors to invent
> their own one-of-a-kind not-really-tested communication stack. God knows
> they are already leaning towards wheel reinventing without any such system
> blessing.
> 
> Set up your own replacement. Make it good, secure and easy to operate.
> Have it widely adopted. And then you can talk about removing the default
> MTA. If it's good enough you have nothing to fear about MTA competition.
> If you do need MTA removal to get your stuff adopted, well, that only
> tells me your stuff is crap.

You seem to be chasing an argument I've never seen anyone produce
before, without a single shred of concrete evidence to back it up: that
the presence or absence of a non-functioning MTA by default in a single
Linux distribution will significantly affect the likelihood of an app
choosing to use local mail delivery as a standard notification
mechanism. Could you please, before pushing this line of reasoning any
further, produce some shred of evidence to substantiate it?
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net