[proftpd-dfsg] branch master updated (dc5027e -> 1e3bae5)

[Francesco P. Lovergine]

On Sat, Jan 06, 2018 at 10:39:19PM +0100, Hilmar Preuße wrote:
> >> 3. We test for the correctness proftp configuration before we start the
> >> init script. If we really need to do that (instead of simply start and
> >> eventually fail) I suggest to move it to the init script and simplify
> >> the postinst script.
> > 
> > Currently I have not a clear tought about that. I don't remember
> > if it is something introduced by me or not, but anyway the idea is communicating
> > ASAP the problems to admin without interrupting the service, but I'm not sure
> > if the service at that stage is *already* stopped in any case.
> > 
> OK, I found a bug for this: #379503. If a postinst fails (and hence a
> package upgrade) due to a wrong config file this needs to be handeled.
> However I tend to leave that job to the init script instead of postinst.
> 

Also, I notice that preinst stops the service, so indeed the postinst
approach is quite pointless. Avoiding the stops at preinst could be
excessively safe, but it could be due to some (past?) issue or corner
case.

> > I would start with a separate 1.3.6 branch with substantial changes to
> > debian stuff, e.g. use of dh approach and auto stuff. 
> > 
> I guess there is still a long way until we can use purely dh. ;-)
> 
> H.

I'm quite optimistic on those regards :-D


-- 
Francesco P. Lovergine