Milter-greylist

Is my per-recipient patch planned to get used?

Having something less than perfect is better than nothing at all, imo...

-Dan

Dan Hollis <goemon@...> wrote:

> Is my per-recipient patch planned to get used? 
> Having something less than perfect is better than nothing at all, imo...

Sorry for the delay, I was attending a conference.

I generally agree that unperfect solutions are better than nothing, but
this solution has an impact on the config file. If we want to later
choose a better direction, we'll have to break backward compatibility on
the config file. That's something I really hate when being in the shoes
of an administrator, so I try to avoid doing it when I'm in the shoes of
a developper.

I'll work on that issue, but my time is scarce to work on
milter-greylist. I have two big pending patches, one with Hajimu UMEMOTO
for IPv6 support, and one with Cyril Guibourg for bind address selection
(Hi Cyril, bad news: the IPv6 patch breaks all our efforts on the bind
problem :o/ It solves some issues we had, however) 

-- 
Emmanuel Dreyfus
Il y a 10 sortes de personnes dans le monde: ceux qui comprennent 
le binaire et ceux qui ne le comprennent pas.
manu@...

manu@... wrote:
> I generally agree that unperfect solutions are better than nothing, but
> this solution has an impact on the config file. If we want to later
> choose a better direction, we'll have to break backward compatibility on
> the config file. That's something I really hate when being in the shoes
> of an administrator, so I try to avoid doing it when I'm in the shoes of
> a developper.

FWIW I have automated script building the config file (there's no other 
way to manage 10,000+ users each with their own greylist settings) so I 
don't think this will be too big of a problem for anyone. If 
greylist.confg format changes later, it's only a few seconds change to 
the script to support different format.

I assume anyone using per-recipient patch would be large ISP like us 
anyway, it doesn't really apply to small systems. (Nobody else had asked 
for it yet, so I'm assuming noone else on this list uses it at an ISP)

You can put warning in the documentation that the syntax is likely to 
change in the future. But this is probably true for most of the 
greylist.conf anyway, not just per-recipient settings.

-Dan

Dan Hollis <goemon@...> wrote:

> You can put warning in the documentation that the syntax is likely to
> change in the future. But this is probably true for most of the 
> greylist.conf anyway, not just per-recipient settings.

As an administrator, I rarely read the doc before updating, I prefer to
upgrade straightfully and cry against the developper that broke backward
compatibility :)

Anyway, I understant what your problem is. In order to reduce the
maintenance trouble for you on each upgrade, we could integrate your
patch with ifdef's that disable it by default. Would that be satisfying
for you? It would be marked as unsupported in future releases in the
doc, and the day we change the way it works, that will only burn the
users that have readen the doc and enabled that feature. 

-- 
Emmanuel Dreyfus
Il y a 10 sortes de personnes dans le monde: ceux qui comprennent 
le binaire et ceux qui ne le comprennent pas.
manu@...

--- In milter-greylist@yahoogroups.com, manu@n... wrote:
> Anyway, I understant what your problem is. In order to reduce the
> maintenance trouble for you on each upgrade, we could integrate your
> patch with ifdef's that disable it by default. Would that be satisfying
> for you? It would be marked as unsupported in future releases in the
> doc, and the day we change the way it works, that will only burn the
> users that have readen the doc and enabled that feature. 

Yes this would work for me. As long as you test new releases that it
doesn't break the patch :-)

-Dan

goemon667 <goemon@...> wrote:

> > Anyway, I understant what your problem is. In order to reduce the
> > maintenance trouble for you on each upgrade, we could integrate your
> > patch with ifdef's that disable it by default. Would that be satisfying
> > for you? It would be marked as unsupported in future releases in the
> > doc, and the day we change the way it works, that will only burn the
> > users that have readen the doc and enabled that feature. 
> 
> Yes this would work for me. As long as you test new releases that it
> doesn't break the patch :-)

Now I just need to find the time to work on milter-greylist. :)

-- 
Emmanuel Dreyfus
Il y a 10 sortes de personnes dans le monde: ceux qui comprennent 
le binaire et ceux qui ne le comprennent pas.
manu@...