[PROPOSAL] Re-bundle (stop producing tarballs for) ldb?

Simo simo at samba.org
Sat Apr 13 21:24:24 UTC 2019


On Sat, 2019-04-13 at 22:02 +0100, Rowland Penny via samba-technical
wrote:
> On Sat, 13 Apr 2019 16:32:12 -0400
> Simo <simo at samba.org> wrote:
> 
> > On Thu, 2019-04-11 at 09:03 +0100, Rowland Penny via samba-technical
> > wrote:
> > > On Wed, 10 Apr 2019 20:01:29 -0700
> > > Jeremy Allison via samba-technical <samba-technical at lists.samba.org>
> > > wrote:
> > >    
> > > > Red Hat is the most popular Linux distro with a large
> > > > user base and we need to take their engineering
> > > > needs into account.  
> > > 
> > > I don't think taking a particular distro into account is a good
> > > idea. I think we should do what is best for Samba and the wider
> > > community.  
> > 
> > I didn't read Jeremy's phrasing as an exclusive. He just pointed out
> > that there are needs from down-streams as well. I would object myself
> > to catering for a specific party exclusively.
> 
> I understood what he meant, I was just saying that we shouldn't mention
> any distro by name.

I do not see any need to self censor on the fact we have important down
streams, I have no qualms mentioning Ubuntu, Debian, Suse or any other
distribution in an argument.

> > > One big problem with talking about red hat and their engineering
> > > needs is, if you do that, you will have to remove most, if not all,
> > > of the AD DC code, they do not need it, they have publicly said
> > > so.  
> > 
> > Rowland, I do not think this kind of disingenuous commenting is useful
> > in any possible way. The fact a downstream does not use a specific
> > feature does not mean it want it removed, nor it diminishes the
> > feature in any way. There are other down-streams of samba that use
> > just the file-system components, some just winbindd, some just some
> > libraries. Your statement is frankly incomprehensible to me.
> 
> No, it wasn't a 'disingenuous comment', it is a fact the that RHEL have
> said that they will not be producing Samba rpm's that can be used to
> provision an AD DC. So, if we take their needs it to account, we should
> tear out all the AD DC code, because they do not need it. This is, of
> course, ludricous, so we will not be doing it, now do you see why we
> shouldn't mention or think about any specific distro ? 

No, and there is no *need* from RHEL to remove anything, your statement
is a non-sequitur. The fact a downstream does not use or even need a
feature doesn't mean we need to remove anything. On the other hand, as 
Jeremy stated, if a downstream *needs* a feature we do factor that need
in our plans to keep and maintain specific features, we always had,
heck we even came up with new pluggable interfaces to help down-streams 
needs, multiple times, and this generally improved Samba, not always,
but often.

> > > The problem with creating public API's (as I see it), you cannot
> > > just stop them from being public.  
> > 
> > It's a trade-off. And anything can be done, but any action imply a
> > reaction.
> 
> Yes, anything can be done, but should it be done ?

If it was done it was because we thought it was a good idea and there
was enough pressure and momentum for us to deliver it.

> Do we have to trade off anything, can we not find another way of doing
> things ? (I think you will find the answer to that is 'Yes')

We can always find other, better ways, especially in retrospect with
the benefit of hindsight ...
My answer is that it is always a trade off, and the perception as well
benefits may change over time.

Simo.





More information about the samba-technical mailing list