openerp-community team mailing list archive

[Raphael Valyi <rvalyi@xxxxxxxxx>]

On Fri, Oct 25, 2013 at 3:05 PM, Ronald Portier <ronald@xxxxxxxx> wrote:

> Personally I think that if the official branches are made less strict,
> in the sense that they accept database changes needed to resolve bugs,
> then there should be no problem to proceed along the lines proposed by
> Olivier.
>
> I think most people who contribute features and fixes to the stable
> branches are well aware of the need to avoid database changes wherever
> possible.
>
> But I fear that having an "experimental" OCB branch in the official
> projects, alongside a "conservative" branch, will just make things
> complicated.



Hello Ronald,

this is my vision too because with 2 branches it will be hard to ensure
both have enough critical mass of "Human Continous Integration"  which is
sadly how most of the important bugs are found today. Eventually as OCB
usage among advanced users grows enough, that could be done, but may be not
before.
I think that OpenERP SA's should be responsible instead to cherrypick
themselves from OCB instead to limit the diff. Let's also remind that for
every OCB merge there is a merge proposal to the official branches. Well
this is my vision. Of course, some may have a different one and I will
respect it as I'm almost a mere user of OCB branches.


Also, to answer Olivier, let's keep in mind that most of these millions of
people that would like to use OpenERP as an ERP just fail or dream (these
for which we may not tolerate a schema change). In fact, among the
relatively few success stories of companies using OpenERP as something
close to an ERP, most of them are indeed driven by these few people we know
in the community. In fact it's so hard that people who don't invest in the
community will most certainly fail, at least this is largely what we
observe.
I could remember last year, Antony and Fabien trying to sell us that vision
in Belgium and mentioning a possible success stories in for a wine
re-seller in the USA. Well that went short when we told them that it was us
behind the project too.
You may sell the 'trusted by millions" image to the investors you are
looking for, but not for the experimented people around who are really
working with OpenERP.

As for the fact that there were a lot of shit happening on the 5.x series
official branches. IMHO that was not a problem of stable policy, that was,
sorry to tell it, a problem of very low end offshore programmers touching
the "stable" branches directly and also a framework even more borderline at
that time.
IMHO, this vision of saying things are stable and we should stick to stable
harms OpenERP a lot. So what is stable? v7 when it was released? What a
joke, right?
IMHO that would be a lot more constructive to admit things are far from
being as mature as sold and just facing the work to be done without lying
to ourselves or new community members. Now yes that would certainly clashes
on the business side and this is where I sometimes used the heretic word
"bubble".
An other thing that would be constructive, would be to release at different
scales. That is for instance say OK the server layer should be stable for
12 months. Now, that purchase_requisition module for instance is just a
POC, let's assume it and accept to have several "stable" versions of it in
the year in order to improve it over the year instead of lying to ourselves
and then having only the range of a few days per year to make the
improvement, a range of days where probably none of the people that would
use and improve the module will be available. And then it stays broken for
ages.

Anyway, this was just my shocking point of view.


Regards