Brazil mailing list archives


Re: Apresentações

- 23/06/2015 03:12:55
Hello Leonardo,

As I said, it was mostly a commercial decision to screw package management. But also if you believe Odoo developers were really good like 4 years ago you are wrong. At the beginning, a few years before, the good developer was Cedrick Krier and he went create Tryton...

Also, I know wheel is going to be okay.

Pip is meant to be not to bad and could certainly be much better than what we have. If you look in the English community, there is some discussion how buildout may use pip cause today it's not using pip and worse it's the freeze option is incompatible if you installed some modules with pip...
But also, even if pip seems to be okay right now, it's totally incredible to see how bugs like this one have been open and blocking many advanced usages.

You can discuss theses topic in the English community, but personally I've been bringing them on the table for years and I'm a bit tired. Given this is a political decision around the appstore, I'm afraid that may improve significantly only if there is an OCA fork or if the OCA becomes so string that it really starts to dictate the future of Odoo to Odoo SA. That may happen in a matter of a year eventually despite few would officially admit that.


On Mon, Jun 22, 2015 at 3:11 PM, Leonardo Rochael Almeida <> wrote:
Hi Rafaël,

Just a commentary regarding your comment:

On 22 June 2015 at 12:10, Raphaël Valyi <> wrote:

[...] package management was always weak in Python (you would even hear that from the mouth of Python evangelists like Tarek Ziade, so that's not me doing flame war).

Package management in Python could use better documentation, and is a moving target, with a transition to "wheel" still ongoing, but it's no longer "weak". And I dare say it hasn't been "weak" for close to a decade now. And basing any decisions on it being weak is counter-productive.

As a newcomer, it's surprising to me that odoo is still using a technique for package management (using configuration or cmdline switches to stitch arbitrary directories under a single package) that Zope2 was using in 2001.

It is possible to do package management properly with standard python technologies, even if you are sharing a Python namespace (like "openerp"), and both the Plone and the Pyramid communities (to mention two examples), have been using them successfully, including recursive dependency tracking.

These days you should just "pip install" (or add to the "eggs =" section of your buildout) all the packages you need. Or "pip install -e" (or add to the "develop =" setting of your buildout) if you're actively developing a package. And yes, it should be one repository per module (or tightly bound collection of modules).

Anyway, this is not the forum where improvements to Odoo add-on packaging are going to be discussed, so what I wanted to ask is: where do we want to go? And how do we move forward? What work is ongoing for our corner of the community (pt-BR localization), and how do I find it? How can a newcomer collaborate?

Increasing the scope a little bit: do we want to have more hands-on collaboration, like sprints? Do we want to call more attention to ourselves in Brazil, for example at PythonBrasil? FISL? Do we ever want to get into the Portal do Software Público [1], which forces government contracts to first search there for solutions before trying proprietary ones?



[1] (ironically, I can connect to that right now)

On the contrary if we had a proper package management, then installing modules from many repos would be trivial. Today it has to be done somewhat manually. The Buildout system helps, but it's still really weak compared to stuff like say Ruby Bundler, it won't pick dependencies recursively or do some kind of smart matching...

So if we had a proper package management and 1 module = 1 branch, then that __unported__ thing would never exist, all would simply be a branch. v8 branch for a given module would exist or not, simple as that...

But today we don't have that proper packaging, so we somewhat keep grouping modules together and we had to find some artifact to explicit migrated and non migrated modules (other solutions were proposed). Ultimately that kind of topic should be debated in the international OCA list.

But may be, that trend to go toward 1 module = 1 branch is more clear now and may be you also understand better why I don't want to go counter stream by merging the fiscal-document project inside the same repo when we should instead go toward splitting the repos and pinning them via package management.


On Mon, Jun 22, 2015 at 10:56 AM, Danimar Ribeiro <> wrote:
I've sent some considerations to some PR four days ago.
I've put the link with some improvements I made in my branchs, it's not necessary to get the commit, just copy the code if it's easy, I don't mind.

I won't introduce myself, (lazy today), I'll just say I'm a programmer and I help in l10n-brazil :)

And just to express my thoughts, I really don't like those __unported__ folder, can we do it differently next time? it's really hard to compare the diffs.

2015-06-22 9:15 GMT-03:00 Raphaël Valyi <>:
Hello guys,

I'm Raphaël Valyi, the founder of Akretion who created the largest part of the Brazilian localization since 2009 and also co-created the OCA since 2013, along with CampToCamp (Switzerland-France), Therp (Netherlands), Vauxoo (Venezuela-Mexico) and Savoir Faire Linux (Canada).

I think we will soon communicate about the v8 and the unfortunate counter-time we had in the OCA v8 localization release. But as I explained I'm only slowly getting back to work, so that's taking a few days. I also beg for your comprehension, mentioning that for instance we got v7 published like 1 year before Spain that is seen as a dynamic localization by many, so it's not like we always did it wrong over the past 6 years. So really we had a tough problem this year, but we nearly sorted it out now.

The Google groups we created has more than 1000 people registered in Brazil. But unfortunately has you may have seen, may be 99% of the audience has a completely wrong vision of the software and think it's just about downloading and putting it on some server while this is may be only 1% of getting a project running. People think the average adminsys is going to make it work when if you look outside from Brazil, Odoo is only successfully implemented by really top notch engineers with a lot of experience...

My opinion is this wrong expectations have been created by Odoo SA themselves with the kind of marketing they started doing back around 2012 to make VC's dream about the quick returns they promised to borrow the money (like "create a free website with Odoo", that kind of message). In Brazil SMB's are little used to the open source eco-system and has poor knowledge about its complex economics. So I would say this is the reason we got that kind of audience. Recently, as we had to re-focus our business at Akretion, we even kind of stopped animating the list as only 1% of it was creating something useful for the eco-system. I would even say letting it quite for some time was pedagogical, as hard as it may have been.

The Brazilian localization is probably one of the very few most complex in the world. It took millions for a company like SAP to create a localization for Brazil. But few people believe that. Sadly even Odoo SA has no clue at all about that fact...

So to avoid being cross topic, probably what we can do is keep this list for development and OCA oriented stuff while keeping the Google group for a lower target audience.

You guys could already help reviewing the 20+ pending PR's for the localization and also try to spot the features we would be missing from the branch done by Mileo and Danimar while for some force majeure internal reasons we couldn't publish the v8 yet.

But soon that will be over, I believe we will finally have an exemplar codebase in the l10n-brazil project, with may be 60 to 70% test coverage, so really it will finally be quite consistent and easier to review third party contribs.

Also guys if you have some topic you guys want to discuss with the OCA, feel free to drop it here as we have 2 people in the board (Sebastien and Sylvain) and I have also been elected a delegate (I'm also a core reviewer). But the OCA is no magic, it's a "do-ocracy", so what is important is to be active in the reviews. You cannot expect people to review your code if you don't do reviews yourself. By that I don't mean only in the Brazilian project, otherwise the OCA which is an international project will never really value your skills. I've been calling about that for years and sadly that never happened here. I hope the recent moves from Odoo SA will open the eyes of the most skeptical people why we created the OCA.

Thanks and welcome to all, I believe exciting times are coming.

2015-06-22 0:21 GMT-03:00 Luis Felipe Miléo <>:
@Leo Please feel free to tell us your story with the ERP5!


Luis Felipe Miléo

+55 35 8876-3663
+55 35 3622-2548
skype: luisfelipemileo
Parceiro oficial no Brasil:

----- Mensagem original -----
De: Leonardo Rochael Almeida <>
Para: Brazil <>
Enviadas: Sun, 21 Jun 2015 22:35:49 -0400 (EDT)
Assunto: Re: Apresentações

Curiosamente, estava a ponto de fazer a mesma coisa que o Luiz,

E acho que não conto como figurinha repetida, afinal, estou começando a me envolver com Odoo agora :-)

Estive na Odoo experience e achei muito legal, em particular o papel que a OCA representa.

O Luis eu conheci faz pouco tempo, numa apresentação online que ele fez do Odoo pra um potencial cliente na qual eu estava presente.

Fora ele, não conheço ninguém mais das pessoas que estão trabalhando com o Odoo aqui no Brasil.

Também vi que esta lista está bem enxuta, umas 16 pessoas só [1].

Que tipo de atividades nós queremos coordenar nesta lista?

E já que eu não sou figurinha repetida, o que vcs acham da ideia de nos apresentarmos por aqui?



On 21 June 2015 at 22:56, Rodrigo Lamar <> wrote:

Até então, só figurinha repetida =P


Post to:

Post to:


Post to:

Danimar Ribeiro

Post to:


Post to:

Post to:

Raphaël Valyi
Founder and consultant
+55 21 3942-2434