devtalk

[rootbeer23]

Hi.

since FAF is going for an open development now there are a few things
i like to say.
i hear that (as expected) some are already planning and coding.
Like before this happens in small circles. open source is not just
about the end product. It is also about education. And there is
some free education to be had here if you can watch over other peoples
shoulders and see how the sausage is made. No question too stupid
(if someone happens to take the time to answer it).

Where there is talk, there is automatically documentation.

And finally: not everybody may be willing to learn 1000 details
to be able to program a computer, but almost everybody is able to
understand the basic problems and offer ideas or ask good questions
after they learn the vocabulary.
Also there are quite a lot here who already do.

[nine2]

https://github.com/faforever

[Ze_PilOt]

I don't know how the new owner is planning to organize it, but I still think that if it's a "community effort" with no lead, the coders will run around like headless chicken.

While someone will implement something, someone else will bloat the server with something, and a third one will fix something that make the first guy job obsolete.

A project that big need a lead, and contributors will have to accept it. Maybe it can work for a while without a lead, but it can only be disastrous in the long run.

When Thygrr was active in FAF dev, he was doing the lobby work while I was doing the server work. We work together on the redesign of the protocol. If a third guy would have been there, it would have to answer either to Thygrr or me.
Actually that happens with Thygrr who had the help of someone (that's the guy who did the autocomplete for the chat).

They are big priority now, and like it or not, these are not fancy stuff to implement.

FA is not compatible with Ipv6. Ipv4 is done. Most of the connection problems, and why the proxy server is used so much is because of that.

The FIRST thing that need to be done is coding a local proxy server next to the client, in c++, that will tunnel ipv6 connections to FA.
That's what I was working on (and that's why it seems FAF was not evolving).

All other problems or function are trivial compared to this issue.

[Frozen_byte]

Well for me i.e. I will not able to fix major issues in C++ nor in Python (since I'm more the Web-Dev)

Anyway my goal in the OS-Project will be to establish basic (unofficial) linux-support. This will not happen from now to tomorrow, but I will investigate.

Also: If some dev-talk happens in future I don't have to ask "How is XY working?! - Sorry I cannot say anything to this".
Now I only quickly take a look myself!

Indeed ZeP: Without a leading-instance nothing will happen.
Anyway it's much easier now to find and fix someone's mistake (like removing bloats)
Many Project work the same way (like Malpile & Tor are in active colab-dev right now)


The biggest issue I see:
We have no docs. Open Source projects are heavily relaying on good documentations.

[Ze_PilOt]

The biggest issue I see:
We have no docs. Open Source projects are heavily relaying on good documentations.

That's why I've said so many time that having the server code is premature. I never refused for help (that's why I don't get it's will be easier now), I've always said than the current state of the server code doesn't allow someone to work efficiently on it.

Also server access should stay in the end of one and one person only.
Nothing good can come if any contributor can push any code in production when he feel like it.

[Frozen_byte]

The biggest issue I see:
We have no docs. Open Source projects are heavily relaying on good documentations.

That's why I've said so many time that having the server code is premature. I never refused for help (that's why I don't get it's will be easier now), I've always said than the current state of the server code doesn't allow someone to work efficiently on it.

Also server access should stay in the end of one and one person only.
Nothing good can come if any contributor can push any code in production when he feel like it.

I still hope that the new "owner" will share his power with one or two PPL he fully trusts (may not now but in near future) So we don't run into such a disaster again.
To stay at Topic:
Now the community can look into the code and make a documentation for it. So the development will get fluid. As far I will see the first server-code I will investigate and make my notes. Share them with others and spread my acknowledge. I'm very sure other will do the same.

[Ze_PilOt]

So the development will get fluid.

A development is fluid when everyone is given a specific task to do, and everyone is moving in the same direction with the same goal.

I don't think people committing code as they want can lead to something good in the end.

[no_need]

So the development will get fluid.

I don't think people committing code as they want can lead to something good in the end.

I just created an account to answer to this thread. I'm a software engineer for some years, and I can say you really should trust Ze_Pilot here.

Some leadership is required for FAF management. It's impossible for a project that big for everyone to commit changes to the code. Same for production server, you can't just give access to the server to people so they can push their updates. Leadership doesn't mean one person only, it means a very small group of persons that can easily communicate and cooperate.

If a random FAF player wants to add some bits of code, the standard way to go for open projects is to submit a "patch" that'll be reviewed by the code leaders. Then, and only then, if the patch is reviewed and accepted it may be integrated in the main branch eventually.

For the builds / roadmap and production server management, it's even more complicated. Only a very small group of people should be in charge. Deciding to release a new version has many consequences (every FAF player upgrades its client app...etc), so it should be done very carefully. If a significant modification must be applied on the server, it should be tested on a pre-production server first, then planned for release on the production one. There is no place for improvisation here, so only a very small team should be in charge.

I hope the new owner won't make the mistake to give too much power to the community. Creating a mod in a thing, modifying / deploying a new version of a core component is another story.

Don't go too fast. Releasing a GPLed code is an important first step, so everyone can have a look. Opening an official project page with a dev wiki / bug tracker / patch submission page / code repository (read only for everyone except the leaders) is a second important step.

That's, of course, only my humble opinion. Hopefully the new owner has experience with software development and project management.

[IceDreamer]

What you just described, no_need, is exactly what is currently being planned and set up. The small group with actual push access on the server is looking to be 4, maybe 5 people, all of whom have collaborated closely in the past and trust each other. And no, I'm not one of those

[rootbeer23]

https://github.com/faforever

all i see is the code. did i miss the link to the mailing list/forum?