Giving back

[rcurtis]

Greetings,

  I work for a company that uses SFML and several of it's libraries every day.  Yesterday the topic of how to give back came up in a meeting.  I would like to hear from the core members on how they think I might be able to support the project.  Giving code is probably not an option, our code is very specific to our use case and is probably no good to anyone else.  We would, however, be interested in giving back in some other way.  I would like to know what the project could use and how my company can help.

Thanks.

[Nexus]

Contributions are always very welcome; code patches are not the only way. Very helpful is testing, especially on systems/configurations that are not common. Bug reports in the forum are also highly appreciated (much more than workarounds in your projects, which help no one else ).

We get most feedback from people working with SFML in their free time. As such, companies that use SFML can provide us with another perspective. Personally, I would find it very interesting to see how SFML is perceived and used -- not primarily with concrete code, but rather to know what criteria/points you consider important in a library like SFML. For example, why did you choose SFML, what features do you consider the most beneficial, and in which places are you most limited? Have there been decisions in the SFML development with an extraordinarily positive or negative impact on your projects? Other opinions of you and your colleagues about the library?

I don't know how open or publicly accessible your products are, but in case they are, a nice option would of course be to present them in the projects forum.

[binary1248]

Like Nexus already said, the most valuable asset to an open source project is its community and user base. Knowing who uses SFML and how they use it helps the library evolve in the right directions. Without feedback, all we can do is make assumptions based on what we perceive and that might not always be accurate.

That being said, we are always looking for people to test new feature implementations and report bugs. The more data we can collect, the more confidence we have in the test results. The last thing we want is progress on an issue to be blocked because of something unrelated on the reporter's system. Getting prompt feedback in regards to development is also nice to have. We've adopted a faster release cycle, which means we are more willing to commit less stable code to master hoping it will get tested by more users that way. The more people use master, and the more people report issues they have with it the faster we can respond and make sure everything gets fixed before the next stable release.

I don't know what policy your company has in regards to the version of SFML that it uses in its products, but if it is possible to test them against a master build of SFML and report whether the issues which have been declared as fixed are really fixed, that would be great. It would add to the confidence of newer code and accelerate the rate at which code can be considered stable. The reverse is true as well, if code is changed that ends up breaking your product(s), it would be nice to get a prompt report as well. The others and I are keen on slowly clearing out the tracker of issues and one way of doing that is by making sure newly reported issues are given higher priority so they can be dealt with while the reporter is still around and can provide information on how to reproduce the issue.

A few other random points off the top of my head:

• Free publicity is always nice . As stated in the licence: If you use this software in a product, an acknowledgement in the product documentation would be appreciated but is not required.
• Maybe this belongs to the first point, but testimonials are also advantageous. Every now and then, people come to the forum asking whether SFML has been used in commercial projects. There are a few projects that are known to us, and there are probably a few/many more that aren't, but without actual first hand proof, we don't have much to back up our claims.
• SFML is always accepting donations and we are working on setting up a somewhat usable CI infrastructure. There were/are discussions on whether we should get a dedicated OS X node, and I'm probably not the right person to ask about that . All I know is that it isn't something that SFML can afford right now considering the team works on it pro bono.

[rcurtis]

Thank you for the replies.

Binary, could you tell me more (or connect me to the right people) to discuss concrete details of your last suggestions (the CI infrastructure).

Thanks,
Robert

[Jesper Juhl]

I'm not connected to the SFML project (except as a user of the library), but I'd still suggest that if you could throw some money at their CI efforts (like a Mac Mini or something) it'd be great. Or if you could perhaps upstream whatever fixes to SFML that you had to do in your own code; that might help others.

I for one would like to know what you use SFML for (what's your product? where can I check it out?). And I'd love to see (when I do check out your product; where is it?) that there's some prominent notice that it uses SFML :-)

In the end you could always just throw money at the project and see what falls out ;-)

[FRex]

If you are a reasonably notable company in your field and mention somewhere on your official website that you use SFML maybe that could serve as a "Reliable Source" so you could give someone (who is willing to try start a "discussion" on wikipedia) an argument to pull SFML wikipedia page off the 'delete' status, which it is in for years, since it's apparently not notable.

[Hiura]

Binary, could you tell me more (or connect me to the right people) to discuss concrete details of your last suggestions (the CI infrastructure).

Ideally, we'd like to set up some regression/quality tests (#623). The first step was to setup a Jenkins (however we're also exploring some other, more flexible solutions atm) to make sure that the codebase is always compilable. But to do that we need slaves for every build configuration (OS, compiler, etc...).

And as you may know, we can't virtualise OS X so we really need a physical Mac machine for that. Currently, I provide a slave running on my laptop but only with a very low uptime. So that's not really ideal. We're considering acquiring a Mac Mini or something similar but it tends to be rather expensive... even second hand ones. Hosting it wouldn't be an issue, though.

[rcurtis]

Thanks for all the feedback.  I'll be back in touch next week.

Robert

[dwarfman78]

(however we're also exploring some other, more flexible solutions atm)

Can you please tell us what Jenkins alternative are you talking about ?

thanks.

[eXpl0it3r]

Can you please tell us what Jenkins alternative are you talking about ?

Buildbot
Originally it was the idea to use Buildbot, because it gives you the full Python language to write your build script, but in the end setting up Jenkins was faster.