[Brainstorm 2010] Garage and developer services
 

This thread is part of the Maemo Brainstorm 2010

Subversion, Gitorious, autobuilders, issue trackers, mailing lists... what can be done to improve the Maemo application developer experience by the community?

Propose your ideas and suggestions here, let's talk about it, and summarise the best ideas and conclusions in the Brainstorm 2010/Garage and developer services wiki page

Re: [Brainstorm 2010] Garage and developer services
 

I'd like an option that the autobuilder does not add the packages to the repository when they were build successfully. Instead I want to get notified and be able to download the package before it's uploaded.
Otherwise you always have to make a new version number when the autobuilder did other things than the local sdk.

Re: [Brainstorm 2010] Garage and developer services
 

Also is it possible if a package build fails, the autobuilder to offer an option to scrap the package, so that we can reuse the version without changing the package itself (many of my build failures are actually unsatisfied dependencies and I don't want to change the version number and the changelog for that)?

Re: [Brainstorm 2010] Garage and developer services
 

Some of the developers dislike Bugzilla, is there a possibility Garage to offer a radio choice of several bug trackers?

Re: [Brainstorm 2010] Garage and developer services
 

It's been a while that I have been wondering about the mid term usefulness of http://garage.maemo.org . It made a lot of sense when it started in 2005, but nowadays there is a variety of rock solid services providing all those services, only not in a single place.

But how relevant is to have an ok-ish single place when with simple links you can have at reach free infrastructure like gitorious.org, transifex.net, wordpress.com, your closest bugzilla, etc?

The last garage project I created had as basically main purpose the obtention of a mailing list. I guess we could also have a default recommendation for project mailing lists.

Conclusion: http://gforge.org itself is not a simple beast to maintain, leave alone to customize, and the benefits nowadays are probably slim. So why not leaving it in pure maintenance mode, pointing developers to the current better alternatives instead of trying to add even more beef there?

Re: [Brainstorm 2010] Garage and developer services
 

I never had this problem. I use always the same version number
if a package build fails. I use the webinterface to upload new
packages.I know packages are rejected if there is one already
with this version, but only if the build for this one succeeded.

nicolai

Re: [Brainstorm 2010] Garage and developer services
 

A garage project can have several extra tabs (like the Lists, Tasks, Docs, Wiki etc). You could just create a tab for "Bugs" and point it to your preferred bug tracking service. See an example.

However this is just a link to a remote service, and no further integration is available. If the remote bug tracker offers RSS feeds for bugs, then we could perhaps list them on the garage pages, but nothing else, really.

Re: [Brainstorm 2010] Garage and developer services
 

Don't forget that granting upload privileges to extras happens through garage. It would take some effort to move this functionality elsewhere..

GForge is not that bad to maintain. We have added quite a bit of new functionality there, we have the know-how.

There are couple of things which our garage offers for free unlike other providers:

• possibility of running private projects
• unlimited storage (for code, docs, etc)

However I agree with the maintenance mode proposal, but if there are some great ideas we can hack on GForge, no worries.

Re: [Brainstorm 2010] Garage and developer services
 

There are some advantages to providing them in a single place, including familiarity between projects, increasing possibility of cross-project collaboration; streamlined tools for Maemo use cases; web-site hosting and single authentication/authorisation point.

However, we could probably turn some of the defaults off for new projects.

The other big advantage is the offer of svn - for many tasks this is preferable (depending on the skills of the developers) - and I'm not aware of any gitorious-like service for Subversion.

However, there any many things around Garage which do need improvement: in particular, the communication around the move to vcs.maemo.org and drop.maemo.org were very poor. With svn relocation commands being discussed when developers had problems.

If it goes into "maintenance mode", it'll bitrot. However, if someone wanted to find the best-of-breed tools which provide the same features:

• Website hosting
• Version control (preferably choice of Subversion or git, but just git'd do for this list)
• Bug tracking
• Mailing lists
• Blogs
• Documents

...we could try and ensure tie-up between the services where appropriate.

Re: [Brainstorm 2010] Garage and developer services
 

Not necessarily. As long as we have staff available maintaining the services and there are projects actively using them I don't think this will be a problem.

Re: [Brainstorm 2010] Garage and developer services
 

Perhaps the best thing would be for Garage to transition to providing an integration point for all the disparate services, and some implementations of those services itself.

Or are there already good services which allow you to combine your bugzilla from x.com, with your source from gitorious.org, with your mailing lists from mailman, your wiki pages on wiki.maemo.org and a sub-forum or thread on talk.maemo.org?

Re: [Brainstorm 2010] Garage and developer services
 

There is no doubt that garage.maemo.org plays a key role in the maemo.org infrastructure. Even something as basic as registering to the website goes through there. But here we are discussing where the new work needs to focus and my only point is that I don't think garage.maemo.org needs to be in the focus. What it needs to do it does it already.

Two generic thoughts to put this discussion in a context:

- It is fair to assume that active Maemo community software projects will point their trunk development to MeeGo as soon as it becomes a stable platform with devices announced. maemo.org will not the main place to think of developer novelties.

- We should even ask ourselves whether Maemo/MeeGo are the places to host project infrastructure in the future. Maemo has such an specific API that it almost becomes an upstream of itself. However, MeeGo takes as a reference the Qt API that is by no means unique to MeeGo. The likeliness of projects choosing MeeGo as primary target but not the only one is high (Ubuntu etc for netbooks, Symbian etc for mobile...)

In this context I think that it makes sense to keep the investment in garage.maemo.org at a maintenance level and move the attention for new features and resources to whatever exists already, is free of charge, free software friendly and well maintained. If there are gaps then the MeeGo project would be the default candidate to fill it.

Looking at the list of services posted by Jaffa it is clear that we are almost there already:

- Website hosting, blogs and documents: developers are increasingly using wiki.maemo.org (and in the future probably wiki.meego.com) and/or wordpress.com or another free blog services. If someone has more demanding needs then the chances are that these needs also hit the limits of our hosting service, and web hosting is quite inexpensive these days anyway.

- Version control: several alternatives for git, svn can be found at least at Sourceforge and Google Code, you can even use Launchpad's Bazaar if you wish.

- Bug tracking: maemo.org already has a preference to use bugs.maemo.org. There hasn't been a discussion about this in the meego.com context but I just added a proposed task at http://wiki.meego.com/Community_Offi..._committed_yet. Right now I'm busy with the MeeGo release, if you want to start this discussion in forum.meego.com you are welcome!

- Mailing lists: probably a real gap, although it would be interesting to see how many active mailing lists are there currently in garage.maemo.org. Seeing all the discussions about mailing list vs forum sometimes I think we should go for Usenet with their well tested web & mail gateways. Anyway, discussions usually go around the current development and last stable versions, and soon these will happen more in the MeeGo context. I just added a proposed task at http://wiki.meego.com/Community_Offi..._committed_yet. Right now I'm busy with the MeeGo release, if you want to start this discussion in forum.meego.com you are welcome!

Re: [Brainstorm 2010] Garage and developer services
 

Well, apart from looking crap. No-one's actually proposed any in-depth work on it, yet, though. And such proposals would still have to get through the prioritisation meeting at the end of this brainstorm.

Well, unless you're willing to go on the record saying one or more of the following, we can't go on that assumption when looking at what to do with the Maemo community.

1. Anyone who's ever bought a Maemo device will be offered an equivalent MeeGo device for free, or nominal cost.
2. The people productively driving the Maemo community (however that's defined, let's say karma > 100) will be offered a top-of-the-range MeeGo device for free, or nominal cost.
3. The N8x0 and N900 will be able to run MeeGo to such a high quality that practically no-one will use Maemo anymore (except, perhaps, 770 owners); with no loss of functionality.

The only reason to believe that Maemo developers will decamp to being MeeGo developers is that everyone will actually decamp to being Qt developers. This then depends on Qt applications being massively cross-platform, and the (assumed) wealth of them for MeeGo being trivially portable to Maemo.

However, who does that porting? Who does the packaging? Who does the testing, and where do people discuss the foibles of the Maemo implementation? Where do they store the debian/ directory for a Qt app written for MeeGo (RPM), Symbian (SIS still?) or Windows?

Yes, MeeGo's coming; but in the scope of this brainstorm we don't know yet what the impact will be on Maemo. The point of it is that we shouldn't spend the next 6, 9 or 12 months thinking "well, anytime soon MeeGo's going to torpedo us". Some of the features being proposed, such as the donation framework for Downloads, are also things which we could offer the MeeGo community and have benefit there.

Perhaps Garage isn't one of them, but the other point of this brainstorm is to ensure that in 6 months things can go into maintenance mode if necessary.

Re: [Brainstorm 2010] Garage and developer services
 

Here I agree. Garage has been a fairly low maintenance, reliable service for a long time. Don't poke at it without a good reason. And if such reasons come up, Ferenc knows where to poke.

I fail to see how those two relate to the topic. MeeGo is MeeGo and discussed somewhere else.