I will attempt to clarify my position, and bring the discussion back on topic.
When I described CouchDBX as a toy, I didn't mean it in a pejorative sense. I actually use CouchDBX a lot for my own work with CouchDB, because it's much easier to set up than installing CouchDB as a system service. It also feels cleaner, because I know that I can just blow it away when I'm done with my testing. When I say that I can see the use case, I am not paying lip-service. Jan did a great job with this OS X app, and if we could provide something similar, for other people, that'd be grand. So, why all the hot air? Well, I am concerned about two things, primarily. I am concerned about the difference between CouchDB as a system service, and CouchDB as an isolated application. My perspective is probably skewed a little, making me value the former over the latter. But still, I think it's important to remember the distinction, and to communicate that to our users. Secondly, I am concerned about the resources available to the release team, and the provenance of the files that we instruct our users to download. Let's assume my first concern can be adequately addressed with good documentation. So what of the second concern. Well, if we were going to provide official binary downloads, we would have to find a solution to two problems. The first is getting reliable access to the platforms that we want to support. The second is finding a way to get a quorum of voters for each of the platforms. We'd need three +1 votes for each platform, which implies three voting users of that platform. We have a problem finding three voting users at all sometimes, so it is my expectation that this would be a difficult task. However, as Jan mentioned in his original proposal, we can side-step this problem by hosting the files elsewhere. That would mean that they were unofficial. Which is where my concerns about provenance come in. The voting thing isn't just a big of red tape we have deal with, it's there for a reason. It ensures the quality of the release artefacts, and prevents foul play. I think we all know that anyway, but it's important to bring it up. Unofficial release artefacts side-step our community safe-guards and, as Ryan points out, contribute to the perception that CouchDB is "fragmented" across lots of different sub-initiatives. That worries me. Having said all that. Let's imagine that Jason and a team of happy gnomes decided to package the output from the build-couchdb tool for lots of different platforms. And let's imagine that you could download those packages, unarchive them, and run CouchDB (as a stand-alone application) in a matter of seconds. And let's imagine that they were made available from a page on the IrisCouch website, organised in a grid, depending on version and platform. And let's imagine downloads.html linked to that page. Would that be a bad thing? Nope, I don't think so.
