Asset pipeline approach autoreload about js-routes

[mjtko]

I'm not certain that it's safe to turn of auto-update in this environment. The trivial approach would be to only register the preprocessor if Rails.env.development? or something, though this would then break for those who wanted compilation to occur in production (ie. weren't using precompiled assets).

It seems that the preprocessor has to be registered before the assets pipeline is finalized - I'll have a look to see if I can work out where and how to hook this in during initialization.

[bogdan]

I am experiencing even more problem:

In recent Rails rc - 3.1.1.rc2 rake assets:precompile loads environment but doesn't initialize it. And seems it will bring even more serious problem.

http://weblog.rubyonrails.org/2011/9/29/ann-rails-3-1-1-rc2

[mjtko]

There is a problem with asset pipeline approach in production environment:

I was wondering why I hadn't seen this issue myself and it seems this is only the case if config.assets.compile is set to true in production. While this is an unusual configuration, I think it's still a use case that should be dealt with. I think we can set the initializer to run at a better point during rails initialization - now I've managed to reproduce the issue, I'll work up a solution. :)

In recent Rails rc - 3.1.1.rc2 rake assets:precompile loads environment but doesn't initialize it.

Indeed we should, again, be able to sort this out with some initializer magic. I'll look into this too.

And seems it will bring even more serious problem.

Anything specific you're concerned about?

[mjtko]

In recent Rails rc - 3.1.1.rc2 rake assets:precompile loads environment but doesn't initialize it.

Looking at the 3-1-stable branch, it seems that they've made the sprockets precompilation without loading an environment optional (I belive it was implemented to workaround some problem that Heroku users were experiencing). By default, in 3-1-stable, the behaviour is as before - load the full environment - so we should be ok. Phew. :)

I've got a fix for the initial issue, will submit a pull request later this European morning. :)

[mjtko]

Good morning @bogdan :-)

I've opened two pull requests with patches for the issues discussed in this ticket. Looks good under my test environments here - hopefully these should cover us under 3.1.1.rc3 and up. :)

Cheers,

Mark.