Startup times & stagedfunctions

[timholy]

The load time for the package are fairly long, largely because lots of methods need to be generated. The load time could probably be drastically shortened using stagedfunctions, which allow the methods to be generated on-demand.

If you get to the point where you are ready to switch your development to 0.4, I will gladly convert things over to using stagedfunctions.

[tomasaschan]

I have been thinking a lot about this too - to the point where I've started writing, then discarded, a post to julia-users asking where the documentation for stagedfunctions is, if it exists, and whether the feature is ready enough for some real-use-case tire-kicking. I have put off dropping 0.3 support until Julia master is moved to 0.4-pre status (by the end of the month?). When we do move to stagedfunctions, I'd be happy to do at least some of the converting work myself - I don't know when next I'll get such a good use-case to try it out on =)

In the meantime, my workaround when testing various things has been to comment out most options for IT and EB in the loops generating getindex and gradient methods, which pulls down the load time to an acceptable second or so for one or two cases.

But yes, we should definitely do this.

[timholy]

I'm being begged to write stagedfunction documentation in multiple julia issues already.

[mbauman]

Bump! Now that @ngenerate has been removed from Julia, Iterations is broken on 0.4… even with Compat.jl since it uses a private ngenerate function that isn't being supported.

[tomasaschan]

@mbauman Thanks for the notification! I've been swamped by other stuff (work, mostly...) lately, so I haven't been able to put as much time as I'd like into this project. I'll see when I can get to it next, but I won't complain if anyone else gets to it before I do.

[timholy]

I likewise plan to get to this if no one beats me to it, but it won't be before Wednesday.

[timholy]

Fixed by 6da1207