Tony Kelman

I finally decided to give this a try and started writing spec scripts for the COIN-OR stack (where most of the open-source solvers that JuliaOpt uses come from) here https://build.opensuse.org/project/show/home:kelman:mingw-coinor...

Yeah, may as well start from one of my spec files, as good as anything. Lots of my configure flags are COIN specific though. I never actually bothered figuring out...

My JuliaCon talk will close this issue.

Putting this here (already posted a few other places but lost track of where) until I get a chance to pretty this up and put it in the readme: ```...

What exactly is the failure locally with precompilation off? I frequently see an http code mismatch intermittent failure where a return code is 503 instead of 200, but these failures...

That ReadOnlyMemoryError looks entirely separate and likely unrelated to precompilation.

Given the `"Error executing request : Failed writing received data to disk/application"` and the segfaults, I suspect this package isn't precompile-safe as currently written.

See https://travis-ci.org/JuliaWeb/HTTPClient.jl/builds/88606158, I just restarted that. Master passes, or at least passes some of the time (there's an intermittent http code mismatch failure, but I don't usually see segfaults) with...

Yeah, the former is explicitly the syntax for a 1-element tuple, the latter gets implicitly lowered to the same thing but that could be pretty surprising.

While it would be ideal not to be changing return types of these operations post-feature-freeze, if the current behavior is considered buggy / incomplete then this can hopefully be worked...