forked from servo/servo
-
Notifications
You must be signed in to change notification settings - Fork 0
Workweek script crate
jdm edited this page Nov 11, 2014
·
1 revision
- zwarich: Should we split it up? Could we split it up?
- jdm: I'm happy to talk about concrete ideas, but I don't think this needs to be a general topic of discussion.
- kmc: we may literally run out of ram while compiling the script crate.
- cgaebel: librustc only takes two 2gb.
- kmc: I suspect the DOM could be bigger than that. What percentage of time/lines is the generated bindings code?
- mbrubeck: Just counting newlines in .rs files in components/script, there are 95k total, including 62k in generated binding code.
- dherman: Is compilation speed THE problem, or are there others? I guess modularization isn't a big deal. An inheritance feature being limited to one crate doesn't matter since you can still split across as many files as you want, right? It's just a speed issue, not code organization.
- zwarich: Feasibility, too. Memory limitations on compiling devices.
- kmc: There's a knock-on thing where I might have multicore but not enough ram to build several things together.
- zwarich: If we think we're likely to want to split it, then we need to make cross-crate inheritance a requirement when discussing that feature for Rust 1.0+.
- jack: Did we decide that only backwards-compatible proposals will be considered?
- zwarich: Compilation strategy everyone wanted was C++ single-inheritance. It was a messaging thing that since it was intra-crate that would dissaude people from using in interfaces.