You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've just been looking into compile times for Blitz. And read-fonts (16.14s) and skrifa (9.61s) both stand out in the profile. This is for a release build:
This mostly gets hidden by parallel compilation (the non-codegen times are not nearly so bad). And perhaps this is the only way to get the runtime performance and type safety of this crate. But nevertheless I though this should be tracked, and considered when implementing changes to these crates.
(note also the compile time for ttf-parser at the bottom of the screenshot - only 5.5s!)
The text was updated successfully, but these errors were encountered:
thanks! the codegen-based approach to the design of read-fonts does mean that we often end up compiling a lot of code, so naively I imagine that is most of what's going on here (and a quick look with -Ztime-passes doesn't show anything egregious). There are some design choices we could reconsider if this ends up feeling serious.
I've just been looking into compile times for Blitz. And
read-fonts
(16.14s) andskrifa
(9.61s) both stand out in the profile. This is for a release build:This mostly gets hidden by parallel compilation (the non-codegen times are not nearly so bad). And perhaps this is the only way to get the runtime performance and type safety of this crate. But nevertheless I though this should be tracked, and considered when implementing changes to these crates.
(note also the compile time for
ttf-parser
at the bottom of the screenshot - only 5.5s!)The text was updated successfully, but these errors were encountered: