fix: C macro memory issue

[MartinKavik]

When the app in quickstart-webpack uses Seed 0.7.0 and C! (instead of class!), then the WASM file size (in debug mode) has 7.58 MB and fails during the start with CompileError: wasm validation error: at offset 26669: too many locals (build is also pretty slow).

Both new helper functions introduced in the PR - _fill_all_classes and _all_classes_to_attrs - reduce the WASM size by +/-2 MB so with this PR the result size is 3.15 MB.

I've resolved this issue by intuition, but I don't really know why this happens, how to prevent it and if we can define a general rule - e.g. "macros should be as small as possible".

Opinions? (cc @rebo, @akhilman, @flosse, @TatriX)

[akhilman]

As far as I understand, each macro call is replaced by the code generated by the macro. Thus, a large number of complex macro calls can significantly increase the number of duplicate code, which is not good. I agree that we need to move code from macros to functions as much as possible.

…

On Tue, 2020-05-12 at 05:24 -0700, Martin Kavík wrote: When the app in quickstart-webpack uses Seed 0.7.0 and C! (instead of class!), then the WASM file size (in debug mode) has 7.58 MB and fails during the start with CompileError: wasm validation error: at offset 26669: too many locals (build is also pretty slow). Both new helper functions introduced in the PR - _fill_all_classes and _all_classes_to_attrs - reduce the WASM size by +/-2 MB so with this PR the result size is 3.15 MB. I've resolved this issue by intuition, but I don't really know why this happens, how to prevent it and if we can define a general rule - e.g. "macros should be as small as possible". Opinions? (cc @rebo, @akhilman, @flosse, @TatriX) You can view, comment on, or merge this pull request online at: #455 Commit Summary fix: C macro memory issue File Changes M CHANGELOG.md (3) M src/shortcuts.rs (35) Patch Links: https://github.com/seed-rs/seed/pull/455.patch https://github.com/seed-rs/seed/pull/455.diff — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.