Thanks Camillo for taking the time to answer this
Very interesting to know you follow this too and knowing you expect a 10% increase per year is a good indicator for to see what s happening.
First to put this into context. I am part of the
Nativescript community. We are using v8 as a Javascript engine for cross platform mobile applications.
We compile v8 then we compile a .so lib which is used inside apps.
That so lib went from 14.2Mb with v8 8.3 to 17.6Mb with v8 9.7
Which means 20% increase i a year and an half. We are not so far from what you explain.
Now maybe you can help us see if we build optimized enough builds of v8.
When building v8 we actually build only `
v8_compiler v8_base_without_compiler v8_libplatform v8_snapshot v8_libbase v8_bigint torque_generated_initializers torque_generated_definitions`
Then we create our v8 lib with all the .o of those modules.
Here is the
build script we use
Here are the build flags we use
`use_goma=false is_clang=true enable_resource_allowlist_generation=false is_component_build=true v8_use_external_startup_data=false is_official_build=true use_thin_lto=false is_debug=false symbol_level=0 v8_enable_i18n_support=false target_os=\"android\" v8_android_log_stdout=false use_custom_libcxx=true cc_wrapper=\"ccache\"`
Do you think we could improve it in any way? is there any .o we can filter from those modules?