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
Now, if we then somehow could get rid of libsass we will have a CGO free build (no need for extended), which would be a massive win. We would want to continue producing theextended archives, of course, to make sure we 1. Avoid breaking builds in the wild (e.g. Netlify) and 2. In case we need to re-introduce it in the future.
The text was updated successfully, but these errors were encountered:
Just a few notes. Currently, there is no way to initialize a Wazero runtime, it will do it lazy if it is not initialized. The Wazero docs clearly state that panic can happen when concurrently initializing the same runtime, and my usage is such that I don't want to do that unconditionally in init() (i.e. I use it for the image.Decode and without concurrent code).
There is sync.OnceFunc is in the code but it is not enough. I have never seen it for Webp because it works really fast, but for other libraries I have (avif, jpegxl, etc.) that is the case. I will add Initialize() when I find time so the user can do it manually.
The other issue is that library will first try to use a dynamic/shared library (if found) via purego. Due to the way purego works (at least on Linux), the binary will always link to libc, even with CGO_ENABLED=0, so there are again issues if you compile on e.g. recent glibc and try to use it on some older system. You can disable purego with the nodynamic build tag, then only WASM will be used.
That's a great idea, since it would also allow (in the long run, since there certainly still be need sort some sort of wrapper / mapper) for more image formats available via WASM, like libultrahdr-wasm for #12781
See https://github.com/gen2brain/webp
Now, if we then somehow could get rid of
libsass
we will have a CGO free build (no need forextended
), which would be a massive win. We would want to continue producing theextended
archives, of course, to make sure we 1. Avoid breaking builds in the wild (e.g. Netlify) and 2. In case we need to re-introduce it in the future.The text was updated successfully, but these errors were encountered: