Files are loaded into memory instead of streamed

[enumag]

Hi, I checked your source code and found this line:

https://github.com/xresloader/upload-to-github-release/blob/8e0ffc2bf70a36035a8b5479d0903041cb11423f/src/index.ts#L721C37-L721C49

Releases can contain several large files so this way the runner can easily run out of memory and crash.

I found this while looking for alternatives when I discovered the same issue in another similar tool.

[owent]

Good point and thanks.

[enumag]

I'm no JS expert but probably you need to use something like this: https://www.geeksforgeeks.org/node-js-fs-createreadstream-method/

[owent]

I'm no JS expert but probably you need to use something like this: https://www.geeksforgeeks.org/node-js-fs-createreadstream-method/

GitHub's APIs do not accept read stream as data parameter. I will try to use @file_path to see if it will use stream internally.

[enumag]

Ouch... that would be rather nasty if Octokit itself requires the data to be loaded in memory. 😳 I hope it's not the case.

[owent]

I have raise #55 to try to use @file_path to upload file. But I have no repo which can be used to test a large asset file. Could you please try v1 branch or main branch to see if it work?

[enumag]

It doesn't work. The action succeeded and seemingly did actually upload the files however the files are just a few bytes large and contain the path to the file in the worker rather than the correct content.

[owent]

Thanks, I will try to find another way to fix this problem later.

[github-actions[bot]]

This issue was marked as stale due to lack of activity.