Skip to content

Use multipart upload instead of a temporary file #47

Open
@gilbsgilbs

Description

@gilbsgilbs

Hi,

If I understand the current implementation correctly, it creates a temporary file on disk, writes everything to it, and just when the file is closed, uploads it to S3. This implementation is not applicable for very large files and is probably slightly suboptimal in some circumstances. For these reasons, I think it would be beneficial should s3fs could take advantage of multipart uploads.

There's a lot of gotchas and boilerplate associated with multipart uploads though. One possibly simpler implementation could be to delegate the hassle to a third-party library that already does the proxying stuff, such as smart_open (great library by the way). Needless to say that all this could be opt-in.

Thank you so much for all the great work on pyfilesystem2 and s3fs 👍

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions