Skip to content

Don't read past end of file #53

Open
@adriangb

Description

@adriangb

I'm dealing with an SFTP server that errors out if you send a read() for more bytes than the file has. Not sure if this is a common thing or my server is just not following specs (wouldn't surprise me).

I solved this by manually tracking the filesize and bytes read, but I'm wondering if there'd be any cons to just having this in the package?

Two options would be:

  • Do what s3fs does and download into a temp file so that users won't encounter discrepancies between real filesystems and sftp
  • Emulate reading beyond the last byte without actually sending requests

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugsomething does not work

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions