You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@daffodil.apache.org by "Steve Lawrence (Jira)" <ji...@apache.org> on 2025/06/05 21:02:00 UTC

[jira] [Created] (DAFFODIL-3002) CLI reading/writing a fifo file fails with 0 bytes available error

Steve Lawrence created DAFFODIL-3002:
----------------------------------------

             Summary: CLI reading/writing a fifo file fails with 0 bytes available error
                 Key: DAFFODIL-3002
                 URL: https://issues.apache.org/jira/browse/DAFFODIL-3002
             Project: Daffodil
          Issue Type: Bug
          Components: CLI
    Affects Versions: 3.10.0, 3.9.0
            Reporter: Steve Lawrence


Daffodil 3.9.0 modified the CLI so that it would mmap file inputs, which gives pretty significant performance improvements for large files (DAFFODIL-2921). However, this does not properly detect that the file is actually normal file where mmaping works.

For example, if you try to use a fifo file as the input like this:
{code}
daffodil parse -s ... fifo_file
{code}
Then we end up mmaping zero bytes (because a fifo file has a length of zero), and daffodil will have zero bytes available for parsing.

We should fix our CLI so it detects if a file is a normal file, and if not fall back to the normal non-mmap processing.

A workaround to disable mmap in the CLI is to parse data from stdin. For example, the above can be replaced with this:
{code}
daffodil parse -s ... < fifo_file
{code}
This does make it impossible to debug data from a fifo file since the CLI debugger uses stdin for debugger commands, so this does need a more permanent fix.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)