hpc / dcp Goto Github PK

Due to the nature of the truncate option being in its own (cleanup) stage, when it fails, it sends control back to the copy stage which has no knowledge of the failure. This causes an infinite loop when the copy stage calls the cleanup stage again.

Configure currently checks for xattrs, but the code uses xattrs anyway. The code should recognize the DCOPY_USE_XATTRS (or similar) define.

Dcp should support all 7 files types in linux, not just regular files, symlinks, and directories.

The man page for the recursive -R option is incorrect.

Just figured y'all would want to fix that link to point somewhere else.

What's going on: If the source file is empty, the destination file is not created.

What should happen: If the source file exists (even if empty), the destination file should be created.

Due to limitations in libcircle, file path lengths are limited to less than 4096 characters (depending on how libcircle was configured).

Preserve permissions is not implemented.

It would be useful to update the README.md text to indicate this project is abandoned and point explicitly to hpc/mpifileutils? It seems that the http://filecopy.org/ URL already directs there, but this is non-obvious.