#310 Archive file containing a file with a colon: windows problem

[John]

Running a job through DPR which contained a TAR file. Inside the TAR file, was a zero-length file called "Ne01:"

Xena exploded the tar file and wrapped the file. The output file was named in the standard manner - which included the name of the original file.

Was able to process the job into the Linux repository - without any problem.

However, when trying to copy the AIPs into the Windows repository, DPR claimed that it could not find the file, despite being able to search for it. It turns out that a colon is an illegal character in file titles for Windows. Obviously the NTFS specification allows you to write files (using linux) with characters that are not legal in Windows. Getting the filesystem specification modified to prevent this happening again would be a good exercise for our new Software Product Manager. :)

Xena should not allow filenames which contain illegal characters in _any_ filesystem/operating system.

We probably should also discuss the issue of zero length files in general - they would be hard pressed to be called "retain as national archives", but we cannot just ignore them wholesale.

[John]

I will also leave the problem of making sure that both the Linux and Windows repositories have the same stuff in them to another time.

[Daniel Black]

since you pseudo assigned it to me anyway :-) Thanks for the detailed comments. I probably will deal with your comment #1 as that's part of a OAIS / TRAC assessment criteria.

[Daniel Black]

Reference: http://msdn.microsoft.com/en-us/library/aa365247.aspx

[Michael Carden]

This looks both old and high priority. Probably worth a look.