Connecting physical to logical names

[DarioE]

The problem I have is that on my disk I can find gigabytes of various hash file folders + maybe some temporary sorter or aggregator files..
The programs that created those files are long deleted and I have no clue what they are.

Is there a way to link the physical names of files and folders on a disk to logical? It would also be of great help to find the name of a program that created them. That way it would be possible to detect which files are in use and which can be safely deleted.

Thanks!
DarioE

[I_Server_Whale]

Is there a way to link the physical names of files and folders on a disk to logical? It would also be of great help to find the name of a program that created them.

I highly doubt it unless the file/folder names (physical) have some kinda relationship embedded in them that relate to the logical.

Whale.

[ArndW]

The file names for pathed hashed files can be extracted by doing a full project export to a .dsx file and parsing the export file. This isn't too difficult to do in a simple DS job using one transform and a stage variable to keep track of the job name being processed.
This will let you find out which programs used which hashed files. As long as the files aren't in a project directory you can remove them from the OS level directly if you are certain they are no longer required and that any jobs that use them will re-create them.

[ray.wurlod]

The $DSHOME/bin/UVfile command will report whether a particular file or directory is a hashed file. The existence of DATA.30, OVER.30, .Type30 and nothing else in a directory signals that that directory is a dynamic hashed file. You can encapsulate this in a script that traverses your file system looking for hashed files.

Arnd's approach is sound until you encounter hashed files whose names and/or parent directory pathnames are given as job parameters.