Need to retain old file name in datastream

Project:RUcore dlr/EDIT
Version:7-x
Component:Code
Category:feature request
Priority:normal
Assigned:martyb
Status:closed
Description

Now that the Label field in a datastream does not contain any data that identifies the original file (e.g., it used to include the title of an ETD), there needs to be a field added that ties back to the original file. So, for example, the original file name for an ETD (e.g., ETD-2013-405 as stored in ALT_IDS in the XML) needs to included in the datastream.

Comments

#1

Category:bug report» feature request

I'm not sure the datatreams have ever included the ETD ids in the label. The labels have traditionally had the mods:title followed by ": DSID", e.g., "The Title: PDF-1". The altids have not been used up till recently. The ETD names such as ETD-2013-405 are saved in a mods:identifier, however, and so could be added as altids, though that would involve versioning the datastreams.

#2

Jeffery,

This is not just applicable to ETD. WMS is writing original file name in ALD_IDS attribute in the datastream section along with filesize. If we have to replace a datastream in Fedora, we are going to lose the original file name and filesize info. We discussed using filesize and alt_ids attributes at several sw_arch meetings.

What would happen if we have to replace a zip file or tarfile for datasets? The original files names will be in the SMAP-PRES but if they are not in the datastream, we wont be able to,easily identify files. I am not even sure if package download will work as expected if the original file name is not in the datastream.

May be I am not understanding the whole purpose of storing original,file names.

This needs further discussion at a sw_arch meeting.

#3

It's a good idea for a feature. We can add in altids with Fedora rest now, though it was not built into this version of the dlr/EDIT interface. As for the zip tar question, even the current changeds will not affect any established altids - they get inherited from the previous versions of the datastreams. Adding new ones will require a setup change and versioning, but will be no more difficult than changing labels. If a pinch, the native Fedora API can be used now to add a new datastream with any valid attributes, but it is not especially user friendly. We will want to build something like the labels form.

#4

Status:active» test

These features are actually in place as of 7.1.

#5

Status:test» active

I tested this on rep-devel today (06-17-2013) and it appears that it is not working correctly. I ingested an ETD (rutgers-lib:201519) and exported the object in dlr/EDIT. The PDF-1 datastream contained an ALT-ID field whose value was the .pdf filename. Then I replaced the PDF in dlr/EDIT and exported the object. The new PDF-1 datastream was missing the ALT-ID field.

#6

Status:active» test

Did you use the change a datastream page? Did you enter the filename you wanted in the "altID for Datastream? (e.g., Original Filename)" field? the dlr/EDIT system preserves the old altID when changing something like a label that is not expected to change the file itself, but if you are replacing the file itself, it does not assume that the filename is the same. You have to put the altID you want (the same or a different name) in field on the form.

#7

Assigned to:triggs» martyb
Status:test» closed

Oops, my bad. I used the original method I learned, which was to purge and add the datastream. When I used the Change Datastream and entered the filename in ALTID, it worked as expected.

Back to top