Wednesday, January 14, 2009

File Under: Reasons not to encode Data in File Names

Consider: Two adjoining papyrus fragments that connect on the horizontal. On one side, a relatively long text that spans the two fragments (text A). From the perspective of this text, there are two additional texts (in two different hands) in the right "margin" of the rightmost fragment, one (text B) above the other (text C). Finally, there is some writing on the other side (text D) of the joined fragments.

You have a cataloging scheme that separately identifies each text on the physical artifact. You have an image file naming scheme that effectively represents the relationship of a catalog record with an image file: {catalog id}.{front/back}.{tile coordinate if applicable}.{resolution}.{format}.

From the perspective of the text A, images of the two fragments might be named A.front.left.high.tif, A.front.right.high.tif, etc. But wait! Isn't A.front.left.high.tif also D.back.left.high.tif? And isn't A.front.right.high.tif also B.front.high.tif?

No comments: