I really have to take issue with this statement: "As stated above the repository is the archive where the original is maintained."
That is simply not true. The repository is, where necessary and not unnecessary textual information, where the document you are looking at can be found.
If you are looking at micro-film of the US Census, the repository is that library - but I contend that putting the library for common and ubiquitous sources like the US census is totally unnecessary (although the word "microfilm" would be a nice touch,as I gave in my example), especially if one is going to put "NARA" or "US Archives" or some such verbiage in either the author or publisher input "box".
A source (ie transcription) of a source (ie original) is simply that.
The two are NOT the same.
For example, I can extract a marriage from one of the most used references in New England genealogy:
Clarence Almon Torrey, New England Marriages Prior to 1700 (Genealogical Publishing Co., Inc., 1985).
If I wish to pursue that source further, I can get a copy and read the preface to find out that the files that Torrey accumulated over a lifetime to come up with that work were donated to the NEHGR and I could go there for a visit one day if I wish. If I kept up with the current literature, I would learn that a team of folks teamed with the NEHGR to publish extracts from those many linear feet of files in book form to give some indication of where he was getting his info. That is another source. But each tidbit continues to go back and back to published genealogies, published NEHGR and other periodical articles, vital records, wills, consultations with other New England genealogists, and many more sources.
The repository for my copy of this book is "my library"; but since this is a ubiquitous and common source, available in libraries across the country, I leave the repository empty. No info is added by my placing "my library" in that box - it is extraneous information.
Now, to get back to the relatively simple example of the US Census. First, the source of the transcribed records of the ancestry.com (which is what users will normally merge into their program) reposes on ancestry's servers. It is owned intellectual property of ancestry and exists nowhere else. The url to the specific page could be a reference for a repository for a transcribed record, or just reference to ancetry.com anywhere as the author, publisher or repository is all that is needed to refind the record. The repository is NOT the original paper copies of the census (at least two generations removed) in the archives in Washington, D.C.)