I am in the process of organising all of my documents (pdf/ps/etc.) and putting them into an annex. It seems like the easiest (and smartest?) way for me to manage my 15,000+ document library... (Yeah, I've not read them all!)
However, one of the issues I have run into is when I want to include something like a software manual...
I'm not even sure if git-annex is the correct way to handle these sort of document, but...
What would you do if the URL stays the same, but the file content changes over time?
For example, the administration manual for R gets updated and re-released for each release of R, but the URL stays the same.
http://cran.r-project.org/doc/manuals/R-admin.pdf
I'm not particularly worried about whether my annex version is the most recent, just that if I want to 'annex get' it, it will pull a version from somewhere.
Any thoughts?
I'd bet that the SHA-hash would change between releases(!) so would a WORM backend be the best approach? It would mess up my one-file-in-multiple-directories (i.e. multiple simlinks to the one SHA-...-.....) approach.
The web special remote will happily download files when you
git annex geteven if they don't have the same content that they did before.git annex fsckwill detect such mismatched content to the best ability of the backend (so checking the SHA1, or verifying the file size at least matches if you use WORM), and complain and move such mismatched content aside.git annex addurl --fastdeserves a special mention; it uses a backend that only records the URL, and so if it's used, fsck cannot later detect such changes. Which might be what you want..For most users, this is one of the reasons
git annex untrust webis a recommended configuration. Once you untrust the web, any content you download from the web will be kept around in one of your own git-annex repositories, rather than the untrustworthy web being the old copy.Thanks!
git annex addurl --fastdoes exactly what I want it to do.Wow. Yet another special backend for me to play with.