Since it's quite possible to forget to upload a file, or to have the file fetched manually, and fail to update SRC_URI accordingly, it would be nice for repoman (with a fuller, network-based scan) to be able to check that the file is indeed there.
Downloading the file might be too much (if the file is huge that's baaad), but at least HEAD it (and maybe check whether the size corresponds to what is saved in the Manifest, if provided) would be a good idea, no?
i dont think `HEAD` follows redirects (3xx) ?
also, we already have FEATURES that control the d/ling of files -- assume-digests. this is specifically designed to avoid having to d/l files even some of the time let alone your proposal of d/ling the files *every* time. i cant see that ever really being useful.
HEAD returns a redirect, you have to follow through a 3xx request and HEAD the further location until you get either 4xx, 500 or 2xx.
And I'm *definitely* not suggesting to download it every time; I'm suggesting we check it's there without full download (what HEAD is for).
ping. I agree with Diegos' latest comment
*** This bug has been marked as a duplicate of bug 295335 ***