We have two opposing problems that are glaring at each other with fists raised:
o Is 2GB enough? Maybe not: how big is an hour of video? how big is the take from an instrument in a typical run on the Large Hadron Collider? o HTTP has problems with the uploading of huge quantities of data. The resolution, to date, has been: "give us the file and all particulars, and we'll put it in for you." Some contributors probably like that, and others probably dislike it. Some sites probably wish they didn't have to get so involved with submissions about which they know very little. A potential problem is that the item importer imports *items*, so someone has to make up item batch-load descriptors for these mondo submissions. IMHO this is incorrect partitioning: the contributor could easily submit everything else, if not for the size of the the bitstream(s). The admin.'s help is only needed for the bitstream(s). Here are a few ideas to stir up our thinking: o Attach a bitstream by URL. A submission could include a URL and credentials to pull a bitstream rather than push it. Then a background process can take as much time as needed, use protocols like FTP, keep restarting from where the connection broke off until it gets the whole thing, etc. There's potential for abuse of the server to attack others' sites with file-transfer requests, so maybe "can request pulls" is an EPerson attribute that an administrator would have to enable for trusted users. o Attach a bitstream by token. "Data to follow. Give me a UUID to identify this bitstream when you see it again." Then the contributor can fill out all the forms and launch his submission without wait or frustration. DSpace can email copies of the UUIDs to the contributor so they don't get mislaid. When the contributor's pallet-load of SSDs arrives from the loading dock and they're cabled up, the admin. tells DSpace, "that file over there is bitstream UGLY-UUID" and DSpace will associate it with the Item sitting in some workspace or workflow. This could be abstracted to facilitate the automation of uptake of files that arrive by some magical means behind DSpace's back. You'd wind up with a package descriptor file again, as with item registration, but look how simple it is: "UUID filename". No dublin_core.xml, no descriptions, no collection-IDs, no EPerson-IDs but the admin's own (for the provenance). No having to track new forms of metadata when they are implemented in the GUIs. Everything but the bitstream itself is already in an existing submission. Probably we should define a bulk-transport provider interface and let people cook up their own pluggable variations on the theme. -- Mark H. Wood, Lead System Programmer mw...@iupui.edu Asking whether markets are efficient is like asking whether people are smart.
pgpbjNDGUhBOS.pgp
Description: PGP signature
------------------------------------------------------------------------------ Keep Your Developer Skills Current with LearnDevNow! The most comprehensive online learning library for Microsoft developers is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, Metro Style Apps, more. Free future releases when you subscribe now! http://p.sf.net/sfu/learndevnow-d2d
_______________________________________________ DSpace-tech mailing list DSpace-tech@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspace-tech