FITS is god-awfully slow. Let's replace it.

[mjgiarlo]

And it's also not great for A/V files. Since we've already got hydra-file_characterization, which wraps characterization tools, and the only benefit of using FITS is that it wraps characterization tools, I vote in favor of leaving FITS behind. Another downside of FITS is that the released versions often wrap outdated versions of the tools. Instead of FITS, we could directly wrap JHOVE(2), Exiftool, MediaInfo, file, DROID, and possibly others (FIDO?).

[dchandekstark]

👍

[jcoyne]

I believe that @mjgiarlo picked FITS for Scholarsphere (which became sufia, which spun off hydra-file_characterization), so I believe he is highly qualified to make this decision.

👍

Is Penn State going to lead this effort?

[mjgiarlo]

I'm also at a meeting where I've learned about some of the inefficiencies of FITS, hence where this came from.

If I were writing code, we'd lead this effort -- our other devs are maxed out on our ScholarSphere 2.0 effort. Maybe it's something I could hack on at LDCX, though?

[dchandekstark]

+1 on LDCX hacking

[ruebot]

subscribes

[mjgiarlo]

Brief update: this was not on the docket for LDCX, but now that we have @awead on staff, who is keenly interested in this work, things may get moving soon!

[ruebot]

Since y'all use Solr and we use Solr in Islandora, the Tika integration is pretty great. But iirc, that comes from the GSearch side of things that y'all don't use. My outsider 2¢.

[mjgiarlo]

We do in fact have Tika in our Solr install, and though Tika wasn't listed above, it is on our Redmine ticket for this (an older ticket that pre-dates Sufia and our usage of GitHub issues). Thanks for adding that into the mix, @ruebot!

[mjgiarlo]

We may attach this work to an upcoming sprint with @jcoyne (et al.) this spring/summer. Stay tuned, and pls keep the great ideas coming!

[ruebot]

Glad to hear!

[ruebot]

Just remembered... y'all know about FITS with Nailgun? Might be worth bench marking with other characterization tools.

[mjgiarlo]

Yeah, thanks for adding that to the discussion. I believe @mistydemeo may have done some benchmarking there and it seemed like she was still leaning towards ditching FITS for Archivematica.