Deal with the case of an optical drive with media.

[trgill]

PR #350 deals with the case of "no disk". We should test and deal with case when there is a disk.

[mulkieran]

If there is an optical drive w/ disk, shouldn't we just read the disk?

[trgill]

I'm not sure, but if we do - we need to deal with the possibility that it is read only. Not sure what that will do to our metadata update strategy etc.

[agrover]

Easiest to just not support this, I don't see how it enables any realistic usage model...

[mulkieran]

@agrover, you're saying we should skip all optical media? Or all RO optical media?

[agrover]

Worth discussing but I currently think we could skip all optical media.

[agrover]

I think the cleanest way to handle this is to issue all block devices a SCSI INQUIRY command, which would then report 0x5 in the peripheral device type field for optical devices. Actually, since we just want block devices, we could filter out all devices with device type other than 0x0 (direct access block device).

There isn't yet a nice Rust library for issuing SCSI commands so we either need to wait until someone develops one or develop one ourselves. I think it's not a high enough priority for us right now to devote significant time to this.