Michael Brown

The problem is that the "index" is not a well-defined concept. We can enumerate all filesystem handles (as we already do in order to support access by volume name), but...

> Is there anything in efi shell that makes that implementation more consistent? If not, random IDs might be better than no ID. Anyone looking for specific FS will have...

> What about using a random Hex four character identifier? > filesegmentFEED > filesegAC1E > filesBEEF How would that be used?

> file://filesegment[ABCD]/path/to/file > > if it’s addressable, then at least it could be searched. > A la your file://device0/... naming and the mention that "fs0:" is unstable I'm sorry, I...

> Here's an idea for generating a label: checking for the existence of a file on the device that would start with some prefix and use the filename as the...

To get any meaningful response, you'll need to say exactly what you modified in your "modified ipxe.efi".

> After further investigation. The file were it was erroring on was was `src/interface/efi/efi_pci.c` (reading error message). After a couple of permutations the commit the build fails on is [9b25f6e](https://github.com/ipxe/ipxe/commit/9b25f6e5cf517f426de80ede618398ef01e385f9)...

May plausibly be the same issue as #250

I can't see a clean way to implement this, sorry. The internal generic data transfer APIs don't really support the concept of a range request, except for the purposes of...

I've pushed commit https://github.com/ipxe/ipxe/commit/0f4cc4b5a based on the first commit in this PR, to add the `--uefi` option for `isohybrid`. In my tests, this produces a working bootable hybrid image even...