[openbeosstorage] More Progress
- From: "Ingo Weinhold" <bonefish@xxxxxxxxxxxxxxx>
- To: "Storage Kit" <openbeosstorage@xxxxxxxxxxxxx>
- Date: Mon, 20 Jan 2003 00:08:19 CET (+0100)
Hey guys,
finally I'm through with what I planned to do (respectively postponed
some bits ;-) and here's a little overview:
* FS flags, i.e. properties a FS add-on has (e.g. support for
initialization and the like): Not yet implemented. I have the feeling,
that there will show up more requirements of that kind, when we
thouroughly read the C++ API again.
* extended_partition_info::{device,mounted_at}: I have little hope that
get_nth_partition_info() will be able to fill them in. For `device' I
actually see no way at all. If we would also (or only, but that implies
opening/closing the device with each invocation) pass the path of the
raw device, things look *much* better. I documented the problem a bit
in the function in DiskScannerTest.cpp.
* The get_nth_*() disk_scanner module hooks now have a
{partition,session}_module_info** parameter for future kernelland
optimizations.
* As proposed by Axel, the FS add-ons' identify() hook are supplied a
callback for retrieving data from disk. There is a very simple cache
implementation, but it should be sufficient for the purpose. The two FS
module are adjusted accordingly.
* I added the hooks concerning partitioning to the partition module
interface. The disk_scanner module support is also there, as well as
the user functions. The only partition module doesn't implement it yet
though. For volume initialization there are only the unimplemented (but
documented :-) user functions. I played with thought to also add the
disk_scanner and FS module hooks, but didn't, since the future of the
FS modules is uncertain. I'm actually not sure how to proceed. When we
want to test the initialization (including parameter editing) properly,
we certainly want to have something we can test in userland. I don't
know, how well that could work with the FS shell.
* I also added the API for the userland GUI add-ons. I adjusted some
things (e.g. got rid of the common base class), but we will probably
change more, once we'll have a clearer idea of the remaining
requirements.
That's it for now. I will re-read the C++ Device API next, but not
today.
CU, Ingo
Other related posts:
- » [openbeosstorage] More Progress
