[maemo-developers] [maemo-developers] Core-iSCSI/Nokia 770

From: Nicholas A. Bellinger nab at kernel.org
Date: Thu Mar 30 00:33:10 EEST 2006
On Wed, 2006-03-29 at 19:28 +0100, Neil Jerram wrote:
> "Nicholas A. Bellinger" <nab at kernel.org> writes:
> 
> > I am very interested in hearing from voluenteers from the Maemo
> > community who would be interested in developing a UI that makes script
> > calls in order to control the core-iscsi (and eventually open-iscsi)
> > stacks.
> 
> When you say "script calls", do you mean shell command line calls -
> and hence things that can be done from within a scripting language
> using system(2) ?
> 

This is correct.  There are currently two methods that Core-iSCSI can be
controlled with:

1) via /etc/rc.d/init.d/initiator (ie: the script)
2) via /sbin/initiator-ctl (ie: the ioctl)

The script reads in config data and then makes calls down
to /sbin/initiator-ctl.  The idea here is to hide the complexity of the
day-to-day iSCSI admin operations inside of the script and allow for
advanced users to call (or write additional programs/scripts
around) /sbin/initiator-ctl.

On a related note, one of the projects for the core-iscsi-tools package
is to control Open/iSCSI via the same set of scripts and configuration
files.  The idea here is to allow Open/iSCSI and Core/iSCSI stacks to be
interchangeable with a minimal amount of complexity exposed to the user.
What this boils down to is that UI should use #1 above in order to keep
future portability between stacks.

Note that a full explanation of the script in question can be located in
Section 3 of the HOWTO:

http://www.kernel.org/pub/linux/utils/storage/iscsi/HOWTO

> If so, I'd be interested in doing this as a test case for my Guile
> packages, which wrap the Gtk and Hildon widget set on the 770.
> 
> You may have been thinking of doing the interface in C, and not want
> to introduce further dependencies, and I can understand if that is the
> case.  It could still be interesting to use my bindings to prototype
> and demonstrate the interface, and then convert it to C later.
> 

This sounds great!  I am not too stuck on the language or libraries that
would be used.  I will however mention that this UI would definately be
used in the field on both existing and future devices, so keeping it
portable between platforms will definately be a requirement.

Thanks for your interest! :-)

-- 
Nicholas A. Bellinger <nab at kernel.org>


More information about the maemo-developers mailing list