Thea Gab wrote:
> Hi Keith,
> Thanks for the comments. Responses are in-line below. We are also
> planning to move the wiki material to the external wiki and will send
> out a link to that as soon as it is available.
> Thea, Danny, Mark, Ryan
>
> On Wed, May 20, 2009 at 1:05 PM, Keith Mitchell
> <Keith.Mitchell at sun.com <mailto:Keith.Mitchell at sun.com>> wrote:
>
> Hi,
>
> I was planning on asking how this was going. I think the
> requirements look good so far. It sounds like you've finalized the
> decision to use the Ncurses set-up, programming in C. I added just
> a couple comments/questions in-line below.
>
> Please note, though, that the wiki link you posted is an internal
> link, and not available to everyone watching the caiman-discuss
> mailing list. In the future, just include a copy in the email. Thanks!
>
> -Keith
>
> *Problem Statement*: Provide installation media for
> OpenSolaris that delivers a purely text-based interface for users
>
>
> *Requirements:*
>
> *Primary:*
>
> 1. Installation media will run on x86 architecture
>
> 2. Same options/configuration as GUI installer
>
> 3. Ncurses interface
>
> 4. Provided as option on current CD
>
> /Non-Functional Requirements:/
>
> 1. Provide proper documentation
>
> 2. Ensure adequate testing is performed
>
> What defines "adequate" testing? In terms of your involvement with
> this project, what degree of testing will you be doing?
>
>
> We meant this requirement more as coordination with the QE's and test
> engineers, but we also plan to do our own testing of our code by
> trying out our install on different computers, using different
> options, etc. We will change this on the wiki to clarify.
A coordination effort will probably be best. I don't imagine you'll have
time to personally test a whole lot of different configurations (and if
you did have that time, you would probably be working on the secondary
objectives).
>
>
>
> 3. Make sure that ncurses interface is user-friendly
>
>
> *Secondary:*
>
> 1. Runs on SPARC architecture
>
> 2. Line-by-line interface
>
> 3. Network installation
>
>
> *Use Cases:*
>
> 1. User has SPARC machine and wishes to install OpenSolaris by
> inserting a CD and selecting configuration options through a
> console.
>
> 2. User has x86 machine and wishes to install OpenSolaris, but
> does not have enough memory to run GNOME. User inserts the
> text-based install CD.
>
> 3. User has x86 machine but graphics card is not supported by
> OpenSolaris, so they run the text-based installer in order to
> install OpenSolaris.
>
> 4. User starts install (situations 1-3) and the install runs
> to completion with no problems. At the end the user is given
> back control of the terminal from which they can explore the
> system or reboot.
>
> What are the reasons for supporting this use case? Or rather, for
> drawing attention to it as desired behavior vs. automatically
> rebooting, for example?
>
>
> We included this case just to indicate a normal flow of events after
> the installation starts. It was suggested to us that we give the user
> the option to cancel out of the automatic reboot, so they can look at
> the log files and other aspects of the system before the reboot
> occurs. We will reword this to be a little more clear.
Yeah, rewording to something like "At the end of the install, the user
is given the option to either reboot the system, or explore the
installation using the terminal" might be a better way of stating it.
>
>
>
> ALT 4: Error occurs during install, user is given log of
> install and error messages, user regains control of terminal
> for troubleshooting
>
>
>
> Thea Gab wrote:
>
> The interns from Colorado School of Mines have posted
> information from yesterday's meeting and an overview of the
> text-based installer project at
>
> http://installzone-wiki.central.sun.com/wiki/index.php/CSMInternResources.
>
> Also, after listening to all of the advice and ways to take
> this project from the meeting today we have determined a
> different approach would probably be better. Since we
> initially did not anticipate how to connect Dialog to the
> installer code, specifically liborchestrator, we feel like
> using this implementation will result in a much more
> roundabout method in order to achieve the goals we're seeking.
> Specifically:
>
> * Interaction with liborchestrator will require some
> intermediate steps between bash and C which will introduce a
> new level of complexity on top of the project
> * This additional complexity would also leave the system
> prone to breaking more easily than a single self contained C
> program.
> * We also believe that using this approach would make the
> text installer easier to implement for SPARC systems in the
> future, because Dialog would not need to be added into the
> repositories. * Further, we can reproduce Frank's mock ups
> to a closer degree.
> * Learning how exactly to tweak Dialog to obtain just the
> right type of interface could take just as long as programming
> up a near perfect representation in ncurses.
>
>
> We now think it would probably be better to just use a
> straight ncurses implementation of the text based installer.
>
> Thea, Danny, Ryan, Mark
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> caiman-discuss mailing list
> caiman-discuss at opensolaris.org
> <mailto:caiman-discuss at opensolaris.org>
> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>
>
>
