Re: [Setup-tool-hackers] Helix Setup Tools and Open Connect.

On Mon, Dec 11, 2000 at 02:56:01PM -0500, Miguel de Icaza wrote:

> I have been reading the mails about Open Connect and Helix Setup
> Tools, and there is something that I still do not think has been
> addressed correctly to make the right decisions.

> This manifests as the discussion in `we share the backend, we have
> different GUIs' debate.  I am not sure why we need two user
> interfaces.  At all.

> Arturo said `They need to handle 10,000 users'.  Which in the context
> of administering users, it lead me to think `If they have 10,000
> users, they most likely will be using NIS and some custom setup'.

Well, since the OpenConnect guys just hacked in the beginnings of NIS support
in the backend, that's exactly the sort of stuff they want to be doing.

>> From a user interface perspective, the questions are:

> 	* What do you need in your interface that is different enough
>           that it will require a different front end?

> 	* And if you need a new user interface, what is the rationale
>           behind the new user interface?

There are a quite a few things that are clearly very different between
managing a handful of users on a desktop box, and managing a couple of orders
of magnitude more users on a large cluster. These can be easily surmised just
from analyzing the problem, and shouldn't require usability testing to
identify (usability testing is more suitable for evaluating the quality of
the proposed solution, when a prototype interface exists).

The main differences seem to be:

 * Different mechanism for selecting the user to edit. On a desktop system,
   browsing a list of all users is quite acceptable. On an installation with
   10k users, it clearly isn't. Searching the user list, possibly also
   arranging users by groups (in a hierarchy) is a must.

 * The ability to change settings for several users at the same time.
   Selecting users matching a pattern, for instance, then changing the
   settings of all of them. You don't want to change the login shell of 10k
   users individually.

 * Support for templates and so on when creating new users.

It's not necessarily clear to me that the frontends need to be two completely
different ones, sharing nothing but the backend, but it seems very clear that
there are significant differences between the two interfaces and modes of
working. If the frontends aren't separate, they'll have to be modal. Modality
isn't generally considered a good thing in UI, and should be used sparingly.
If we end up with only a small part of the frontend functionality shared, it
would probably be better to use two frontends. A suggested feature spec
should be an excellent way to proceed, "usability testing with sysadmins"
seems vast overkill.

Joakim Ziegler - Helix Code web monkey - - Radagast@IRC
      FIX sysop - free software coder - FIDEL & Conglomerate developer

Setup-tool-hackers maillist  -

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]