>> On Wed, 8 Mar 2006 01:33:21 -0600, Roger Deschner <[EMAIL PROTECTED]> said:
> In looking at this implementation, I am left to wonder, "What's in > this for me?" Sounds like putting it up is quite a bit harder than > it is worth. I'm exasperated by the GUI efforts too, but I think that's a little too strong. There are lots of folks who want a straightforward graphical view of what their stuff is doing, and I don't see any reason to just refuse them. But since it comes from IBM, there's a natural sense of "Why can't I do <feature q> through the GUI?", so IBM tries to cover all of it. Unfortunately, "all of it" is hideously complex. Think of the range of exposed interface complexity for an everyday task: registering a node. At the minimal level, we've got something that tastes nearly zeroconf: the default signup policy domain. Just point your client at a server, and *ping* you're done. On the other hand, we've got the web interface, with it's encyclopedic --21-- options! In my docs, I say >>> On this screen, you should presume that no change is necessary to >>> any field unless it is mentioned herein. and I still get blank looks and a scared expression 3 times out of 4 from a new domain admin. Now, I know why they've got all the options there; it's probably generated by the same syntax engine that generates the docs, and that helps ensure that it's complete and accurate. Rock on. But that makes it specifically wrong for the class of users (new,timid) which was your reason for webifying the bloody thing in the first place. So if IBM wants to deploy a web interface for newbies and operations staff, have it be limited and relatively weak, and make no bones about it. Let the aftermarket folks accept the slings and arrows of outraged admins, and cherry-pick the ideas that come from them. - Allen S. Rout
