Re: [webmin-l] Ramblings after reviewing the pole

[Barry <we...@i1...>]

John Hinton wrote:

> Jamie Cameron wrote:
>
>> Hmm .. it sounds like I need to add some popup help to that module. To
>> be honest, I have trouble thinking from the perspective of a
>> non-Unix-expert :)
>>
> I just had to leave that comment there. ;) Yeah, you only have to deal 
> with us dumb sysadmins that can't find what's right in front of them 
> in the Webmin interface! We on the other hand, have to deal with the 
> public who for the most part doesn't know the difference between POP, 
> alias, incoming mail server and outgoing mail server.. and thinks 
> 'email client' is a command not a program. "OK, I'll email that client 
> of mine right away!"



My 2 cents:

O have mentioned here before and will mention again that I was one f the 
early employees/managers at Cobalt Networks.

We wrestled with the same issues every day back in the early dot com era 
when we needed web based software to support our Linux server appliances.

I am not saying that I would either architect that systems the same way 
today, nor would I make the UI exactly the same, so let's keep that in mind.

But from a marketing perspective, like most wildly successful startups, 
had a laser focus on who our users were and what the tasks they were 
trying to accomplish were.

At the top line, all our products were designed so they could be fully 
functioning 5 minutes out of the box. In most cases, as I recall, the 
only typing that was required was to enter an Admin password. You had to 
set an IP address from buttons on the box, and that was pretty much it.

Beyond that, the focus was on specific tasks, and tying them together in 
some coherent interface.

We identified several distinct groups, roughly as I recall:

- end user home networking (the Cobalt Qube product line).
- the ISP. (Cobalt Raq) I think Cobalt payed a large role in the growth 
of Mom and Pop ISPs, and as a result played a large role in enabling the 
low price monthly hosting model. Think back to 1997, and 1998 and lots 
of people with minimal technical skills were able to deploy Raqs out of 
the box and then provision hosted web sites with all the features of the 
day: mail, web, ftp, dns, and so on...
- the sysadmin: As we had customers with more and more Raqs installed 
(1000s?) there grew a class of more sophisticated admins, and they 
needed to be able to manage their boxes from a single interface.


there were other product lines, and many others that were identified, 
test marketed, and even prototyped but never saw the light of day.

My point to Paul and Jamie is to identify users, the tasks they need to 
do most often, and to design products around those needs, while keeping 
the flexibility. Ultimately, this is more of a marketing task than a 
technical task.

One way to gather some data, aside from the polls, is to build a version 
of webmin that is instrumented somehow, and reports back to you (at a 
certain level of abstraction) what real people are doing. Call this a 
beta program if you will, but let people opt in after explaining to them 
the way the data will be collected and used (and not used :). Give them 
something in return and I am sure you will get some good info.

>
> Actually though, that is the bottom line. No pun intended.
>
> And, in all fairness, I would really like very much to hear the 
> reactions to my original post by other sysadmins using Webmin. I am in 
> a unique niche myself which may be enough unlike what others are doing 
> so that there is no justification to spend time on my wishes. Our 
> clients are hosting clients, predominantly in the bed and breakfast 
> industy. This means they are business owners, in a business that 
> requires a high degree of specialty knowledge and many simply don't 
> have the time to become proficient on computers. At the same time, 
> bookings run at about 80% directly from the internet with some 
> reporting over 80% from online booking engines. Email is as important 
> as water and electricity to these folks and equates to the person 
> behind the cash register at a store. Many of them can make french 
> toast to die for, or make the army turn green with envy when they 
> bounce a dime off a properly made bed sheet.. but computers? So, yeah, 
> I have to step back to view my ideas about 'how stuff should be', as 
> I'm sure my opinions are somewhat biased by my clients as well. So, 
> really, I do hope others will chime in with yes or no, that would help 
> or not.


That is precisely the type of customer we had on our target list at 
Cobalt. We wanted the boxes to be as functional (and as good looking) as 
a routine kitchen appliance. I wish Webmin had the same kind of 
attitude...I notice Paul mentioned a great book about UI, I recommend to 
him the work on "personas" that started with Alan Cooper in his books 
and is now more pervasive in the UI area ...

>
> Again, with regards to Webmin, yeah, some additional help pop ups 
> would be great, but I really don't see any of my users knowing what 
> the heck to do within the Apache module. Well, I actually might have 
> one, but he was a sysadmin in the past and got burned out. :) Hmm... 
> maybe even two now that I think more.


Frankly, I agree with John here, and have kicked around the idea of 
incorporating the Cobalt approach at the user with the underlying 
functionality of Webmin. Any potential sponsors out there?

>
> I've looked at just about every module/package available and don't 
> find anything that fully fits, or really even gets pretty close. When 
> I went back in to look through Usermin again, I realized it was pretty 
> close. Heck I even discovered that https://users_domain_name.com:20000 
> got them in! I never even thought to even try that until today. Now 
> that's pretty neat! I don't have to look up which server they're on 
> and give them one of my convoluted server names as the domain to 
> visit. Easy for them to remember... And it is much easier to tune down 
> Usermin vs. Webmin/Virtualmin. What I have left running in Usermin for 
> my users at this point is
>
> Change Language (might get used by a few of our clients and it's 
> always good to be PC)
> Mail Forwarding and Replies
> Read Mail (great for clearing out that email that causes "My Outlook 
> is stuck!" syndrone. Although a view oldest mail first switch would be 
> nice. Yeah, the go to page ### works, but then an email comes in and 
> interrupts the process and you're back on page one unless you figure 
> out and then remember to use your back button)
> Scheduled Emails (nothing like 'Elvis' sending you a message each 
> month on the day sales tax is due.. or better yet, on your 
> anniversary! This should bring back some memories for some of you old 
> 'nixers!)
> SpamAssassin Mail Filter (probably not a good idea.. but..... I'm 
> trying it to find out.. another of those with no help and over the 
> heads of most end users)
> MySQL Database (well... who knows.. but for the ones that have it.. at 
> least they can look at what they've created)
> Disk Quotas (OH yeah.. hope they look at this... often!)
> System Documentation (betcha none of them ever read any of this but 
> you can't blame me for trying)
>
> What is missing...
> Add/edit/delete a POP account
> Add/edit/delete an Alias
>
> Now I know those are going to be tough without some tie back into 
> Webmin/Virtualmin to get them into their virtual domain.. get the 
> account tied to the domain name and still exist within the limits set 
> in Virtualmin, like number of POP accounts and so forth. But what the 
> heck.. I can wish!!! It's almost Christmas isn't it?
>
> And....
> And the dumbed down interface that 'anyone can understand' or at least 
> almost anyone.
> The hardest task is to keep it simple... or make it simple. Many times 
> I've started a simple easy to navigate website and it grew.. and then 
> it became complex.. and it was no problem at all to navigate... but 
> only for those that built the darned thing.. Somewhere along the line 
> the new visitor's view became confusing, they were lost and left.. 
> This scenario seems to be the natural order.. and one has to interrupt 
> the natural order to get it back to a simple and easy to follow flow.


Again, to go back to the Cobalt example, internally we had people with 
conflicting responsibilities that had to compromise on the final 
product. But with the laser focus on user *tasks* and a keen eye for 
creeping features and creeping UI, we were not engineer driven at the 
feature level, even though we had an army of Jamies. The engineers were 
fully involved, don't get me wrong, but they all bought into the notion 
of the appliance itself - that was built into the corporate culture.

>
> So, come on you others on the list... What's your thoughts on what 
> I've said? Is there reasonable need for these things or am I just off 
> in left field here? I can take it.. I am left handed even. Am I the 
> only one out there that would like to have these tools?


No, not in left field, and I am right handed. I want them, I *know* 
others want and need them, because I know that Cobalt did not finish its 
job, and now the net is just that much bigger....

>
> Thanks to both Paul and Jamie for your open minded attitudes.


Likewise. I use Webmin all the time on many servers, and that is great, 
but I also look at it every single time with the above thoughts and 
background in mind, and I find lots of room for enhancements in the 
approach and experience...

Best,

Barry Caplan