On 6/3/07, Chris Custine <firstname.lastname@example.org> wrote:
I have been thinking about a web based admin app for quite some time as well... I think maybe we are even talking about 2 different things here (basic internal admin app and larger enterprise admin app). I have even gotten to the point of thinking that the basic embedded Jetty app that you are already discussing should be part of the OSS project, but maybe a larger Enterprise app is a seperate thing altogether, almost like Studio (maybe Studio Web).
Yes this is certainly a possibility. Let's not close the door on this but I do not think such a
large application should be hosted directly on the ApacheDS server's Jetty service. Perhaps
The hooks could be placed on the ApacheDS instance via web services or some other
management interface like LDAP or JMX. Then this studio web app could be an application
deployed on a standalone web server.
I think we are obviously occupied with many other more important things at the moment, but I can tell you that my experience with client preference has been the opposite of yours. My larger clients would count a web based admin app as a postitive feature, and an installed GUI as a negative in product assesment. This is mainly due to the strict deployment and evaluation policies for desktop applications since neither of them allow direct install of software and require automated software push for inventory and license control, even for niche admin apps like this.
You're totally right Chris. Big companies lock down desktops but do they do it for those
select few power users like administrators that will be the ones using the studio application.
My reason for not thinking too highly of using a web based administration application
stems from this fact. Of the population of employees in the company a very small fraction
of power users (administrators) will be using this application. From my understanding
one of the main strengths of a web based application is in providing access to a large
population of users without having to deploy it on their desktops along with centralized
administration and maintenance. Here we're going to only have a small population of users
and hence I feel a web application might be overkill.
There might be another slightly larger population of non-administrator type users like
developers which may use Studio to develop schema or stored procedures. Most companies
now use Eclipse for development. Studio as an eclipse tool can be added to an existing
Eclipse installation as a set of plugins without requiring the need for such approvals to install
new applications. Meaning the plugin update process in eclipse will not require the developer
to request the installation of a new application on their workstation.
But I do agree some organizations will still insist on having a web based platform for this. This
is why I'm not abandoning the idea but for me it is merely a matter of prioritization. I think we
can get by with an Eclipse RCP application for a while. Having a web based Studio will be
a great thing to have but not required.
I think SUN is writing a OpenDS web application because they're stinkers when it comes to
using Eclipse. This is one of the reasons why they chose a web based administration console
since they X'd the option to use eclipse.
Also note that building a Web application verses a fat client is a bit more involved.
(1) server apps must run forever and leaks can add up whereas client apps are restarted
(2) lots more moving parts in a webapp
(3) webapp dev is less agile than fat client development
I think if we mature the RCP based Studio fat client rapidly through user input and solidify the
feature that are deemed the most useful then we have a great set of requirements already
in hand for building the web based studio application. Knocking it out then will be much easier
since the requirements are clear and all we need to do is apply some mechanics to whip it
Anyway, this is a complicated discussion, but at some point I would like to re-visit this when we can give it more time. I have a long list of features that I have been building in my head, so maybe at some point we can document some ideas and evaluate it from there?
Sounds good and I hope you don't think I am shooting down your idea. I do think it is a good one but it just comes down to prioritization, time to market (can't believe I just used this term on an OS mailing list :) ), and the impact that will result. I do want to do it though but the when and the how is what I am concerned with.