no flame intended on this side 
I agree with you and with everybody out there saying that the e-smith layer is “strange”… it’s, simply, unique…
when you know it, you can managfe your server without a web gui (there are situations where you can’t) without dealing with conf files, their syntax and so on
At the contrary the discussion is very helpful because many people here might have the same doubts and needed of explanations on why doing things in this way and not in another, being always open to suggestions on improving it.
I’ll put my 2c in.
I’ve been building ubuntu servers with packages from repositories, why, because the documentation points to the .conf files, mod the file, restart, done.
Or use, webmin, you guys know what that does.
Point being, I’ve posted enough here about this need, or issue, and most times it means create a file in the esmith environment… wtf?, huh, sounds like trying to do something in Zentyal… only not as bad, Zentyal is bad…
My problem, I don’t have time to sit and understand e-smith, and worse, there are no decent examples, documentation is sparse and all over.
So, if I need a simple mail, file server, especially one I’ll hand over to someone else… NS is good, if I need to do something out of the box, which is often, I have to go back to the basics I know, Ubuntu. It’s a matter of time.
I’d honestly love to understand NS better, but the info isn’t readily available, easily consumable, and I have work / family that prioritizes over learning a distro that’s serves a minor purpose.
So, I like .conf… esmith, not so much.
I think few developper have enough knowledge to made software package for CentOS, RHEL, Fedora, Suse but don’t invest time to the e-smith, Koozali SME or Nethserver because it’s time consuming, it’s not worth.
They prefer stay develop for the main stream.
Other could have the knowledge in CSS, HTML to improve the webgui but don’t for the same reason don’t want to invest in the e-smith layer.
This e-smith layer is too specific, too restrictive for awake interest for the majority of developer in my point of view.
I’m now curious to know how ClearOS and Yunohost deal with software package.
Another way to look how to made things easier is how to deal, how to generate the .conf files in the /etc directory.
Is someone of the dev team already look how Cockpit or Webmin deal with such files?
my 2c: learning e-smith layer is just a matter of study… and you DON’T need weeks, neither days.
so, here, it seems to me that the majior probloem is, don’t get me wrong, lazyness…
AFAIK Cockpit does not deal with config files directly. It provides the abstraction to invoke some API on various protocols (HTTP, dbus, plain exec()…). In other words, even if NethServer UI is Cockpit, we’ll need e-smith or its alternative for the business logic layer!
Webmin? I don’t know it…
Perhaps people coming from Zentyal can tell us about webmin…
But it’s seem that webmin don’t need an e-smith layer to deal with the system
Heh, laziness, yeah, that’s me. smh.
I see laziness too, whenever I look at Nethserver documentation.
Is Python can do the job?
Use Python for…
Web Programming: Django, Pyramid, Bottle, Tornado, Flask, web2py
GUI Development: wxPython, tkInter, PyGtk, PyGObject, PyQt
System Administration: Ansible, Salt, OpenStack
Guys, I’m against such sentences and behaviour, please no personal attacks here as our guidelines already said. Don’t post anything that a reasonable person would consider offensive.
We’re discussing about e-smith with good and polite contributions from everyone so far, please let’s keep going in this way. 
I think that the Developer manual has been written by some developer.
While I’m not the author, I know developers tend to be terse. We probably need a non-developer to adjust the manual.
I also think that the upcoming tutorial(s) will be of great help.
I really love Python and I’d like more Python in NethServer 
But we’re developers facing an already existing system. We can’t pick our favourite language and start from scratch with it.
We must take care of a colourful codebase, speaking Perl, PHP, Bash, JavaScript, C, Objective-C…
I use Webmin on two RPi2’s and one RPi B and quite frankly, not sure how the inner workings work. They are used as DNS/DHCP servers; I found them to be “better” in the fact of setting up those types of configuations, especially bind9.
But why start from strach?
There’s the existing Salt
sorry but, no personal attack intended…
with “laziness” I mean the difficulties to defeat approaching a new and unknown task.
I happily met Ansible to develop and deploy packages.nethserver.org
I was asking myself if its templating system could replace e-smith one, and came to the conclusion they’ve different purposes. How implement a key feature like custom templates? What about modularity?
Indeed managing an infrastructure as code is far from administering a single server. Do you agree with me?
this is not true…
since the e-smith layer is used by only 2 distro, you can find documentation here and there (I already linked the wiki page 2 times in this 3ad, so I won’t link it anymore)
Jim, it seems that you are missing the point…
even if NS is almost all new, it is based on something that exists and (simply) works since 15 years now…
you obviously can’t image how much code is behind the surface you see.
moreover, re-basing all the management system (forget about the web UI) to a new infrastructure means re-writing all the packages, and falling back to a pre-alfa stage…
the error that many of “us” are doing here is to thin in the UI perspective… forget about it… when you have the underlying logic working in the right way, the UI can be made with every kind of framework/language
in the end, if I was able to learn not only how the system works but how to create a package (mine was smeserver-password, imported now here as netheserver-password) I’m quite sure that everyboy can do it…
Of course my opinion will differ from yours, but if I can’t find the answers I seek to what I feel are fundamental usage cases…
I appreciate that, and have added your links to my list, and when I can get caught up enough to justify spending the time to study and understand the NS template system I will be pleased to do so, of course that will be after I solve the email filtering and a couple of other NS issues, lol! Of course, this priortization is defined by how useful NS 6.* is to me, becuase I’m much more interested in 7.
I just got an email from ClearOS for their v7 RC with Samba 4…
Where do I put my free time???
I do understand that in the end it’s whether or not NS serves my needs, will continue to do so, and if there’s long term stability and value to invest myself in it, so it’s up to me to take it or leave it as is.
Yes, in this case it’s changing a 6 for an half-dozen 
When I initiate this thread it’s only a tentative to do a brainstorming about the weakness, or perhaps the apparently weakness of Nethserver.
t’s not the base CentOS.
It’s not the gui wich is a good idea
It’s the lack of developer due to the complexity of the e-smith layer.
Perhaps tutorials could/would resolve…
But don’t forget why SME server was in dificulty during the time… it’s the same problem: the lack of developers. This affirmation is not mine:
http://www.clearcenter.com/News-Articles/comparing-sme7-and-clearos.html?Itemid=456
I don’t want to denigrate the e-smith layer but we must analyse the forces and the weaknesses of such system.
Perhap the way ClearOS deal with webconfig is easier for developer