Zentyal Desktop Distro

[stuartiannaylor]

If a Zentyal desktop was to arrive what would you require?

My opinion is that M$ rule the desktop at the moment and it makes sense to differ and offer a lightweight desktop to gain a niche in thin clients and low spec hardware.

Lubuntu would be my choice with lxde providing enough of a GUI whilst remaining capable of zero to fat client.

[ichat]

i chose ubuntu lts while im not sure if lxde and specially pcFM and lxsession  provide enought features to support single signon as well as other features that will make it into zentyal 3.. but im sure that we want a modular base, thats build on stable and long term supported software.  in my vieuw we should try and support at least gnome 3 (either with or without unity), since thats the ubuntu desktop default.   and one lightweight alternative desktop either lxde or xfce based on system requirements and zentyal compatibillity.

in any case  all supported desktops and software should be easily installed  maintained and supported by both zentyal partners staf and community members. and we should always help and support any effort to add aditional support  whether that would be as complex as a fullblown desktop like kde (kubuntu) or just a diferent email client.   

at some point we should probably find a way to list supported software, packages and configurations.   an example could be taken from: wineHQ.

[stuartiannaylor]

I am not sure about the current state of linux desktop. Gnome3, unity and kde all have criticisms at the moment. I think they are in for some possible forks and might be static for a while.

My thought is that linux desktop doesn't compete with M$ so why even try. Concentrate on a lightweight desktop that provides more ROI in hardware, maintenance and administration. Stay away from shiny features and concentrate on a core.

Anyway anyone with some alternatives to those already in the poll?

Stuart

[robb]

As far as I'm concerned the ZD distro must meet some primairy demands:
- stable
- intuitive
- long term support (at least 2 years)
- as lightweight as possible
- centrally manageable
-

Then there are some conveniences that make life easier:
- Zentyal is Ubuntu LTS based. It would be convenient if the Desktop version has the same base.

Some considerations: There a TONS of Debian and Ubuntu derivates. Some are focused on being a 'light' desktop, some on higest possible user experience. I personally would favor to focus on user experience, but first priority should be stability and upstream support.

Therefor I think I would come out on Ubuntu LTS. The question remains if the Desktop manager should be Unity, Gnome or one of the many other flavors. Since Ubuntu comes with Unity by default, I think that is a strong point to stick with that. Don't forget that all the services that Zentyal Server provide should be supported out of the box on Zentyal Desktop.
An option _could_  be xubuntu, but I'm not completely sure about supporting all those services with xfce. Another option _could_ be Bodhilinux That distro is a full ubuntu derivate and lightweight, but it might lack the solid upstream support we want.

I am convinced that end users don't 'want' Microsoft XP/7 or MS Office. They neither 'want' Open/LibreOffice. They want an OS that runs their pc and they want an Office suite that has enough features to do their jobs. Bottomline: they want functionality. And THAT is what we should focus on when developping the Zentyal Desktop. And when we manage to do that, Zentyal Desktop WILL be competitive against a Microsoft desktop at a fraction of the costs.

[stuartiannaylor]

I have been playing with Lubuntu and it is quick and much less resource hungry than many 3d window managers. I really do suggest a look see.

From a 30 minute play of Lubuntu it seems feature rich and perfect for the SMB environment as an alternative to Windows.

Saying that I can't see why any mainstream ubuntu derivative shouldn't be capable of a Zentyal Desktop.
I mention Lubuntu as I can see a market for low cost "Thinner" clients as it fits the current economic mood.
Also I have mentioned LTSP before and this is a welcome roadmap 3.0 entry as when it comes to centralised client management and rollout... Huge benefits and I don't think there are any arguments about that.

"The Zentyal Desktop" and LTSP have much common ground and I think it would be prudent to keep LTSP in mind whilst toying with the idea of a Zentyal Desktop.
My own thoughts go along the lines that a light weight bare minimum (Zero Application) distro should be the basis of a Zentyal Desktop.

Then we have method of install from disk to PXE https://help.ubuntu.com/community/Installation/LocalNet#From_a_local_server and Terminal Services. This needs discussion just to see where we stand?

The Zentyal Desktop to myself isn't the actual client distro but the methods of control and distribution. I would like to know if anybody has any ideas on this? I have seen solutions forwarded such as Puppy and I am unsure if they are overkill.
Maybe something simple as they below.
If we created a Distro that would use a domain controller to provide a local repository as the distro's singular software source.
Had a Webadmin GUI entry to allow custom scripts based on user/group to push configuration settings and populate a local repository with .debs.
Then wouldn't this cater for the 80/20 rule of all our needs?

Stuart

PS Whilst sat here contemplating could not a share present a set of "Install scripts" through standard user/group security. Maybe something that simple?

[ichat]

did you try setting up intergrated smb-shares, nfs shares, and  kerboros,   all these nice little desktop things look a lot like single user, simple internet connection no network-filesystem  workstations.   those may be usefull for a simple desktop at home where you use a usb drive for remote storage but not in the office. ...

its not all about just being small,  you dont really want to mount any shares (from zentyal) via  fstab,  its not verry secure, and entirely not 'dynamic' ...   

there are developers out there that do understand this better than i do, and can probably explain in more detail,   

[stuartiannaylor]

I am convinced that end users don't 'want' Microsoft XP/7 or MS Office. They neither 'want' Open/LibreOffice. They want an OS that runs their pc and they want an Office suite that has enough features to do their jobs. Bottomline: they want functionality. And THAT is what we should focus on when developping the Zentyal Desktop. And when we manage to do that, Zentyal Desktop WILL be competitive against a Microsoft desktop at a fraction of the costs.

I am just not sure that developing "Yet Another Distro" is the way to go. I am more in favour of the "Zentyal Desktop" being a means to control what is a client.
Zentyal includes a VPN download package suited for each user, maybe a similar mechanism could be employed per object?

Stuart

[ichat]

i would say that we need a basic  desktop interface. to be able to support users.

it should contain at least:
> linux core (kernel drivers filesystems etc - managed and supported by ubuntu and us). 
> a desktop envirmoment, (or the choice between a few).  with a window manager, dekstop manager session mannerger grafic display manager  and a  configuration tool to setup user preference. and a lockdown system to enforce other settings  that where setup by the network administrator...

these should be installed from any usable source  netboot or cdrom  or whatever you choose.

im im not sure if you can already call a basic desktop installation without any apps a distro,  or you could merly call it an base install  like you would have done using the 'ubuntu alternate install  pressing f5'   with the small exemptiont hat we include a grafical desktop  an smb and nfs client   a proxy client (if needer) and a few more of those features.

the rest (installing firefox chromium or even epiphany,  libreoffice or gnumeric,  thunderbird or gnome evelution or even just a prism to the zarafa webapp,  thats not really interesting to  enforce upon the users.

we could at least for a first development release,  push a group of appications so that at least all testing is done using the same versions and applications.   but once it gets stable we should alow people  create there own desktops from a large repository with supported (and preconfigured) applications.    presented by an installer (wizzard) or a post install script?   thats up to the user...

[stuartiannaylor]

i would say that we need a basic  desktop interface. to be able to support users.

it should contain at least:
> linux core (kernel drivers filesystems etc - managed and supported by ubuntu and us). 
> a desktop envirmoment, (or the choice between a few).  with a window manager, dekstop manager session mannerger grafic display manager  and a  configuration tool to setup user preference. and a lockdown system to enforce other settings  that where setup by the network administrator...

  Not that I am disagreeing but I fail to see that we haven't already got this with any Ubuntu derivative?

Also if we are aiming at the SMB market then applications are more likely to be dictated rather than the choice of the user?

[robb]

I am convinced that end users don't 'want' Microsoft XP/7 or MS Office. They neither 'want' Open/LibreOffice. They want an OS that runs their pc and they want an Office suite that has enough features to do their jobs. Bottomline: they want functionality. And THAT is what we should focus on when developping the Zentyal Desktop. And when we manage to do that, Zentyal Desktop WILL be competitive against a Microsoft desktop at a fraction of the costs.

I am just not sure that developing "Yet Another Distro" is the way to go. I am more in favour of the "Zentyal Desktop" being a means to control what is a client.
Zentyal includes a VPN download package suited for each user, maybe a similar mechanism could be employed per object?

Stuart

Stuart, I am not suggesting a whole new distro should be developed. I think I even suggested the opposit: use a stable, preferably lightweight, with decent upstream development  distribution and adapt it to our needs.

Maybe the suggestion ichat made by profiding the option to choose a fullblown unity/gnome/kde desktop is possible, but also, to support older hardware, make xfce or lxde available. Ofcource provided all services offered by Zentyal server can be supported. We should keep that in mind, the ZD is there to make use of Zentyal Server in the easiest way possible, with an as high as possible user experience.

[stuartiannaylor]

Sorry as to my confusion as many seem to mention to ZD and I am with you, there is little point in redeveloping a distribution. We have had ZD mentioned many times but the idea of implementation seems a little hazy.

The user corner seems a logical place for a configuration and some discussion is needed.

As far as changing the samba client workgroup = %domain_name%"

email aliases & connection details

network shares

Jabber

Guess asterisk should be included.

What else are we talking about ?

https://help.ubuntu.com/community/SingleSignOn ?

http://www.ahinc.com/linux101/loginout.htm ?

[ichat]

automatic mounting of  /home  in userspace rather than fstab... ?
managing apt pushing updates and stuf.
controlling the allowed settings apps and preferences.
managing network-deamon.  setting up cubsclient for networked printers (per user). 

there are lots of other settings that we need to support and control..   and these are only just the 'verry basic desktop settings .. im nog talking yet about shared keyring support (so that your stored passwords for remote services are avail across all desktop logins (per user). or  per application settings..


bringing the user to a fully network intergrated desktop is a lot mor than just controlling who gets access to it.   

[stuartiannaylor]

Just to get down to it and clear my fuzzy head 

1...automatic mounting of  /home  in userspace rather than fstab... ?
Why? for what purpose? Honest question as I am unaware why this is prefered?
2...controlling the allowed settings apps and preferences
Can you expand on that?
3...managing network-deamon.  setting up cubsclient for networked printers (per user).
After DHCP what is required?
4...controlling the allowed settings apps and preferences.
Can you expand on that?

Sorry if the questions sound obtuse as the ZD has never been discussed down to specifics.

5...managing apt pushing updates and stuf

If we could create various repositories on a Zentyal server and have the clients point to that as a singular software source then much of the above is accomplished.
Each repository could represent a group of network objects or users.
From a base layer distro you build up the client from the centralised repository.
The sudoers accounts needs to be controlled and you can stop any changes this way.
DHCP options could pass the required variables.
login scripts could provide apt-get-install...upgrade and if synaptics and the software center have no GUI or root access then this is controlled.


Sorry for being dim, but the Zentyal Desktop isn't setup at the client it should be a module on the server?

Stuart 

[stuartiannaylor]

This is my wish list in terms of a Zentyal Desktop.

Basically its a server based module that has various repositories representing network group objects.
Software lists are created representing user groups.

The actions it would be provide:-

PXE automated install of "application free" desktop via boot menu choices
Custom DHCP to provide a custom software source via apt.
Apply a software list to a apt-get-install...upgrade script @ login based user/groups
Control manual or periodic updates of the centralised repositories.

Why is a good question but from a blank client the installation and configuration would be purely to connect to the network.
From the PXE boot menu choose a repository and let DHCP provide the rest.
Create the necessary network object and apply to group.
Job done.

[ichat]

1: because in userspace when you log of back to the  login screen another user can login without having the same network shares pre mounted this is for safely as you dont whant a computer having access to secure information for example...    you could use an smb browser instead but intergrating it into the desktop (like most users are acustomed to would be prefered in my support experience.

2 i was refering to controll the remote installation of updates (rather than having ubuntu auto-update you may want to instead managed this from a central place you may even want to push the newest version of  APP x to all the people in accounting  but not to the users in  development  (for example because they might brake some compatible device (printerpres, or some machinery.   (or any other reason for that matter.

3 dhcp is not going to support per user / group  dicrimination ...  and specially not setting the default printer  for you...  so when a sysadmin installs a new printer in  an office room he might wants to update the user's profile in that office to include it...      - using these kind of settingautomation would involve multiple vlan taggs  witch would be come quite costly and complicating..

im not saying that all features will be like this,  but we should desing ore 'basic image in a way that allows easy adding of any of these services.  automatic discovery would only work with one alternative ... so what happens if you have 2...