"mini roadmap" for userland for 3.7-dev branch

[John Marino]

On 11/22/2013 03:30, Justin Sherrill wrote:
> On Thu, Nov 21, 2013 at 3:19 PM, John Marino <dragonflybsd at marino.st
> <mailto:dragonflybsd at marino.st>> wrote:
>     Right now, "maintenance", is about the best we can hope for unless we
>     either get new blood or "vets" start producing something substantial.
>     And new blood has to start somewhere, so ...
> Honestly, the project now is as busy as it's ever been.  There's less
> list traffic, but that's because IRC has been very active. 

A large percentage of that IRC traffic are from folks that don't
contribute code.  Speaking for myself of course, I judge activity first
on the number of active committers, then on the frequency and content of
commits.  Intuitively I'd say we rank about 10% compared to NetBSD which
is about 50% of FreeBSD, and that's just kernel + userland, I'm not
including ports/pkgsrc only commits.

So by those estimates we're getting by with 5% of the manpower of
FreeBSD, which also means that any resignation will have a huge impact.
 The one thing that DragonFly is terrible about is recruiting and that
should change.

The project may be busy as usual, but it's the usual suspects carrying
the load which leads to burnout.  This is not a healthy situation.


>     I'm the first to cheer on "DragonFly Firsts" but I just don't know who
>     is going to spearhead them right now.
> I don't know... maybe suggesting ideas in a publicly posted list, or
> 'roadmap'? 
> 
> :)
> 
> Anyway, I had better supply crazy ideas if I'm going to ask for them:
> (note that these are not focused around just code)
> 
> - I'd like to see some management utilities around Hammer.  When people
> talk about liking ZFS, they are often referring to the DWIMmy tools for
> using it; Hammer has a lot of features that could be shown that way.  A
> bit harder with Hammer 2 on the way, but not a bad goal.

I think people see HAMMER as a dead end, something that is already in
maintenance mode because HAMMER2 is the future.  I agree with you,
except that all these things should be based on H2.

> - DragonNAS. 

yes

> - Freshports adapted to dports.

Freshports is overkill, all we need is a visual catalog describing the
ports and its relationships, home page.  The actual commit history is
not pertinent.  I believe Joris already started on something like this.

Also this is project infrastructure (not a DragonFly OS or first)

> 
> - mdocml used to replace our online man pages.

project infrastructure

> 
> - A DragonFly image with preinstalled packages - similar to the GUI
> image, but set up with monitoring and server utilities instead.

I don't know about this.
1. packages will get obsolete quickly
2. somebody will always complain about package selection
3. It's a lot of work

Maybe some kind of basic GUI package with "profiles" which is just a
list of packages that you can install the latest versions of the
packages on the profile list.  You could have many different profiles
(web and file server, games, development, etc)

> There probably could be crazier, but they are useful.

The issue is manpower (as always) not a lack of ideas!

John