[gentoo-dev] What sort of infrastructure programs/features do we want?

[Kurt Lieber]

Folks --

Over the past few months, there have been a number of ideas tossed around
about having this or that feature available to the developers.  Off the top
of my head, I can recall:

* shared calendaring
* shared addressbooks
* web mail
* workflow management/project management
* wiki (internal and/or external)
* php support for developer ~/ directories

and the list goes on.

The thing is, most of these are being requested by one or two developers
and I have no idea if the group as a whole wants or needs them.  Do we want
a full-blown groupware system like phpGroupWare and/or eGroupWare?  Do we
want to implement the Horde framework?   Or are we happy with what we have
now?  There used to be a lot of wiki proponents out there, but the verve
seems to have died down a bit.  Do folks still want one?

We're finally getting to the point where we have some hardware to start
offering new features and capabilities.  Now we just need to decide what.

I have a feeling that this email is going to bring a deluge of responses
about "we should have this or that" with little, if any, consensus.  So,
I'm going to be a bit ruthless in this.  If you simply say, "we should have
<some feature>, kthanx" and leave it at that, your request is likely to
receive very little attention.  

If, on the other hand, you really believe we need some sort of capability
and/or feature, then do your research, find the top 2-3 apps that provide
that functionality, draw up the pros/cons in a summary of sorts and create
a GLEP[1] for it. The other developers can discuss/vote on it and we'll
take it from there.  I don't promise to implement everything that gets
voted up, simply because we need to ensure that all apps fit within our
overall architecture, are secure, aren't resource hogs, etc.  I do promise,
however, to give those requests a lot more consideration than the "kthanx"
requests. :)  In this case, GLEPs speak louder than words.

Some guidelines to consider related to our current infrastructure:

* We (the infrastructure team) are responsible for supporting whatever new
  features that we (Gentoo) decide to implement.  Thus, when it comes to
  deciding what does/doesn't get implemented on our servers, a very
  significant aspect of the overall decision making process is how easy
  things will be to maintain.  Thus, suggestions that contain phrases like
  "and then write a series of custom modules to hook them all together" are
  far less likely to be approved than ones that contain phrases like
  "simply untar it, edit two conf files and it runs from there".  

* One thing I consider very carefully when implementing new applications is
  how well-supported they are in the OSS community.  Apps which haven't
  been updated in over a year, have little activity on their mailing lists
  (or have no mailing lists) and do not have a strong community following
  are less likely to be approved.  This is a *general guideline*, mind you,
  and not a firm rule.

* Your participation in all of this is crucial.  I realize we don't all
  have the time to research options, write up proposals, etc.  Fine.  But
  at least take the time to comment and vote on the other proposals that
  get put forth.  I don't know what sort of infrastructure we want.  I know
  what I want, but you all need to tell me (using the process above) what
  you want.  Then, we can figure out what the best solution is for the
  entire team.

 --kurt

 [1] http://www.gentoo.org/proj/en/glep/