Re: Please stop sending Github messages to the ML

[Leslie Hartten (NOAA Affiliate)]

Jeff and Dave, thanks for your recent contributions to this thread.
I will chime in as "one of those people who would not want to deal with 
the additional complexities of Github".

I'm a scientist who in the last few years has needed to create data sets 
for broad release, and wanted them to be CF-compliant.  That meant 
learning more about CF naming conventions, and how they were applied, 
and how they were expanded.  That led me here, where I mostly lurk but 
do learn things about how CF naming conventions are developed and are 
applied.

GitHub is "a development platform .... [where] you can host and review 
code, manage projects, and build software alongside 31 million 
developers." (source: https://github.com/)  GitHub is "a web-based 
hosting service for version control using Git. It is mostly used for 
computer code. It offers all of the distributed version control and 
source code management (SCM) functionality of Git as well as adding its 
own features. It provides access control and several collaboration 
features such as bug tracking, feature requests, task management, and 
wikis for every project." (source: https://en.wikipedia.org/wiki/GitHub)

I don't want to do the things GitHub is built for; I just want to listen 
in to the conversation about CF naming conventions and occasionally 
chime in.  There are probably many others like me, people for whom 
understanding and applying CF naming conventions is one small part of 
our job, occasionally.  Maybe David's steps 1-4 are as simple as he 
makes them sound (whats a CF repository?  how does one find them?  how 
do they differ?).  But all the other things he lists are not things that 
intrigue me.



On 11/7/18 3:49 PM, David Hoese wrote:
1. You create an account.
2. You go to the CF repositories you want to keep updated on and click 
on "Watch" in the top right corner.
3. You receive emails.
4. You either reply to those emails (github will automatically write 
the comment on the site for you) or you click on the link in the email 
and go to github and type your message their directly.

Some basic pros:
 - With github, you can link to other issues and pull requests by 
using a # symbol followed by the number of the issue/PR.
 - You can mention other users and anyone who doesn't explicitly say 
they don't want to get notifications, will get an email/notification 
that they were mentioned (ex. @djhoese).
 - You can use github markdown syntax to add URLs, format code/literal 
blocks, shorten chunks of text/code (<detail> tag), add images, create 
sections headers, bulleted lists, etc.

Dave

On 11/7/18 4:44 PM, Painter, Jeff wrote:
Someone (I forget who) told me that many of the people discussing 
standard_names would not want to deal with the additional 
complexities of Github.

*From: *Brian Eaton <ea...@ucar.edu>
*Date: *Wednesday, November 7, 2018 at 2:41 PM
*To: *"Painter, Jeff" <paint...@llnl.gov>
*Cc: *"e.r.g.quaegheb...@tudelft.nl" <e.r.g.quaegheb...@tudelft.nl>, 
"cf-metadata@listserv.llnl.gov" <cf-metadata@listserv.llnl.gov>
*Subject: *Re: Please stop sending Github messages to the ML

Erik,

 > I don't have a view of the whole setup, so I may be missing some 
aspects
 > that complicate this. But even then, it may be worth the effort.

The piece you're missing is that I'm trying to get Mailman support 
off of
my plate; not make it a more integral part of the system.  The CF
conventions activities are supported at LLNL, not at NCAR.  The 
original CF
website which I created at NCAR was moved to LLNL quite a few years ago,
but for reasons I've never understood it wasn't possible, at least at 
that
time, to support a Mailman server there.  So it stayed at NCAR, and has
been stuck here ever since.

Again, I don't really see why moving all discussion to github 
wouldn't be
preferable to having discussions on both github and a Mailman list
(wherever that list might reside).  But I'm not a github expert so don't
know whether there are drawbacks to having all the discussions there.

Brian

On Wed, Nov 7, 2018 at 3:14 PM Painter, Jeff <paint...@llnl.gov 
<mailto:paint...@llnl.gov>> wrote:

    It certainly is a hack!  When we set up this two-list system long
    ago, LLNL's lists were managed by Majordomo.  No they are run by
    LISTSERV.  I don't know whether LISTSERV supports flexible topics
    within the list.

    On 11/7/18, 1:58 PM, "Erik Quaeghebeur"
    <e.r.g.quaegheb...@tudelft.nl <mailto:e.r.g.quaegheb...@tudelft.nl>>
    wrote:

         Dear Brian,


         > I don't believe that cf-metadata@listserv.llnl.gov
    <mailto:cf-metadata@listserv.llnl.gov> is a mailman list.  The
         > cf-metadata mailman list (which I administer at NCAR) is
         > cf-metad...@cgd.ucar.edu 
<mailto:cf-metad...@cgd.ucar.edu>.     The LLNL server is the one 
that gets the github
         > notifications and sends them to *mostly* the same people who
    are members of
         > the mailman list.  I say mostly because the membership in
    these lists is
         > synchronized by hand.  When people sign up or remove
    themselves from the
         > mailman list, I send this information to Jeff Painter who
    then makes the
         > changes at LLNL.  We originally set things up like this so
    that people
         > didn't need to sign up in two places to follow the
    discussions that were
         > happening on the mailman list and on the trac server at
    LLNL.  As github
         > has replaced trac the current system was set up.

         I understand the logic, but this sounds like a hack. It seems
    this can be
         dealt with as follows:

         * You as cf-metadata admin creates topics as needed
         * LLNL list traffic is sent to the cf-metadata list using one
    or more
         specific topic that people can (un)subscribe (from) to.

         It seems to me such an approach would have multiple advantages:

         * No need to (manually) sync users
         * Users are in control of selecting topics
         * Reverse traffic is likely possible by users sending mail 
to the
         appropriate topic (how exactly this works, I don't know, but it
    would
         surprise me if this were not possible)

         I don't have a view of the whole setup, so I may be missing
    some aspects
         that complicate this. But even then, it may be worth the 
effort.


         Best,

         Erik