Propsing a move from Google Groups to Discourse

Development
21

Putting what I think is the more important part first …

> What we are running is a mailing list here.

We are running a mailing list today, and you're right, I want to
change that. This may well be where we simply have to stop and let
others join in the conversation, as I don't think either of us is going
to back down :wink:

> It's a mailing list, we are open source developers and this is our
> way of communication.

It's your preferred way of communication, yes (and probably mine too,
to be honest). Don't forecast that onto all 2500 users signed up to our
lists, though - we may be in a minority. I'll quote from the Art of
Community here:

"Each type of contributor will have different preferences. Software
developers generally prefer content to be delivered directly to them.
They are typically most comfortable with mailing lists and RSS feeds
(updated content from websites and online resources) and don’t like to
have to refresh a browser to see if new content exists. This is part of
why many (typically Western) developers don’t get on very well with forums."

"Users are (typically) different. Users often love forums for their
accessibility and simplicity. The conversation flow is clear, the
interface is friendly, and the web browser is a familiar window
to that world. Users are used to having to refresh their browser to see
if updates exist. They are used to visiting many websites to find
content, and they generally feel uncomfortable about technical barriers
to these discussions and content. Users just don’t like to jump through
hoops, particularly technical hoops that can easily trip them up."

Now our users are more technical than most, but still, this has been my
experince as well - users want shiny, developers do not. Honestly I'm
surprised no one else has complained along with you yet :slight_smile:

> I understand how this is supposed to work technically, but will my
> gmail.com handle this correctly? This is not how e-mail lists are
> supposed to work. Will my MUA work correctly? Won't I see broken
> threads because folks will introduce new shiny feature into Discourse
> that does not play with plain emails anymore? I can't tell. Yeah you
> set up a testing instance, but there is almost no real traffic and I
> really don't know how it will look like with thousands of emails. My
> MUA handles millions of them just fine.

These are well documented standards. In my emails I'm seeing what I
think are the right headers:

List-ID: <hiscategoryimtesting.community.theforeman.org>
List-Archive: <long discourse url>
List-Unsubscribe: <another long discourse url>
Auto-Submitted: auto-generated
In-Reply-To: <90019b41-28a7-babc-65da-5843e20980d8@emeraldreverie.org>
References: <90019b41-28a7-babc-65da-5843e20980d8@emeraldreverie.org>
Precedence: list

That, I believe, should be sufficient, and things are getting threaded
here, at least with the volume I have.

As you say, we need more volume to be sure, which is exactly why I've
made a test category for people to mess with before we take any
decisions. Please try it out! We won't know unless we try, and that
means at least a few people making a thread in the testing area, with
mailing list mode enabled (under personal prefs > email)

(Note there's a 2 min delay in creating new topics, 30s delay betweem
posts to the same topic, and a 5s delay for any new post - going faster
than that will mean you'll get a bounce)

Also, it seems Ohad didn't get around to enabling the mailforward I
requested, which is why topic creation by email was failing (yay DNS).
I've re-used the dev one for now, emails to community-dev@theforeman.org
seem to be working fine for creating topics in the test area (i've
started one to test inline replies with too, which see to be persed nicely).

> Ever been to XdaDevelopers searching for Android ROM?

Oh god, XDA is awful, yes, we are not going there. But the fact is we
have some issues with our users list (the dev list benefits less from
this, again because of that quote above, but it does benefit, and I
don't want to run a forum and a list). You may not like recommended
topics, but it could be of significant benefit to the users group (and
actually it can be restricted by category). Like any tool, if you use it
wrong its a bad thing (like XDA).
> My response is not about "look here is an alternative HyperKitty". I'd
> be fine with any other mailing list

And there's our impasse, I want to move away from a list. But it doesn't
change the choice in front of us:

  • do nothing
  • switch mailing list for minimal improvement
  • switch to a forum, big upheaval but potential big payoff

> Yeah, feels like I am the only one. Not a good feeling, really.

Give it time. I doubt you are alone, and this debate is not done.

Cheers,
Greg

··· On 03/11/17 13:30, Lukas Zapletal wrote:
22

One more thought occurred to me while I was out on the nursery pickup, so I'll drop here before I bow out for the weekend.

Lukas, I think part of our disagreement is our different goals. As I highlighted in the last mail, users behave differently to devs. These days I consider myself more user than dev (when did I last contribute code), so I have a different world view.

You want to protect a tried and trusted workflow, likely used by many here - that's fine. My job is to promote and develop the user community, so I see room for improvement.

Here's the catch though… Our future devs, as a community, come from the user community. If we don't focus there, then we risk stagnating the dev community too.

I won't deny this change is a larger net benefit for the user group. The case for the dev community is harder to argue. But there is benefit, and compared to running a list (for dev) and a forum (for users) I think the better argument is to use a forum for both.

I don't expect to convince everyone, so this is going to come down to a group decision - but not for a while yet. We need to do more tests.

Have a great weekend all,
Greg

··· -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
23

I also think it's important to note that just because things like mailman
have existed for years and will continue to exist for years does not mean
they're always the best tool for the job. As open source developers it's
also important to be open to change, and always to evaluate what the best
tool to get the job done is. A false dichotomy for sure, but: can I
continue running CentOS 5 or Solaris 10? Sure, but it probably will
stagnate me from doing what I want to do. It's like doing "DevOps" by using
clusterssh to run commands on all your servers instead of a config
management tool.

Greg is right–many would-be new developers are turned off from things
like mailing lists. I know because I used to be turned off by them–and to
a certain extent, still am. There's nothing worse than finding a -dev
list on google for an issue you're having and not being able to find or
indeed know if there's a response to a thread: Forum-type solutions don't
have that issue, and as long as they support email, even if it's not
"first", then I think any distruption in workflow, if limited, is worth it
to encourage new voices. I recently graduated from University, and know a
lot of people that wouldn't step near a mailing list for development–and
would be especially reluctant to contribute to them.

To many "next gen" developers, email is not the workflow they will choose
to adopt, and sticking with listservs because they have always been and
will always be is not a good enough reason in my opinion.

(Please note this is mostly random rambling on my behalf, and entirely
opinion based to attempt to give insight from someone who's only recently
joined the whole process)

··· On Fri, Nov 3, 2017 at 1:45 PM Greg Sutcliffe wrote:

One more thought occurred to me while I was out on the nursery pickup, so
I’ll drop here before I bow out for the weekend.

Lukas, I think part of our disagreement is our different goals. As I
highlighted in the last mail, users behave differently to devs. These days
I consider myself more user than dev (when did I last contribute code), so
I have a different world view.

You want to protect a tried and trusted workflow, likely used by many here

  • that’s fine. My job is to promote and develop the user community, so I
    see room for improvement.

Here’s the catch though… Our future devs, as a community, come from
the user community. If we don’t focus there, then we risk stagnating the
dev community too.

I won’t deny this change is a larger net benefit for the user group. The
case for the dev community is harder to argue. But there is benefit, and
compared to running a list (for dev) and a forum (for users) I think the
better argument is to use a forum for both.

I don’t expect to convince everyone, so this is going to come down to a
group decision - but not for a while yet. We need to do more tests.

Have a great weekend all,
Greg

Sent from my Android device with K-9 Mail. Please excuse my brevity.

–
You received this message because you are subscribed to the Google Groups
“foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send an
email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

24

Greg, I absolutely understand the motivation, every two years amount
of programmers doubles. That is a crazy amount of newcomers. But these
new people are not idiots and some technical level is required even
for soft roles in our community. And we can make lists approachable
very much like forums.

Do not put me into position of blind and angry dev who can't accept
something different or new. I understand all contexts and I say
Discourse is an overkill that will bother me and possibly others. God
I wish Google Groups are gone, but not for this.

> * do nothing

Honestly, yeah.

> * switch mailing list for minimal improvement

s/minimal/reasonable/

> * switch to a forum, big upheaval but potential big payoff

Sure, because there are no downsides.

It's not about a list standard e-mail headers. The forum has different
workflow and features and there will be new features as well while
mailing list will stay the same. This will screw my inbox. This will
but a wall between e-mail users and web forum users. This is what's
this all about. And I think we don't need to go that direction.

LZ

··· On Fri, Nov 3, 2017 at 6:45 PM, Greg Sutcliffe wrote: > One more thought occurred to me while I was out on the nursery pickup, so I'll drop here before I bow out for the weekend. > > Lukas, I think part of our disagreement is our different goals. As I highlighted in the last mail, users behave differently to devs. These days I consider myself more user than dev (when did I last contribute code), so I have a different world view. > > You want to protect a tried and trusted workflow, likely used by many here - that's fine. My job is to promote and develop the user community, so I see room for improvement. > > Here's the catch though... Our future devs, as a community, *come from* the user community. If we don't focus there, then we risk stagnating the dev community too. > > I won't deny this change is a larger net benefit for the user group. The case for the dev community is harder to argue. But there *is* benefit, and compared to running a list (for dev) and a forum (for users) I think the better argument is to use a forum for both. > > I don't expect to convince everyone, so this is going to come down to a group decision - but not for a while yet. We need to do more tests. > > Have a great weekend all, > Greg > -- > Sent from my Android device with K-9 Mail. Please excuse my brevity. > > -- > You received this message because you are subscribed to the Google Groups "foreman-dev" group. > To unsubscribe from this group and stop receiving emails from it, send an email to foreman-dev+unsubscribe@googlegroups.com. > For more options, visit https://groups.google.com/d/optout.

–
Later,
Lukas @lzap Zapletal

25

I think about this in two ways:

  1. Who are our lists for?
  2. How can we provide the most value through our lists to the audience?

Often I find digging up old threads to reference to users more painful than
it should be. Often, I am curious about popular or trending threads and
cannot find this information easily. This latter point I think can be
important for users and developers if there is a hot button breakage or
workflow topic being discussed. Often, when I want to answer a user, I find
the interface of email lists to be limited when it comes to including
screenshots, or writing code blocks. Further, I find asking users for
information to help with debugging difficult because they have limited
options for attachments or screenshots for inclusion. Often I find writing
structured emails for things like proposals or recaps difficult. 75% of the
time I am going to prefer email given it is what I am used to and my
primary interface for everything else. But 25%, I want something more.

If Discourse can help solve these problems, and make the users of the lists
experiences better when interacting amongst themselves as well as
developers then a big ole +1 from me. Mailing lists are great, times
change, users change, requirements change.

Mailing lists are for communication, and whatever increases communication
the most I am all for.

Eric

··· On Fri, Nov 3, 2017 at 2:29 PM, Lukas Zapletal wrote:

Greg, I absolutely understand the motivation, every two years amount
of programmers doubles. That is a crazy amount of newcomers. But these
new people are not idiots and some technical level is required even
for soft roles in our community. And we can make lists approachable
very much like forums.

Do not put me into position of blind and angry dev who can’t accept
something different or new. I understand all contexts and I say
Discourse is an overkill that will bother me and possibly others. God
I wish Google Groups are gone, but not for this.

  • do nothing

Honestly, yeah.

  • switch mailing list for minimal improvement

s/minimal/reasonable/

  • switch to a forum, big upheaval but potential big payoff

Sure, because there are no downsides.

It’s not about a list standard e-mail headers. The forum has different
workflow and features and there will be new features as well while
mailing list will stay the same. This will screw my inbox. This will
but a wall between e-mail users and web forum users. This is what’s
this all about. And I think we don’t need to go that direction.

LZ

On Fri, Nov 3, 2017 at 6:45 PM, Greg Sutcliffe greg@emeraldreverie.org > wrote:

One more thought occurred to me while I was out on the nursery pickup,
so I’ll drop here before I bow out for the weekend.

Lukas, I think part of our disagreement is our different goals. As I
highlighted in the last mail, users behave differently to devs. These days
I consider myself more user than dev (when did I last contribute code), so
I have a different world view.

You want to protect a tried and trusted workflow, likely used by many
here - that’s fine. My job is to promote and develop the user community, so
I see room for improvement.

Here’s the catch though… Our future devs, as a community, come from
the user community. If we don’t focus there, then we risk stagnating the
dev community too.

I won’t deny this change is a larger net benefit for the user group. The
case for the dev community is harder to argue. But there is benefit, and
compared to running a list (for dev) and a forum (for users) I think the
better argument is to use a forum for both.

I don’t expect to convince everyone, so this is going to come down to a
group decision - but not for a while yet. We need to do more tests.

Have a great weekend all,
Greg

Sent from my Android device with K-9 Mail. Please excuse my brevity.

–
You received this message because you are subscribed to the Google
Groups “foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send
an email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

–
Later,
Lukas @lzap Zapletal

–
You received this message because you are subscribed to the Google Groups
“foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send an
email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

–
Eric D. Helms
Red Hat Engineering

26

For monitoring of what is going on on the foreman-dev/users I prefer to
consume it as a mailing list. It is lightweight and efficient and fits well
to my mail-centric workflow. I understand the benefits of the forum so I
gave Discourse a try to see how it works and if its mailing-list mode
promises smooth transition.

Things I like:

  • searching during new post compose
  • existence of "groups"
  • likes (even work when sent via mail)
  • rich-text messages, syntax highlighting, markdown
  • easy to share links to individual posts

Things I didn't like (I guess some are likely interference with the Gmail
client and some can be tuned up):

  • for some reason the threads are not kept together in my Gmail and the
    messages from one thread are split into multiple threads even if they seem
    to have same subject. I'm not sure why, it may be because I tuned the
    account settings. I'll keep testing this
  • it took about 15 min since I sent mail to the time I received it from
    the list (not sure what are the reaction times on the list today but this
    won't improve it)
  • mails from Discourse take too much visual space - the footer saying how
    to unsubscribe, reply or visit the topic is included in each message.
    First post should be enough. There is also extra username with avatar and
    forum role next to User name in the From field. Is this configurable?
  • "likes" are only indicated in forum notifications but not in emails. If
    you send '+1' to the list the like is added but no message is sent to the
    users (just the notification)

So far for me it is difficult to follow the Discourse discussion using just
Gmail. For further testing I'd like to see more traffic in the Testing
area. I'd also appreciate experience with mailing-list mode testing form
others.

M.

··· On Fri, Nov 3, 2017 at 7:29 PM, Lukas Zapletal wrote:

Greg, I absolutely understand the motivation, every two years amount
of programmers doubles. That is a crazy amount of newcomers. But these
new people are not idiots and some technical level is required even
for soft roles in our community. And we can make lists approachable
very much like forums.

Do not put me into position of blind and angry dev who can’t accept
something different or new. I understand all contexts and I say
Discourse is an overkill that will bother me and possibly others. God
I wish Google Groups are gone, but not for this.

  • do nothing

Honestly, yeah.

  • switch mailing list for minimal improvement

s/minimal/reasonable/

  • switch to a forum, big upheaval but potential big payoff

Sure, because there are no downsides.

It’s not about a list standard e-mail headers. The forum has different
workflow and features and there will be new features as well while
mailing list will stay the same. This will screw my inbox. This will
but a wall between e-mail users and web forum users. This is what’s
this all about. And I think we don’t need to go that direction.

LZ

On Fri, Nov 3, 2017 at 6:45 PM, Greg Sutcliffe greg@emeraldreverie.org > wrote:

One more thought occurred to me while I was out on the nursery pickup,
so I’ll drop here before I bow out for the weekend.

Lukas, I think part of our disagreement is our different goals. As I
highlighted in the last mail, users behave differently to devs. These days
I consider myself more user than dev (when did I last contribute code), so
I have a different world view.

You want to protect a tried and trusted workflow, likely used by many
here - that’s fine. My job is to promote and develop the user community, so
I see room for improvement.

Here’s the catch though… Our future devs, as a community, come from
the user community. If we don’t focus there, then we risk stagnating the
dev community too.

I won’t deny this change is a larger net benefit for the user group. The
case for the dev community is harder to argue. But there is benefit, and
compared to running a list (for dev) and a forum (for users) I think the
better argument is to use a forum for both.

I don’t expect to convince everyone, so this is going to come down to a
group decision - but not for a while yet. We need to do more tests.

Have a great weekend all,
Greg

Sent from my Android device with K-9 Mail. Please excuse my brevity.

–
You received this message because you are subscribed to the Google
Groups “foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send
an email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

–
Later,
Lukas @lzap Zapletal

–
You received this message because you are subscribed to the Google Groups
“foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send an
email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

27

My views do align with Martin here: email does feel like a second class
citizen. Sending email does work properly (likely because they could
just use the github parser gem) but what it sends out is … barely
enough.

On the other hand (and this has been pointed out by other people), it
might be much better for users. Given that foreman-dev is not a very
high activity mailing list (which I like, good signal/noise ratio) I
think it can be a good trade-off for better user interaction. While we
could consider splitting into a discourse for -users and a mailing list
for -dev, IMHO the downsides of that a much bigger than the upsides.

Right now we have done an experiment and with these initial findings I
think we can approach the Discourse community to ask them what their
views are. Perhaps we can work together to improve it (yay open source).

To state what's perhaps obvious: we're never going to find a perfect
solution that makes everybody happy. We should strive to find a local
maximum that makes most people happy / the least people unhappy. For
-dev we can ask the devs since that's a pretty consistent group and we
know most. That's not true for -users since they might be unhappy now
but not tell us. I'm willing to trust others that it's the better
choice even if I see some downsides for me personally.

··· On Sun, Nov 05, 2017 at 07:51:45PM +0100, Martin Bačovský wrote: >For monitoring of what is going on on the foreman-dev/users I prefer to >consume it as a mailing list. It is lightweight and efficient and fits well >to my mail-centric workflow. I understand the benefits of the forum so I >gave Discourse a try to see how it works and if its mailing-list mode >promises smooth transition. > >Things I like: > - searching during new post compose > - existence of "groups" > - likes (even work when sent via mail) > - rich-text messages, syntax highlighting, markdown > - easy to share links to individual posts > >Things I didn't like (I guess some are likely interference with the Gmail >client and some can be tuned up): > - for some reason the threads are not kept together in my Gmail and the >messages from one thread are split into multiple threads even if they seem >to have same subject. I'm not sure why, it may be because I tuned the >account settings. I'll keep testing this > - it took about 15 min since I sent mail to the time I received it from >the list (not sure what are the reaction times on the list today but this >won't improve it) > - mails from Discourse take too much visual space - the footer saying how >to unsubscribe, reply or visit the topic is included in each message. >First post should be enough. There is also extra username with avatar and >forum role next to User name in the From field. Is this configurable? > - "likes" are only indicated in forum notifications but not in emails. If >you send '+1' to the list the like is added but no message is sent to the >users (just the notification) > >So far for me it is difficult to follow the Discourse discussion using just >Gmail. For further testing I'd like to see more traffic in the Testing >area. I'd also appreciate experience with mailing-list mode testing form >others. > >M. > > > >On Fri, Nov 3, 2017 at 7:29 PM, Lukas Zapletal wrote: > >> Greg, I absolutely understand the motivation, every two years amount >> of programmers doubles. That is a crazy amount of newcomers. But these >> new people are not idiots and some technical level is required even >> for soft roles in our community. And we can make lists approachable >> very much like forums. >> >> Do not put me into position of blind and angry dev who can't accept >> something different or new. I understand all contexts and I say >> Discourse is an overkill that will bother me and possibly others. God >> I wish Google Groups are gone, but not for this. >> >> > * do nothing >> >> Honestly, yeah. >> >> > * switch mailing list for minimal improvement >> >> s/minimal/reasonable/ >> >> > * switch to a forum, big upheaval but potential big payoff >> >> Sure, because there are no downsides. >> >> It's not about a list standard e-mail headers. The forum has different >> workflow and features and there will be new features as well while >> mailing list will stay the same. This will screw my inbox. This will >> but a wall between e-mail users and web forum users. This is what's >> this all about. And I think we don't need to go that direction. >> >> LZ >> >> On Fri, Nov 3, 2017 at 6:45 PM, Greg Sutcliffe >> wrote: >> > One more thought occurred to me while I was out on the nursery pickup, >> so I'll drop here before I bow out for the weekend. >> > >> > Lukas, I think part of our disagreement is our different goals. As I >> highlighted in the last mail, users behave differently to devs. These days >> I consider myself more user than dev (when did I last contribute code), so >> I have a different world view. >> > >> > You want to protect a tried and trusted workflow, likely used by many >> here - that's fine. My job is to promote and develop the user community, so >> I see room for improvement. >> > >> > Here's the catch though... Our future devs, as a community, *come from* >> the user community. If we don't focus there, then we risk stagnating the >> dev community too. >> > >> > I won't deny this change is a larger net benefit for the user group. The >> case for the dev community is harder to argue. But there *is* benefit, and >> compared to running a list (for dev) and a forum (for users) I think the >> better argument is to use a forum for both. >> > >> > I don't expect to convince everyone, so this is going to come down to a >> group decision - but not for a while yet. We need to do more tests. >> > >> > Have a great weekend all, >> > Greg
28

> No, I don't think we can. We've been over this in the preceeding emails.
> We can give an archive a shiny GUI, but fundamentally they have a
> different feature set, and a different target user group. Your position
> on Discourse is clear, and recorded here for all to see. If enough
> people agree with you, it won't happen.

HyperKitty is not just a nice archive, you can easily just use mailing
list as a forum. There are other options like Gmane and others, but I
find HyperKitty as the nicest and smoothest experience. It is the
right interface into mailing list, which some of us want to continue
using. I just want to repeat it here in the new thread for others.

It's not just in hating Discourse, I particularly do not like big bang
migration. I would not write a single word if you'd propose installing
Discourse next to our existing communication channels. We have a lot
of people who subscribed for something and you are forcing them onto
something they haven't subscribed for. And we won't read their opinion
here until you do the migration and they will see (the terrible - I
think) experience. I'd prefer slower-paced rollout with old services
fading out (might not happen), therefore I'd like to propose other
options:

  1. Ditch Discourse and move to Fedora Hosted mailinglists with
    HyperKitty interface (*)
  2. Phased rollout: Install Discourse and keep our -user and -dev lists
    on Google Groups
  3. (2) + integrating Discourse with -user list in some way (daily
    summary mail to -user list)
  4. Migrate only user list and keep -dev list on Google Groups

(*) Yeah still my preferred way, you can easily do new posts without
subscribing or doing any e-mail:

https://lists.fedoraproject.org/archives/list/test-mailman3@lists.fedoraproject.org/message/new

··· -- Lukas @lzap Zapletal
29

Thanks for returning to excellent debating form :slight_smile:

> HyperKitty is not just a nice archive, you can easily just use mailing
> list as a forum. There are other options like Gmane and others, but I
> find HyperKitty as the nicest and smoothest experience.

It's a lovely interface, but it's still a mailing list, and as I've
said, I don't think a mailing list is a one-to-one equivalent to a
forum. The features are fundamentally different, and I don't think it's
possible to make both things first-class citizens (or we'd all be using
that project already).

> It is the right interface into mailing list, which some of us want
> to continue using.
Absolutely, it's a fantastic interface for a mailing list - but it's
still a mailing list. It's time to move to a forum, in my opinion.

Your opinion is valid too though, of course, I don't want to say it
isn't. As Ewoud said, we can't make everyone happy, it's about figuring
out what the majority of opinions are, and acting on that.

> It's not just in hating Discourse, I particularly do not like big bang
> migration.

100% agree, that's why I've waited over a year to propose this. I've
been over and over it, and there's no other option (see below). Again, I
don't do this lightly, for fun, or because I like upsetting people. It's
because it's needed (in my view). I will set out the "how" of the actual
migration later this week, so we can discuss how to minimize that
big-bang (if we go ahead).

> I would not write a single word if you'd propose installing
> Discourse next to our existing communication channels.

This is a classic fallacy I'm afraid - putting two services side by side
and seeing if users move does not account for resistance-to-change.
There are so many examples of people resisting a change, only to
accept the change was good afterwards. Here's just one [1] (and then
read the posts above it, some more good takes on the benefits of a forum).

> We have a lot of people who subscribed for something and you are
> forcing them onto something they haven't subscribed for.

You're right that we haven't asked them … yet. I plan to open up the
instance to the users-list at some point soon and get their feedback
too. Community buy-in is the right way to make change, I've merely
started with foreman-dev because it's the more active group (and usually
decisions are taken here).

> And we won't read their opinion here until you do the migration and
> they will see (the terrible - I think) experience. I'd prefer
> slower-paced rollout with old services fading out (might not happen),

We're also not reading the opinions of the people who haven't joined our
community because the mailing list is a problem for them. Which group is
larger? We have no way to know, so all we can do is hope they cancel
out, until we can get better data (probably never).

> therefore I'd like to propose other options:
>
> 1) Ditch Discourse and move to Fedora Hosted mailinglists with
> HyperKitty interface (*)

As you already know, that's -1 from me.

> 2) Phased rollout: Install Discourse and keep our -user and -dev lists
> on Google Groups
> 3) (2) + integrating Discourse with -user list in some way (daily
> summary mail to -user list)

Both of these will suffer the resistance-to-change problem above.

In addition this causes a split in the community, and will lead to
repeat questions asked in both places, and potentially higher workload
in answering questions - I'm not in favour of that when we're all
already busy enough. The split in the knowledge archives is bad too.

As an example of this, consider that there is a Foreman StackOverflow
board [2] (this may be news to some people here). These questions look
remarkably similar to ones asked in other places, and no-one is
answering them. Splitting channels splits users, devs, knowledge and
ultimately the community. Also, this may be wishful, but the fact that
people choose to post on StackOverflow rather than our mailing list
again shows a desire for more modern forms of interaction.

> 4) Migrate only user list and keep -dev list on Google Groups

Ewoud mentioned this, and I agree with him - the downsides are more than
the upsides. It's splitting the community (devs not in communication
with users) but is at least possible. I'm not sold that the barrier
between the users and devs is worthwhile, but I'm definitely less
against this than the other options. So I guess that's -0.5 compared to
the other choices :slight_smile:

[1] https://groups.google.com/d/msg/opendatakit/gG6D4Gfwh44/YUhGedXaCAAJ
[2] https://stackoverflow.com/questions/tagged/foreman

Greg

··· On 07/11/17 08:50, Lukas Zapletal wrote:
30

Just a quick update on threading, since that's causing the most pain…

> - for some reason the threads are not kept together in my Gmail and the
> messages from one thread are split into multiple threads even if they seem
> to have same subject. I'm not sure why, it may be because I tuned the
> account settings. I'll keep testing this

In short, it appears Mailjet (the SMTP service I'm using) are mangling
outgoing Message-ID headers :confused:

Discourse has some fairly sane code around re-using the incoming
message-id if a post comes in by email, or generating a sensible one if
it was written in the UI. That means you'd get a header like:

Message-ID: <topic/12345/4@community.theforeman.org>

Sadly what's arriving in my inbox has a header like:

Message-ID <uuid@mailjet.com>

That's what's breaking the threading, not Discourse itself ( I think,
it's hard to separate currently). Interesting the Redmine email does not
do this, I see this:

Message-ID: <redmine.journal-uuid@theforeman.org>
References: <redmine.issue-uuid@theforeman.org>

So Mailgun is leaving it intact. I'll see about switching over to
Mailgun shortly (sadly that means I have to sort out new DKIM and SPF
records, sigh), and hopefully this threading mess will go away.

Greg

··· On 05/11/17 18:51, Martin Bačovský wrote:
31

I will just reply to phased rollout.

> Also, this may be wishful, but the fact that
> people choose to post on StackOverflow rather than our mailing list
> again shows a desire for more modern forms of interaction.

Why are we creating a duplicate of StackOverflow then? There is a
community already, let's just embrace it, advertise on dev channels,
create links from site, blog post and integrate more tightly with
-user list. (*)

I don't really see much value in building another StackOverflow. You
are not sure that these people which are comfortable with discussing
there will move to Discourse.

(*) https://stackoverflow.blog/2010/12/20/subscribe-to-tags-via-emai/

··· -- Later, Lukas @lzap Zapletal
32

Not reacting to everything here, but I would like to keep the
discussion focused on facts.

> Thanks for returning to excellent debating form :slight_smile:
>
>> HyperKitty is not just a nice archive, you can easily just use mailing
>> list as a forum. There are other options like Gmane and others, but I
>> find HyperKitty as the nicest and smoothest experience.
>
> It's a lovely interface, but it's still a mailing list, and as I've
> said, I don't think a mailing list is a one-to-one equivalent to a
> forum. The features are fundamentally different, and I don't think it's
> possible to make both things first-class citizens (or we'd all be using
> that project already).
>
>> It is the right interface into mailing list, which some of us want
>> to continue using.
> Absolutely, it's a fantastic interface for a mailing list - but it's
> still a mailing list. It's time to move to a forum, in my opinion.
>
> Your opinion is valid too though, of course, I don't want to say it
> isn't. As Ewoud said, we can't make everyone happy, it's about figuring
> out what the majority of opinions are, and acting on that.
>
>> It's not just in hating Discourse, I particularly do not like big bang
>> migration.
>
> 100% agree, that's why I've waited over a year to propose this. I've
> been over and over it, and there's no other option (see below). Again, I
> don't do this lightly, for fun, or because I like upsetting people. It's
> because it's needed (in my view). I will set out the "how" of the actual
> migration later this week, so we can discuss how to minimize that
> big-bang (if we go ahead).

I'm afraid the amount of time waiting doesn't have any influence on the
feeling if something is big bang or not. So far, it seems there are issues
raised, and discussion going (regardless of the fact, that it's just a
mailing list:).
Let's not try to harsh to a decision: keep in mind that there might be folks
not repsponding just because there are other things to do as well.

>
>> I would not write a single word if you'd propose installing
>> Discourse next to our existing communication channels.
>
> This is a classic fallacy I'm afraid - putting two services side by side
> and seeing if users move does not account for resistance-to-change.
> There are so many examples of people resisting a change, only to
> accept the change was good afterwards. Here's just one [1] (and then
> read the posts above it, some more good takes on the benefits of a forum).

Sry, but I can't resist, as the fallacy I see here is the straw-man: you can
make an argument for every change based on the fact that people in general
hate change: it's irrelevant here. Please don't :slight_smile:

>
>> We have a lot of people who subscribed for something and you are
>> forcing them onto something they haven't subscribed for.
>
> You're right that we haven't asked them … yet. I plan to open up the
> instance to the users-list at some point soon and get their feedback
> too. Community buy-in is the right way to make change, I've merely
> started with foreman-dev because it's the more active group (and usually
> decisions are taken here).
>
>> And we won't read their opinion here until you do the migration and
>> they will see (the terrible - I think) experience. I'd prefer
>> slower-paced rollout with old services fading out (might not happen),
>
> We're also not reading the opinions of the people who haven't joined our
> community because the mailing list is a problem for them. Which group is
> larger? We have no way to know, so all we can do is hope they cancel
> out, until we can get better data (probably never).
>
>> therefore I'd like to propose other options:
>>
>> 1) Ditch Discourse and move to Fedora Hosted mailinglists with
>> HyperKitty interface (*)
>
> As you already know, that's -1 from me.
>
>> 2) Phased rollout: Install Discourse and keep our -user and -dev lists
>> on Google Groups
>> 3) (2) + integrating Discourse with -user list in some way (daily
>> summary mail to -user list)
>
> Both of these will suffer the resistance-to-change problem above.

Again, not a valid argument for me.

>
> In addition this causes a split in the community, and will lead to
> repeat questions asked in both places, and potentially higher workload
> in answering questions - I'm not in favour of that when we're all
> already busy enough. The split in the knowledge archives is bad too.

Watch out for slipper slope (but let's stop before somebody brings up
fallacy fallacy:)

There are other communities
having both forum and mailling list (didn't have to go too far,
my first attempt just hit Get Help | GitLab)
and honestly, I'm not afraid of any community split. And if we find ourselves
using forum for everything, good for everyone. If we find that the forum-like
is not for us, we still have the backup plan.

Anyway, having forum focused for users and mailing list on developer discussion
(for now at least) sounds like a natural split and possible evolution.

>
> As an example of this, consider that there is a Foreman StackOverflow
> board [2] (this may be news to some people here). These questions look
> remarkably similar to ones asked in other places, and no-one is
> answering them. Splitting channels splits users, devs, knowledge and
> ultimately the community. Also, this may be wishful, but the fact that
> people choose to post on StackOverflow rather than our mailing list
> again shows a desire for more modern forms of interaction.

Do you think having discourse would actually stop people asking on
stack overflow?
Why? That might be just a sign that people are used to ask about just
anything on stack
overflow.

>
>> 4) Migrate only user list and keep -dev list on Google Groups
>
> Ewoud mentioned this, and I agree with him - the downsides are more than
> the upsides. It's splitting the community (devs not in communication
> with users) but is at least possible. I'm not sold that the barrier
> between the users and devs is worthwhile, but I'm definitely less
> against this than the other options. So I guess that's -0.5 compared to
> the other choices :slight_smile:

What I find appealing about the forum-like approach is being able to
categorize threads
based on topics. This can have a potential of bringing attention to
the right folks when needed.
So far, I haven't seen if the Discourse can actually be used
effectively for that: let's give it a try.

Anyway, I don't think it's a silver bullet and all of our problems
just go away. I'm definitely for
at least a transition/evaluation period and once proven, we can decide
on whether the benefits
are as good as promised or we want to re-evaluate: this is not
resistance to change: this is
demage control.

The bad scenarios are:

  1. some of the developers community will refuse contributing to the
    forum and insisting
    only on mailing list: this is really bad and if anyone plans doing so,
    please speak up now

  2. we would have the forum as the only way to discuss things and it
    would make the
    discussion harder, not easier, for whatever reason: we can't know that
    until we try IMO

We can eliminate 1. by having this discussion and not rushing to the
decision and
2. can be eliminated by running the two side-by-side.

In the meantime, foreman-users and redmine notifications want you :slight_smile:

– Ivan

··· On Tue, Nov 7, 2017 at 5:17 PM, Greg Sutcliffe wrote: > On 07/11/17 08:50, Lukas Zapletal wrote:

[1] https://groups.google.com/d/msg/opendatakit/gG6D4Gfwh44/YUhGedXaCAAJ
[2] Newest 'foreman' Questions - Stack Overflow

Greg

–
You received this message because you are subscribed to the Google Groups “foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send an email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

33

Apparently my arguments aren't coming out coherently - I think we've
misunderstood each other on a couple of points…

> I'm afraid the amount of time waiting doesn't have any influence on
> the feeling if something is big bang or not.

Sure, that's not what I was trying to say, I was just trying to give
some insight into my thinking.

I knew the likely reaction to a big change would be negative (in some
respects), and that partial solutions would be proposed. So I spent time
looking at softer alternatives and indeed some of the partial solutions
already proposed, and didn't really come up with any that I thought
would work. So, faced with doing something hard, I instead kept waiting
to bring this up, and pushing it back, and pushing it back. I knew I was
in for a tough discussion, and so I didn't do it. That's a personal
failure - I saw a need in the community, and I chose not to address it.

However, I don't feel it can be pushed back any longer. So here we are.

> Let's not try to harsh to a decision: keep in mind that there might
> be folks not repsponding just because there are other things to do
> as well.

Agreed, we have a good debate going, and we need time to evaluate. I
will also post a summary of the discussion so far for those who are
super-busy. That will probably be tomorrow, since it'll be 1 week since
I opened up the Discourse instance for testing. Please do make sure I
represent things fairly (I know you will :P)

> Sry, but I can't resist, as the fallacy I see here is the straw-man:
> you can make an argument for every change based on the fact that
> people in general hate change: it's irrelevant here. Please don't :slight_smile:

I've not made myself clear, I think, sorry for that. I'm not using this
argument to justify the move to Discourse itself, that would indeed be a
strawman. A change must be justified on its own merits, and hopefully my
arguments for that are clear.

Here I'm specifically arguing against running Discourse alongside an
existing list, and for that I think it's not a strawman. The usual
criteria for something like this is looking at how many users migrate to
the new system, despite having the old one available.

However, if you put a new system alongside an existing system, you are
pretty much guaranteed to have very few people move to the new one after
some time. That's human psychology at work, sadly, and not any comment
on the quality of either solution. People will largely choose no-effort
over doing-something even when there's clear benefits (look at how many
people don't switch to lower tariffs on their utility bills, even though
it will save them money).

Therefore, I argue that a proposal to run Discourse alongside an
existing list is set up to fail from the beginning, and I would be
against that.

>> Both of these will suffer the resistance-to-change problem above.
>
> Again, not a valid argument for me.

Hopefully I've clarified that argument now.

> There are other communities having both forum and mailling list
> (didn't have to go too far, my first attempt just hit
> https://about.gitlab.com/getting-help/#discussion)

Interestingly they also use Discourse & Gougle Groups (nice styling on
that instance, I need to copy…), and I see that the mailing list gets
~5 mails per month, and Discourse seems to have ~250 posts per month. I
won't repeat the mistake of trying to draw conclusions about our
community (see reply about S.O. below), but it certainly seems their
community has a strong preference. Not making any arguments here, I just
find the raw data interesting.

> and honestly, I'm not afraid of any community split. And if we find
> ourselves using forum for everything, good for everyone. If we find
> that the forum-like is not for us, we still have the backup plan.
>
> Anyway, having forum focused for users and mailing list on developer
> discussion (for now at least) sounds like a natural split and
> possible evolution.

Oh absolutely, this is possible, and as I said I'm just about OK with
this. It's not my preferred choice ofc, but so long as we agree to
review it every so often, I can live with it.

My main concern is that we still get a good interaction between the
users and the devs. We can't actually ensure that today - the lists are
entirely separate, so I see it as a positive that I can @mention someone
to involve them in a thread (whereas today I have to ping them on IRC or
forward the mail to them). That's an improvement in communication
options if we move everything, especially with @group support.

Basically, if we can agree some guidelines around the interaction
between -dev members and Discourse then I think it can work. But as
Ewoud said, I think having it all in one place is better if we can
handle the change.

> Do you think having discourse would actually stop people asking on
> stack overflow? Why? That might be just a sign that people are used
> to ask about just anything on stack overflow.

A fair argument, I am making assumptions - perhaps my point was unwisely
made. I was simply trying to say that I was aware of other groups (there
were also Facebook and LinkedIn groups at one time, there may be more)
who chose to use an unofficial platform (i.e. not something we list on
our support page) rather than interact with us on our official channels.
Why they chose to do so is an assumption, yes, and I don't think you
can reduce that to zero, no, but it can be minimized.

However, I do still think it implies that there is a group of people we
are not serving (as Neil outlined), who would like something other than
a list. We can solve that need while still maintaing something that
looks (at least a bit) like a list for those that want it.

Note: please do give me a chance to sort out this email mangling mess
before we conclude threading in list-mode is broken. As it looks right
now, it's not Discourse's fault, it's Mailjet's fault.

> What I find appealing about the forum-like approach is being able to
> categorize threads based on topics. This can have a potential of
> bringing attention to the right folks when needed. So far, I haven't
> seen if the Discourse can actually be used effectively for that:
> let's give it a try.

Obviously we can create entirely new categories as & when we choose, so
I'll assume you're looking for something more fine-grained within a one
or more categories.

There's a number of ways of doing this, depending on exactly what you're
after. Firstly posts can be tagged, and those tags can be used as
notification types (so I can be notified on, say, "salt"-tagged
threads). If that doesn't bring the right people in, then a mention in
any post (either a group, @salt, or a person, @gwmngilfen) will
also generate a notification for the appropriate people. Tags can't be
applied to a post by email (there's a feature request for that[1]), but
@mentions will definitely work in any format.

Both tags and groups are set up, and for now everyone has rights to
create new tags. Groups are admin-created, I made a few
(hamburger->Groups to join), let me know if you want more groups
created. Give both a try, but be aware that mailing-list mode already
sends you everything. You will need to disable that if you want to
test selectively getting some things and not others.

There may be other methods too, depending on our usecase. Discourse is a
Rails app, and many of us here know how to write Rails engines… [2]

> Anyway, I don't think it's a silver bullet and all of our problems
> just go away.

Agreed, but it brings a lot of tools to help that we don't currently
have. Like any tool, it must be used correctly, but we're really good at
discussing that kind of thing. I think we'll be able to find
features/plugins that solve needs as they come up - but first we need a
platform that lets us be that flexible.

I'm definitely for at least a transition/evaluation
> period and once proven, we can decide on whether the benefits are as
> good as promised or we want to re-evaluate: this is not resistance to
> change: this is demage control.

I prefer "risk mitigation" to "damage control", no damage has yet been
done :slight_smile:

What kind of trial are you thinking, particularly in reference to the
problems I outlined in side-by-side testing above?

> The bad scenarios are:
>
> 1. some of the developers community will refuse contributing to the
> forum and insisting only on mailing list: this is really bad and if
> anyone plans doing so, please speak up now

This is true, it's bad. But we must accept that we are highy unlikely to
get 100% agreement on this. If there's a significant majority in favour
of moving forward, we should. Aiming for 100% agreement on such a big
change is unrealistic and not something we've done on other
controversial decisions, either.

> 2. we would have the forum as the only way to discuss things and it
> would make the discussion harder, not easier, for whatever reason:
> we can't know that until we try IMO
>
> We can eliminate 1. by having this discussion and not rushing to the
> decision and 2. can be eliminated by running the two side-by-side.

I'll repeat my dislike for side-by-side in a single channel, that's not
going to work, at least long-term. As stated, I can be persuaded to do
-users separate to -dev if there are significant votes for that.

Greg

[1] https://meta.discourse.org/t/add-tags-by-email/25796/7
[2] https://meta.discourse.org/c/plugin

34

Another update on this, almost there :slight_smile:

Threading now apears entirely functional. Ewoud and I ran a private test
using Mailgun last night (so as not to spam everyone in case it was
still broken). As you can see, in this Gist, it appears to be working in
Mutt:

Email support now feels pretty much the same to what we have today.
There's even a List-ID in the headers to filter on.

I've also reduced the email send delay to 5 mins, and I'm waiting for
Ohad to set up an MX record for me so we can handle inbound mail
directly. That will bring us from 15mins (5 min polling + 10 min delay)
down to just 5 min, which is acceptable I think.

I'll update again once I have the MX record. Only Ohad can do this, so
my hands are tied.

Greg

35

> Apparently my arguments aren't coming out coherently - I think we've
> misunderstood each other on a couple of points…
>
>> I'm afraid the amount of time waiting doesn't have any influence on
>> the feeling if something is big bang or not.
>
> Sure, that's not what I was trying to say, I was just trying to give
> some insight into my thinking.
>
> I knew the likely reaction to a big change would be negative (in some
> respects), and that partial solutions would be proposed. So I spent time
> looking at softer alternatives and indeed some of the partial solutions
> already proposed, and didn't really come up with any that I thought
> would work. So, faced with doing something hard, I instead kept waiting
> to bring this up, and pushing it back, and pushing it back. I knew I was
> in for a tough discussion, and so I didn't do it. That's a personal
> failure - I saw a need in the community, and I chose not to address it.
>
> However, I don't feel it can be pushed back any longer. So here we are.
>
>> Let's not try to harsh to a decision: keep in mind that there might
>> be folks not repsponding just because there are other things to do
>> as well.
>
> Agreed, we have a good debate going, and we need time to evaluate. I
> will also post a summary of the discussion so far for those who are
> super-busy. That will probably be tomorrow, since it'll be 1 week since
> I opened up the Discourse instance for testing. Please do make sure I
> represent things fairly (I know you will :P)
>
>> Sry, but I can't resist, as the fallacy I see here is the straw-man:
>> you can make an argument for every change based on the fact that
>> people in general hate change: it's irrelevant here. Please don't :slight_smile:
>
> I've not made myself clear, I think, sorry for that. I'm not using this
> argument to justify the move to Discourse itself, that would indeed be a
> strawman. A change must be justified on its own merits, and hopefully my
> arguments for that are clear.
>
> Here I'm specifically arguing against running Discourse alongside an
> existing list, and for that I think it's not a strawman. The usual
> criteria for something like this is looking at how many users migrate to
> the new system, despite having the old one available.
>
> However, if you put a new system alongside an existing system, you are
> pretty much guaranteed to have very few people move to the new one after
> some time. That's human psychology at work, sadly, and not any comment
> on the quality of either solution. People will largely choose no-effort
> over doing-something even when there's clear benefits (look at how many
> people don't switch to lower tariffs on their utility bills, even though
> it will save them money).
>
> Therefore, I argue that a proposal to run Discourse alongside an
> existing list is set up to fail from the beginning, and I would be
> against that.
>
>>> Both of these will suffer the resistance-to-change problem above.
>>
>> Again, not a valid argument for me.
>
> Hopefully I've clarified that argument now.
>
>> There are other communities having both forum and mailling list
>> (didn't have to go too far, my first attempt just hit
>> Get Help | GitLab)
>
> Interestingly they also use Discourse & Gougle Groups (nice styling on
> that instance, I need to copy…), and I see that the mailing list gets
> ~5 mails per month, and Discourse seems to have ~250 posts per month. I
> won't repeat the mistake of trying to draw conclusions about our
> community (see reply about S.O. below), but it certainly seems their
> community has a strong preference. Not making any arguments here, I just
> find the raw data interesting.

And, if we meet in some time (let's say 6 months from the kick off)
and look at the numbers,
I think it would be much easier discussion if we should ditch the list or not.
Additionally, we would have 6 months of hands on experience with Discourse.

>
>> and honestly, I'm not afraid of any community split. And if we find
>> ourselves using forum for everything, good for everyone. If we find
>> that the forum-like is not for us, we still have the backup plan.
>>
>> Anyway, having forum focused for users and mailing list on developer
>> discussion (for now at least) sounds like a natural split and
>> possible evolution.
>
> Oh absolutely, this is possible, and as I said I'm just about OK with
> this. It's not my preferred choice ofc, but so long as we agree to
> review it every so often, I can live with it.
>
> My main concern is that we still get a good interaction between the
> users and the devs. We can't actually ensure that today - the lists are
> entirely separate, so I see it as a positive that I can @mention someone
> to involve them in a thread (whereas today I have to ping them on IRC or
> forward the mail to them). That's an improvement in communication
> options if we move everything, especially with @group support.
>
> Basically, if we can agree some guidelines around the interaction
> between -dev members and Discourse then I think it can work. But as
> Ewoud said, I think having it all in one place is better if we can
> handle the change.
>
>> Do you think having discourse would actually stop people asking on
>> stack overflow? Why? That might be just a sign that people are used
>> to ask about just anything on stack overflow.
>
> A fair argument, I am making assumptions - perhaps my point was unwisely
> made. I was simply trying to say that I was aware of other groups (there
> were also Facebook and LinkedIn groups at one time, there may be more)
> who chose to use an unofficial platform (i.e. not something we list on
> our support page) rather than interact with us on our official channels.
> Why they chose to do so is an assumption, yes, and I don't think you
> can reduce that to zero, no, but it can be minimized.
>
> However, I do still think it implies that there is a group of people we
> are not serving (as Neil outlined), who would like something other than
> a list. We can solve that need while still maintaing something that
> looks (at least a bit) like a list for those that want it.
>
> Note: please do give me a chance to sort out this email mangling mess
> before we conclude threading in list-mode is broken. As it looks right
> now, it's not Discourse's fault, it's Mailjet's fault.
>
>> What I find appealing about the forum-like approach is being able to
>> categorize threads based on topics. This can have a potential of
>> bringing attention to the right folks when needed. So far, I haven't
>> seen if the Discourse can actually be used effectively for that:
>> let's give it a try.
>
> Obviously we can create entirely new categories as & when we choose, so
> I'll assume you're looking for something more fine-grained within a one
> or more categories.
>
> There's a number of ways of doing this, depending on exactly what you're
> after. Firstly posts can be tagged, and those tags can be used as
> notification types (so I can be notified on, say, "salt"-tagged
> threads). If that doesn't bring the right people in, then a mention in
> any post (either a group, @salt, or a person, @gwmngilfen) will
> also generate a notification for the appropriate people. Tags can't be
> applied to a post by email (there's a feature request for that[1]), but
> @mentions will definitely work in any format.
>
> Both tags and groups are set up, and for now everyone has rights to
> create new tags. Groups are admin-created, I made a few
> (hamburger->Groups to join), let me know if you want more groups
> created. Give both a try, but be aware that mailing-list mode already
> sends you everything. You will need to disable that if you want to
> test selectively getting some things and not others.
>
> There may be other methods too, depending on our usecase. Discourse is a
> Rails app, and many of us here know how to write Rails engines… [2]
>
>> Anyway, I don't think it's a silver bullet and all of our problems
>> just go away.
>
> Agreed, but it brings a lot of tools to help that we don't currently
> have. Like any tool, it must be used correctly, but we're really good at
> discussing that kind of thing. I think we'll be able to find
> features/plugins that solve needs as they come up - but first we need a
> platform that lets us be that flexible.
>
> I'm definitely for at least a transition/evaluation
>> period and once proven, we can decide on whether the benefits are as
>> good as promised or we want to re-evaluate: this is not resistance to
>> change: this is demage control.
>
> I prefer "risk mitigation" to "damage control", no damage has yet been
> done :slight_smile:
>
> What kind of trial are you thinking, particularly in reference to the
> problems I outlined in side-by-side testing above?
>
>> The bad scenarios are:
>>
>> 1. some of the developers community will refuse contributing to the
>> forum and insisting only on mailing list: this is really bad and if
>> anyone plans doing so, please speak up now
>
> This is true, it's bad. But we must accept that we are highy unlikely to
> get 100% agreement on this. If there's a significant majority in favour
> of moving forward, we should. Aiming for 100% agreement on such a big
> change is unrealistic and not something we've done on other
> controversial decisions, either.

I'm not suggesting 100% agreement, on the other hand I'm serious about
listening carefully to the people that actually ARE active in the community.

Anyway, I think it's clear now that I'm for the side-by-side period, and I'm ok
with having both ways around. It seems to me like the best way to avoid
crystal-ball guesses and assumptions, and focus more on the data. And honestly,
if people will not start using it just because there is still mailing list, than
there might be some wrong assumptions to begin with. Otherwise, we will
have the data and everyone will be ore confident about the change, even
when not agreeing 100%.

– Ivan

··· On Wed, Nov 8, 2017 at 2:42 PM, Greg Sutcliffe wrote:
  1. we would have the forum as the only way to discuss things and it
    would make the discussion harder, not easier, for whatever reason:
    we can’t know that until we try IMO

We can eliminate 1. by having this discussion and not rushing to the
decision and 2. can be eliminated by running the two side-by-side.

I’ll repeat my dislike for side-by-side in a single channel, that’s not
going to work, at least long-term. As stated, I can be persuaded to do
-users separate to -dev if there are significant votes for that.

Greg

[1] Add tags by email - #7 by mcwumbly - feature - Discourse Meta
[2] plugin - Discourse Meta

–
You received this message because you are subscribed to the Google Groups “foreman-dev” group.
To unsubscribe from this group and stop receiving emails from it, send an email to foreman-dev+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

36

> And, if we meet in some time (let's say 6 months from the kick off)
> and look at the numbers, I think it would be much easier discussion
> if we should ditch the list or not. Additionally, we would have 6
> months of hands on experience with Discourse.

I'm going to assume we're talking about side-by-side for a single
channel (i.e the users list). I'll focus on that, but I remain open to
the option of running -dev on a list and -users on a forum, we can make
that work. Still not my top choice, but doable.

So, I'm all for following the data - in fact I'm actually studying Data
Science at the moment in my spare time so I can be better at this for
our community metrics in general (this course [1], good fun).

What I think you're proposing is that we measure the amount of activity
on both platforms after 6 months, and then use that as an indicator of
how much people like each platform. The problem is that this
experiment contains both systemic bais and a confounding variable.

Firstly, the systemic bias: people will stay where the conversation is
(i.e. the network effect). Unless everyone moves, no-one does. If we had
zero users on both platforms at the start, this could probably be
accounted for, but that's not the case.

Secondly, the counfounding variable (nemesis of all data scientists).
You're suggesting that "amount of activity" on a given platform (X) can
be used to infer "willingness to use" that platform (Y). But this
doesn't account for the procrastination problem (there are studies, I
picked a couple [2,3]). People don't change if they don't have to,
however much better the alternative is. So there's a variable affecting
X but not Y that we can't account for, which means we can't use it for
inference.

> I'm not suggesting 100% agreement, on the other hand I'm serious
> about listening carefully to the people that actually ARE active in
> the community.

100% agree with that, that's exactly what this thread is for. I'll post
that summary I mentioned shortly to try and loop some more voices in.

Cheers,
Greg

[1] https://www.coursera.org/specializations/jhu-data-science

[2] Opt-in vs opt-out organ donation - much higher donation rate with
opt-out. People could save lives by filling out a form, and they don't.

[3] Electricity costs. 14 million houses in the UK could be saving £200
per year (2016 data), but they don't switch. UK is actually
considerating putting automatic tariff switching into law because people
are so bad at this.

··· On 08/11/17 17:01, Ivan Necas wrote:
37

> Greg, I absolutely understand the motivation,

Good. However, we're not the only two people on this list, and
over-communication is key to reducing misunderstanding in remote
relationships. You know me pretty well, but others do not, so I lay my
position out clearly to benefit them.

> Do not put me into position of blind and angry dev who can't accept
> something different or new.

I have not said that. I praised your first email as a model of rational
debate, proposing a alternative route with clear arguments. Following a
deeper dive into the pros & cons of each solution, you now appear to be
getting defensive. Let's not derail a productive discussion.

> every two years amount
> of programmers doubles. That is a crazy amount of newcomers. But these
> new people are not idiots and some technical level is required even
> for soft roles in our community.

Sure, they need some technical knowledge. They need to know Git & have a
GitHub account, probably a Redmine account, maybe they contributed on
Transifex, maybe they even set up a development environment (non-trivial
for sure). There's probably more, and thats before we consider plugins.
Even adding to our calendar needs you to use GitHub.

I don't think a requirement to join a mailing list proves any further
knowledge on their part. In other words, some barriers to entry serve a
purpose, but where we can remove them, we should. To do otherwise means
we lose potential contributors.

We should also be open to ways to help educate people with the knowledge
needed to participate - I get the feeling you expect people to
self-educate before they come to us, but my own history in this
community started with a completely innocent email to Ohad about how to
test something. We should be open to tooling that makes such teaching
easier.

> And we can make lists approachable very much like forums.

No, I don't think we can. We've been over this in the preceeding emails.
We can give an archive a shiny GUI, but fundamentally they have a
different feature set, and a different target user group. Your position
on Discourse is clear, and recorded here for all to see. If enough
people agree with you, it won't happen.

Greg

··· On 03/11/17 18:29, Lukas Zapletal wrote:
38

Thanks for the testing help, Martin.

  • for some reason the threads are not kept together in my Gmail and the
    messages from one thread are split into multiple threads even if they
    seem to have same subject. I'm not sure why, it may be because I tuned
    the account settings. I'll keep testing this

I'm seeing this too, and it appears to be specific to the very first
email that starts a thread - the rest are threaded underneath. I suspect
some interaction with the Sent folder and also BCC-to-myself which I use
quite liberally at the moment. I don't see this in my GitHub folder (see
my other email for the relevance) but that's probably because we don't
start PRs by email…

> - it took about 15 min since I sent mail to the time I received it
from the list (not sure what are the reaction times on the list today
but this won't improve it)

There's a 10 minute poll time on the incoming mail inbox, so that's
likely the source of the delay.

> - mails from Discourse take too much visual space - the footer saying
how to unsubscribe, reply or visit the topic is included in each message.

Agreed, these are all templates that can be altered. I'll leave it for
now so others can see the defaults, and if we go ahead we can alter to
suit our tastes. All such styling will need sorting before any migration
can happen.

> - "likes" are only indicated in forum notifications but not in
emails. If you send '+1' to the list the like is added but no message is
sent to the users (just the notification)

That's true - since Likes are intended to specifically to not clutter
the topic with just agreements, the post isn't recorded. I'll open a
discussion upstream about it (maybe we can get a setting for that) but
for now, it's one extra character to send "+ 1" as a workaround (I just
tried that).

Greg

39

And this is what happens when you edit drafts as new messages. You don't need fancy software to break threading, just a silly mistake will do :slight_smile:

··· -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
40

Update on threading…

> Thanks for the testing help, Martin.
>
> * for some reason the threads are not kept together in my Gmail and the
> messages from one thread are split into multiple threads even if they
> seem to have same subject. I'm not sure why, it may be because I tuned
> the account settings. I'll keep testing this

From the testing Martin did, and Ewoud supplying the headers he got
without participating, we were able to track down what we think is
happening [1].

We're trying out a hacky patch to fix it, and in a 2-email test I just
tried, it looks like it's working (at least, they were threaded when the
same test yesterday was not). More testing required please :slight_smile:

I've also opened this upstream to discuss, we'll see what they say.
Discussion and patch at [2] if you want to follow along.

Greg

[1] https://pastebin.com/u0cDzWbW
[2]
https://meta.discourse.org/t/threading-for-email-only-topics-seems-broken/73523

··· On 06/11/17 11:43, Greg Sutcliffe wrote: