search

r-spatial / discuss

a discussion repository: raise issues, or contribute!
54 stars 12 forks source link

RSpatial communication places #51

Open Robinlovelace opened 2 years ago

Robinlovelace commented 2 years ago

I know there is a cost to new platforms but having multiple platforms to chat can encourage diversity. People who are more used to using chat platforms such as Discord may be more likely to ask a question and get involved if there's an easy option to do so, compared with the current options (StackOverflow, GIS stack exchange, r-sig-geo, RStudio community and GitHub although the last isn't really designed for Q&A notwithstanding the Conversations feature).

In any case, I open this issue with a proposal: that we set-up an R-Spatial discord server, and we already have a proof-of-concept that is free to join and in my experience easy to use compared with other chat platforms (notably Slack and Gitter): https://discord.gg/Te3gWeDwmf

Context: @Nowosad and @jannes-m set this up thinking of getting more people involved in the 'Geocompr book' and the wider community but then thought the effort may be better spent building a chat place for the whole R-spatial community. I've seen examples where Discord is a large part of open source ecosystems, e.g. https://github.com/logseq/logseq . I'm not suggestiing any 'switch' in comms just another option for people who like this way of communicating (I expect it will be one platform too many for many people here), especially beginners. All we need to make it work is moderators who can be assigned and, in fact, that could help reduce the barrier to entry to the r-spatial world. A neat thing about Discord is you can see how many people are online.

discord

Interested to hear what people think and happy to take next steps if it's a general :+1: . Thinking of geocompr + transport channels in there if the plan flies, could also imagine sf, raster, ... other channels but I may be getting ahead of myself! Just an idea, hope it's of interest and think it has potential on the community building side of things, not least in relation to the OSGeo links (on that note just checked and found an unofficial QGIS discord group with ~50 members).

rsbivand commented 2 years ago

Who owns it, what are its licenses, is it open source, are posts externally searchable? GH issues are bad enough, as they cannot IIUC be found by web search, so trapping things inside someone-elses currently popular software privatises them in a bad way. This affects most of these fashionable platforms, I think mattermost is the same.

Blogs and SO are searchable, and the mailing list framework is searchable and readable (through archives) without subscribing. R-Ladies went with slack, but there was a good deal of discussion about whether that was a sensible choice. Nabble has all but died, but mailing list technologies continue to function. My clear preference is for text-only mailing lists, because 1) carbon footprint (servers are as bad as diesel) 2) open source code 3) open archives 4) text not visual, 5) no likes (being popular certainly does not make answers reliable), 6) but learning curve requires weaning potential users off GUI.

florisvdh commented 2 years ago

@Robinlovelace I hope you don't mind I make a sidestep in this discussion, about the challenge of 'staying tuned to various web-based channels'. I recently encountered a way for the desktop that I thought others may find useful. (There may still be other solutions.)

one platform too many for many people here

It is challenging to manage multiple accounts on different fora, let alone staying notified about new messages, either by keeping browser tabs open (and auto-open them on launch, or using specific browser extensions), or installing forum-specific desktop clients (which should then ... all autolaunch?), or limiting it to the smartphone. Further, browser tabs sitting side-by-side are potentially spying your habits in other tabs, although that can be remedied with e.g. Firefox container tabs nowadays.

Only weeks ago I discovered Rambox (community edition) and this solves most of these problems. It's an Electron-based desktop app, with an open-source community edition (https://github.com/ramboxapp/community-edition), it keeps the respective web pages open (and remembers that you logged in), while keeping each web app in complete isolation and presenting those in a desktop-like manner (see screenshot). It supports a large number of messaging services, in specific ways. It has an icon in the system tray (remains there after closing window) that becomes red if you receive a message.

See also 'How does Rambox work' and https://github.com/ramboxapp/community-edition/releases for all types of installation package (including debian packages).

A remaining downside is the RAM occupation of the concurrent web apps (currently around 500 MB in below case...). But when needed, one can quit Rambox.

afbeelding

Robinlovelace commented 2 years ago

Who owns it, what are its licenses, is it open source, are posts externally searchable?

It's owned by Discord Inc, it is proprietary although they have a decent amount of open source code at https://github.com/discord , and it's posts are not externally searchable as far as I can tell.

I agree with points about carbon impact, the importance of open source platforms, and search-ability for definitive answers (but not tentative questions from beginners which is the use case I have in mind). I was thinking that a friendly R-Spatial chat place could be a way to sign-post new users to the more established fora. Use case: someone has a question about a particular pacakge but no reprex. They post a question on the chat, someone with more experience points them to the docs, and the problem is solved. Or they uncover a bug and get feedback on creating a reprex, e.g. using data in the spData package. Just ideas. In terms of platforms that are free, open and searchable, Discourse is the best I know of, and is the basis of the RStudio Community I believe. I would advocate that as a platform but for the resource involved in setting it up. Discord basically makes it very easy to set a place for real time chat for those who like that way of communicating (with the understanding that many will not). One thought that may be attractive to experienced developers: a new platform for comms could take some of the communication load away from GitHub issues. The pathway for a new user could be something like:

Something seems to not work -> Ask a quick post on a platform with a low barrier to entry -> Get feedback that directs them to a solution

I'm not sure that ideal scenario will happen but I think it's better than another common scenario:

Something seems to not work -> Person does not know how or where to ask, or does not have the confidence to post in a public forum like R-SIG-GEO or Gis stack exchange so gives up

I also think it could in some cases be better than

Something seems to not work -> Open an issue on the relevant issue tracker taking valuable developer time

I think it's worth exploring alternatives. As I see it these are options:

It's interesting to see the list of chat options for the OSM community, I've also heard there is a big community of OSM contributors/developers on a Slack: https://wiki.openstreetmap.org/wiki/Matrix I notice that OSM community has a diversity of ways to communicate, including Matrix https://matrix.to/#/#osm-space:matrix.org Discord https://discord.com/invite/openstreetmap , Slack https://wiki.openstreetmap.org/wiki/List_of_OSM_centric_Slack_workspaces and of course mail servers. The point is that diversity of platforms can encourage diversity of contributors. Based on what I've seen Discord seems a low-barrier-to-entry and quick-to-set-up option that is slightly better than Slack and I don't see negatives associated with it, assuming there is demand from the R-Spatial community. If not I do think it's worth thinking about alternatives. Simply better signposting to existing decent communication options would I think be a non-controversial quick win but I also think the +s of another platform would likely outweigh the -s. Also interested in what others think, anyone had good experience with Matrix? Could be a good option but would take some setting up I guess based on my limited experience with it.

Looking at the list above I think GitHub Discussions could be the lowest hanging fruit in this space, instead of or in addition to other options that can be explored in parallel.

Robinlovelace commented 2 years ago

I guess the key questions for the community are:

Probably best to gauge interest in that before focussing on which platforms, multiply platforms can solve the same problem.

edzer commented 2 years ago

Another question is who is going to maintain it, to moderate it (spam, managing accounts), and will the contents remain accessible and searchable to the community in the longer term. I got used to mattermost during the OGH summer school last week (and earlier years), and OGH might be capable and willing to set up and maintain a public mattermost for this purpose. Mattermost is open source, so there is no vendor lock in AFAICS.

Robinlovelace commented 2 years ago

Mattermost sounds like a good option, especially given prior experience with it. Moderator roles can be assigned I think looking at the docs:

There are six types of user roles with different permission levels in Mattermost: System Admins, Team Admins, Channel Admins, Members, Guests, and Inactive accounts.

From what I've seen on successful communication channels, having moderator (Channel and Team admin I guess) roles allocated to people who have been around for a while works. Another potential + of mattermost is that people who get used to it at the OpenGeoHub Summer School could continue to use it so :+1: from me on that suggestion.

rsbivand commented 2 years ago

Yes, if something is already running, there is some experience to be shared. I do not think that synchronous systems (chat) make any sense, because most suggestions take at a minimum several minutes to check (they often take hours or days because one needs to deconstruct the questioner's context from fragments of often irrelevant information). Too low a threshold removes the incentive to raise one's skills for questioners. Probably a short description (or flipped-style video) explaining how to frame a question, how to search for existing answers, and how to make a reprex from the example on a function help page would be less effort than administering yet another channel. Or at least it should be promoted as what to do first, before posing a question on a channel. I was/am on MM from OGH, but only used it at the 2019 meeting, so do not follow or post, but it would be more sensible than alternatives. The key distinction is asynchronous vs. synchronous, where the former works but the latter largely doesn't (time-zones, finding solutions takes time, people have other things to do).

mdsumner commented 2 years ago

An open chat would be good I'm into it. Let folks who can use it positively do so doesn't have to be a consensus about what should be :)

Sometimes you have to use the old mailing lists as many folks won't pay attention anywhere else (that's fine, that's the terms of the project ...). Twitter is good but I think people hold back a lot as it's completely public. There was a great series of chat rooms on SO but they got flamed out by toxic players. The old "#r" IRC was really good too but I think that era is gone.

rsbivand commented 2 years ago

OSGeo use Libera for chat (which we might join if we join OSGeo), but maybe this has too high a threshold for users, being more aimed at developers, I think. https://libera.chat/about/

Robinlovelace commented 2 years ago

+1 for Libera Chat also, great to see how quickly the community switched after "Freenode staff resigned en masse".

image

Source: https://en.wikipedia.org/wiki/Libera_Chat

TBH I've come around to being in favour of open solutions for R-Spatial. Geocomputation with R was always aimed at getting people started, a stepping stone for people struggling with ASDAR as we saw it in early conversations. With that in mind my plan is to use the Discord channel above for discussion of the geocompr book, which can be seen as kind of mini side community supporting r-spatial. Hoping we can 'onboard' new r-spatial members and developers through that community and planning to point people towards r-sig-geo and repos on this GitHub org. Great to see how ideas are discussed and thinking moves forward, an indication that this is a strong community to me!