
Hi all. Let me try to sum up the state of the discussion and propose next steps -- please feel free to disagree with my summary if you are so inclined. I'm trying to capture the current state and help move discussion, so if people feel this doesn't capture it, its good to know sooner rather than later. == Key Points == * We distinguish between "download method / installer" and "getting started". We add a new page "Getting Started" with links at the top level just like "Downloads / Community / Documentation / News". * The Downloads page points to the minimal platform installer (however named) that includes GHC, stack, cabal-install, and a few other binaries (but no extra global libraries). Below that we can have links to other mechanisms as well, but the minimal platform is the clearly distinguished recommended way. * The "Getting Started" page proceeds to describe next steps for new users to get up and running right away after installing the minimal platform. Since many people have made the case that for someone who hits the page without some prior guide, a stack workflow is the easiest to get running on and best documented at this time, I imagine this workflow will be stack-oriented. == Additional Considerations == * People have pointed to some issues in the windows platform installer. We are going to work to resolve these. A patch for one (which improves stack behavior when it can find ghc but not msys) is already submitted. Sorting out the other (which pertains to enterprise installs, not single-user installs) is underway. * As in the past, we shouldn't focus our discussions on what the answer is "for all time" -- tech will change, things will evolve. We should think of this as something that does the job reasonably well for say the next six months, give or take. == Next Steps == * There is a mock-up of a new download page already on github (Thanks so much Jacco!): https://github.com/haskell-infra/hl/issues/176 Under its "getting started" I would propose something like: 1. Download and run the Haskell Toolchain installer 2. Follow the _getting_started_guide_ to start running some code! with the latter linking to the new getting started page. I hope this is acceptable to people? Maybe some other bullet points should be there too... * I would like to solicit a kind volunteer or two to draft up what they think a "getting started guide" should look like so we can spawn a separate discussion on that. Given that we've decoupled this guide from the download page, it can afford, I think, to have a bit more text than just how to compile and run "hello world"... == a few other thoughts == We've had a productive discussion on this list, I think. But as many probably know, not all the discussions over this have been so positive. I just want to make a general point that has been nagging at me, and apologies for the sappiness that follows. While there is some commercial support, Haskell is an open-source language and community at heart that historically has and currently still relies on people volunteering their time. Be it in writing code, writing documentation, teaching others, and especially in working on tooling and infrastructure. (This goes all the way up to GHC which also runs on donated time and effort in many ways). Being run on volunteer steam is great in many ways. But it has one important caveat. People have to want to volunteer, because they enjoy the technical challenges, but especially because they want to help others, and because they find it rewarding to contribute to things which bring joy and benefit to others. All of us, working on installers, working on fundraising efforts, working on documentation, working on compilers and libraries, or trying to sort out weird issues with mailservers and proxying content delivery networks or php bugs impacting wikis or google-analytics tokens or whatever else, we're all here not because we have to be, but because we want to be. And that's great! But it is also precarious and fragile, because if it becomes not fun, or not rewarding, or we feel that what we do is being disparaged and attacked (and this goes for everyone, i think, on all sides of recent discussions) or diminished, then we change our calculus, and we stop wanting to participate, and stop wanting to fix things for others. And that's understandable. So, if we let our discussions or language play out in a way that can hurt others, can make them feel that their work is no good, or not appreciated, then they have no reason to want to help anymore. And they won't. And I can't blame them. And if we all start to hurt one another too much, it will become no fun for any of us, and then the whole endeavor falls apart. So I want to extend my gratitude and thanks to everyone working on any element of our open-source tooling and infrastructure, or even lending a voice to give insight into how to improve it. And I also want to ask us all to remember this in our interactions with one another, and to remember that people will say things we find unbelievably wrong at times, and discussions may stall at times, and things may get stuck for longer than we'd like. And we have every right to get frustrated when that happens. But we can't let our frustrations at difficult _situations_ turn into words that diminish other _people_, other potential collaborators, others also freely giving of their time trying to improve and help others, whatever our disagreements. I have more to say, but I'm not sure how, and this is long and inarticulate enough. Anyway, onto the next steps (I hope), and please everyone consider stepping forward to help with this stuff, in drafting language, or helping with design and implementation of the actual changes to the site (on which more volunteers welcome -- please email me!), etc. (By the way I want to put out more calls for help on more concrete infra stuff soon, not at all web related -- but in the immediate future, if you have experience running a mail server and dealing with the many painful hassles therein, and are willing to help a bit, please contact me soon). Best, Gershom

On 9/1/16 2:12 PM, Gershom B wrote:
We've had a productive discussion on this list, I think. But as many probably know, not all the discussions over this have been so positive.
+1 I am really happy about the positive, conciliatory tone the discussion has taken in the past few days. I had been disheartened by the schism that has grown in the Haskell community over the past year or so. As a user, I was especially turned off by packages whose sole description was essentially "go to Stackage if you want to read the description": https://hackage.haskell.org/package/http-conduit-2.2.0.1 I am a happy user of both caball-install and stack, and ideally I'd like to see the two ecosystems work more closely in the future, but at the very least, I'd like to see them coexist peacefully and respectfully, and not have individual packages "take sides."
So I want to extend my gratitude and thanks to everyone working on any element of our open-source tooling and infrastructure, or even lending a voice to give insight into how to improve it.
In that vein, is there a place to file bugs against HaskellWiki? Or is that one of the areas in need of a volunteer? My own personal itch I'd like to scratch is the lack of diffs on the wiki: https://mail.haskell.org/pipermail/haskell-cafe/2016-April/123701.html Many thanks to the Haskell Committee, to FP Complete, and to all of the volunteers who work on Haskell! --Patrick

(changing the subject line to reflect this thread) On September 1, 2016 at 7:48:46 PM, Patrick Pelletier (code@funwithsoftware.org) wrote:
In that vein, is there a place to file bugs against HaskellWiki? Or is that one of the areas in need of a volunteer? My own personal itch I'd like to scratch is the lack of diffs on the wiki:
https://mail.haskell.org/pipermail/haskell-cafe/2016-April/123701.html
Many thanks to the Haskell Committee, to FP Complete, and to all of the volunteers who work on Haskell!
For issues with the haskell wiki as well as all other infrastructure problems in general, the admin@haskell.org mail alias hits a group of infra maintainers. A github tracker that’s more widely advertised as well might not be a bad idea, since its become so standard. Over the years our current infra/ops team has gotten rather busy in other arenas, so we could use volunteers nearly everywhere (anyone with any sysadmin experience, even if just on a personal box, and who feels they’ll have ongoing time to donate, please email me! and again, anyone who can help deal with mail stuff would be especially appreciated at this moment). We keep things humming, but haven’t really freed up the time to tackle much in the way of pushing things forward. But you’re correct that the wiki could absolutely use a volunteer. I made a bit of an effort last year to get things rolling, but it stalled out pretty quickly (https://mail.haskell.org/pipermail/haskell-cafe/2015-April/119219.html). Perhaps the call was too ambitious and people didn’t know where to start. Maybe it just needed more energy and will to drive it than any of us had in supply at the moment. We _did_ get a team of people approving new accounts instead of just one overworked individual. (Thank you to all who have done that and continue to! It irks me that we need such a manual process, but the waves of spam we got before periodically despite our capcha efforts were far worse). I can’t quite infer from your message if you are volunteering yourself. If you are, thank you so much for your offer! Please email me and we’ll coordinate how to get you the right login credentials, etc. Best, Gershom (Note, by the way, those who may be hesitant to volunteer not out of lack of time or desire to help, but out of fear you don’t have enough experience or knowledge — most admin work [to be honest like most programming] isn’t knowing what to do right away, but instead about knowing some desired end goal, and then being willing to wade through a bunch of searching and documentation to come up with a plan, talking it over with folks, perhaps experimenting a bit, and eventually finally knowing what to do. And, of course, knowing which experiments are irreversable and shouldn’t be conducted :-). And the hardest step usually isn’t in fixing any problem, but in making the mental leap from “this is a problem someone should fix” to “this is a problem I will fix”.)

On 2 September 2016 at 01:48, Patrick Pelletier
wrote:
On 9/1/16 2:12 PM, Gershom B wrote:
We've had a productive discussion on this list, I think. But as many probably know, not all the discussions over this have been so positive.
I am really happy about the positive, conciliatory tone the discussion has taken in the past few days. I had been disheartened by the schism that has grown in the Haskell community over the past year or so. As a user, I was especially turned off by packages whose sole description was essentially "go to Stackage if you want to read the description":
One can debate that description, but the issues with docs on hackage have been ongoing for a few years. I'm sure any help would be appreciated. One pointer to get started is https://github.com/haskell/hackage-server/issues/478. -- Paolo G. Giarrusso - Ph.D. Student, Tübingen University http://ps.informatik.uni-tuebingen.de/team/giarrusso/
participants (3)
-
Gershom B
-
Paolo Giarrusso
-
Patrick Pelletier