Threads for luke

  1. 1

    Are there any associated rooms on the matrix instance? I haven’t been in any Haskell-related room since I left the freenode bridged #haskell room.

    1. 1

      Just getting started so you can create whatever rooms you want. Nothing is currently bridged.

    1. 6

      After the whole Freenode debacle, I was motivated to set up an experimental web, Matrix, and Pleroma set of services to give the Haskell (and Haskell-adjacent) community an alternative to centralized Meetup, Slack/IRC/Discord, and Twitter, etc.

      Matrix and Pleroma certainly have their downsides - IRC and Twitter are pretty hard to beat - but they have their upsides as well and I figured it’s a fun experiment at the least. I won’t be leaving IRC any time soon. Feel free to join me and/or contribute if you’re interested. I’m always up for hearing feedback as well.

      1. 3

        Have you set up bridging between IRC and Matrix? That’s what we do for the Fennel language channels, and it works pretty well other than occasional lag.

        1. 1

          I haven’t yet but was looking into it. Any recommendations?

          1. 3

            I don’t really use Matrix myself (never found a client that felt as nice as my IRC client), but I’ve found it to be a useful way to broaden the appeal of the chat and make it more accessible to “kids these days”. Of course since I’m lazy I just created a channel on the flagship matrix.org server, which is bad because centralization is bad, but I don’t have the bandwidth to run my own server for a service I don’t personally use.

            Anyway it works pretty well other than occasional lag. The other thing to watch for is that message edits from the Matrix side get represented in a really disruptive/ugly way on the IRC side. We just put a “no message edits” warning in our topic but I really wish it could be disabled entirely on a per-channel basis. Those are basically my only complaints.

            1. 1

              I’ve had the same experience regarding Matrix clients. There seems to be a lot of work going into the client side thought. That said, mobile apps, persistent chat history without the need to run your own bouncer etc. for me really makes Matrix more of an alternative to Slack or Discord than IRC.

            2. 2

              You might want to look into https://gitlab.com/kazarma/kazarma

              1. 1

                Excellent - thanks for the rec!

          2. 2

            Why pleroma over mastondon, glitch? Just system hosting requirements?

            1. 6

              Yeah exactly. Pleroma is the lightest weight and the easiest to maintain for me personally.

          1. 3

            Using the edge router sounds neat, but one of the Amazon commenters said the default firewall rules are pretty basic. Did you add a lot of rules or do the default set seem to suite your needs?

            1. 1

              Yeah it’s definitely not meant to have a ton of preprogrammed rules. It comes with a basic drop all incoming but that’s kind of the reading I got it.

            1. 3

              This is neat! Do you have any logging / monitoring systems in place?

              1. 2

                Thanks! -Yes, everything has Prometheus metrics and Grafana for monitoring. For logs specifically I just log into the box and look at journald.

              1. 3

                @luke you mention a mesh wifi. Is that provided by any of the components you list there, or is it something you built separately before this setup?

                1. 5

                  It’s the Plume mesh wifi. I didn’t mention it directly because I didn’t want to promote it. I had that first then built the rest of the network around it to isolate it out and prevent/limit it’s data exfiltration. But you could use any wifi AC you want. If I hadn’t had that first, I would probably use a Ubiquiti AC. The costs seem to be about the same but it seems so much more robust.

                  1. 3

                    I was wondering specifically about the data exfiltration from the wifi hardware itself. I’ve got Orbi right now, which works great except that it phones home to Netgear all the time.

                    Oh yeah, and the admin interface now shows me ads for other Netgear products, with no apparent way to turn them off.

                    I hadn’t heard of Plume before, but a quick look at their website says they market to ISPs with remote control as a major selling point. So I can totally get why you didn’t mention them before!

                    1. 3

                      Yeah the Plume is the same. It auto-balances the mesh network so if I cut it off from calling home, it has no way for me to manually lay out the network.

                      Plume actually has a “privacy” mode which I use and they seem to honor. But I’d prefer to know for sure. I don’t know of many consumer mesh wifi systems that don’t phone home.

                1. 3

                  This is a really great setup! Is there anything you are looking to add soon?

                  1. 3

                    Plex is pretty bad for user privacy and surveillance so I’d like to give Jellyfin another shot. It fork-bombed my server when I just did a basic install the first time around.

                    1. 2

                      I’m a mostly-happy user of Jellyfin - there are a few quirks involving music tagging I’ve run into, but otherwise it works pretty well at letting me stream my own music and videos from my media server to any computer with a web browser or a smartphone.

                  1. 4

                    You might be interested in my creating a completely Tor-ified home network article if you want to add mixnets into the picture. :)

                    1. 1

                      Awesome - thank you. Yeah TOR has definitely been on my list.

                      1. 1

                        Sorry for piggybacking on @luke post but I’m still somewhat a newbie when it comes to Tor. Can you direct me to some resources regarding (or elaborate on if you’re knowledgable enough) law & responsibility and how safe it is to run Tor public relay node? I mean scenarios like this: how liable I am if some pedo looks through child pr0n through Tor and I end up being an exit node? I guess it varies from country to country but still. Thanks.

                        1. 4

                          and I end up being an exit node

                          You’d have to volunteer to be an exit node. It takes explicit configuration to do so.

                      1. 4

                        Company: Bitnomial

                        Company site: https://bitnomial.com

                        Position(s): Software engineer, operations engineer

                        Location: Chicago, IL No remote unfortunately

                        Description: We’re building a Bitcoin derivatives exchange all written in Haskell. We use ansible and terraform for our operations automation. Trading industry experience is a plus.

                        Contact: careers@bitnomial.com

                        1. 2

                          Check out some of our open source projects at https://github.com/bitnomial

                          1. 1

                            maybe page @jcs to merge?

                            1. 1

                              Oops good catch. Thanks. @jcs merged them.

                            1. 1

                              Where’s the source code? The page had no links.

                              1. 1

                                It looks like the source code is only available through Git. In “Getting started”, the page says you can run git clone http://git.coders.se/kore.git to download the code. (As that link reveals, http://coders.se/kore/ is an alternative URL for https://kore.io/.)

                                I would mirror the code on GitHub so you could browse it, but neither the page nor the code contains a license yet. But here’s a preview of the file tree.

                              1. 2

                                I like it. Do you think little voting arrows are an artifact of a mouse and not very touch friendly? I know you can swipe to vote but what about down and unvoting? All in all great job though. Can’t wait until it hits the App Store. You planning on free?

                                1. 2

                                  Swipe right to upvote or undownvote, left to downvote or unupvote maybe? Long tap to send to instapaper?

                                  1. 1

                                    That’s one way I’m thinking of approaching it, Clear and Twitterrific 5 have made me a huge fan of using those swipe gestures to act on an item. Another approach would be the Tweetie 2 style interaction where you slide a cell off to the right to reveal buttons behind it, but I’ve never been a huge fan of that interaction.

                                  2. 1

                                    The arrows here do still serve a purpose in letting you know whether you’ve voted on a story, but tapping them does nothing. I’ve been trying to come up with a better way to show users they’ve voted on a story that’s better suited to the touch-based context with no luck so far. I’m curious if anyone else has a good idea how this could be handled.

                                    Also as far as price I’ll probably charge $1-2 for it although the project will remain entirely open-source.

                                    1. 2

                                      I hate to mention BaconReader again, but it solves this problem by lightly coloring the background of voted stories. Light orange is used for a positive vote, and light blue is used for a downvote.

                                      1. 1

                                        I think that works well for Reddit where those arrows already have a strong correlation to certain colours, but here both upvote and downvote arrows are red. Maybe a very light yellow background on stories you’ve vote up and faded text on ones you’ve voted down? I think that would be enough to visually emphasise/deemphasise them. I’ll try mocking it up later.

                                        1. 3

                                          Okay I quickly mocked it up, here’s what that might look like, the arrows are still there because I didn’t have to time to go through and delete them all but I’ll probably get rid of them entirely now.

                                          1. 1

                                            Nice! I like that

                                  1. 1

                                    This is a must-read for any Haskell dev as well as interested compiler designers. Compared to other compilers I’ve seen, GHC has some pretty interesting design choices like “everything is a Haskell library” and “there’s no symbol table”. Though long, it’s worth the read.

                                    1. 1

                                      What about changing the Lobsters “economy” a bit to promote the type of community we want?

                                      1. Comment votes (up and down) count for more karma than story votes. Promote discussion.
                                      2. Karma per story/comment gives extra abilities. Promote contributions.
                                      3. Absolute-karma and karma-per-story/comment are both shown side by side to promote quantity and quality of contributions. Promote quality.
                                      4. Meta stories/comments votes don’t count towards karma (but still can move the comment to the top of the list). Promote normal discussion.
                                      1. 7

                                        I prefer the current “if you can’t find a tag for it, it probably doesn’t belong” situation.

                                        1. 4

                                          How should I tag something about Haskell? Last I checked there was no tag for that. Meanwhile, the “science” tag seems like quite a catch-all that doesn’t help distinguish topics.

                                          1. 2

                                            Agreed. I think the real issue is currently the finite set of tags is being used as a filter. Instead, if we focused on tags just being metadata, tags could provide more value. I realized my main use for tags is not filtering but I just like them for visually grepping the stories.

                                          2. 3

                                            I think that’s what the voting is for: making sure the stories are things the community wants to read.

                                          1. 2

                                            The main use-case here would be a ton of stories being posted about a new topic I don’t care to read about. The current process would the new topic to be recognized as needing a tag and requested by someone. We’d discuss and hopefully an admin would pick it up. User-defined tags, on the other hand, with the auto-lookup feature would allow this process to happen naturally and in real time. Maybe even allow other readers to add tags to stories after posting.

                                            1. 1

                                              Thanks for the comments :)

                                              It’s interesting that there seemed to be many people on HN clamoring for an invite, and now it seems that they mostly just wanted Yet Another News Source. But you can get that without being registered, so what’s the point?

                                              I suppose the next meta question is how do you encourage people to comment. And how to do so in way that does not promote bullshit participation.

                                              My own reasons for not commenting tend to be either a decent comment would take more time then I have available, or I don’t know enough to say anything useful. Or I’m reading on my phone and commenting anything of value is too tedious.

                                              1. 2

                                                IMHO comments will happen naturally. We’re all still getting to know each other. It took me a long time reading HN before I felt I had comments people may find useful. It will just take time.

                                              1. 1

                                                Anyone have any other recommendations for facial recognition?

                                                1. 2

                                                  Proposed tag: other programming language. For programming language specific posts that do not have their own tags (like javascript, lua, c, go etc have, and scala, haskell, ats etc lack). This saves adding tons of tags for more programming languages.

                                                  1. 1

                                                    What would the (short) tag be? “otherprog”?

                                                    1. 1

                                                      I would suggest ‘otherlang’, though that is getting long for a short tag.

                                                    2. 1

                                                      I’m not a big fan of negation categories. It’d be better to have a “PL” tag for programming languages and then an additional tag for the specific PL if it’s available.

                                                      1. 1

                                                        Okay, how about ‘PL’ for stuff about programming language design, and ‘lang’ about stuff specific to a programming language that is otherwise not in the list of languages?

                                                    1. 2

                                                      Proposed tag: haskell

                                                      Not sure if there’s enough interest but that’s what the voting is for =)