2026-08-17T22:04:09Z https://njump.me bryan newbold https://njump.me/npub1tull74k2f75pwlc39yg2xn6tve9k6g854yjxafuu4x9rqggcnwas6tf77h https://pds.robocracy.org/xrpc/com.atproto.sync.getBlob?did=did:plc:44ybard66vv44zksje25o7dz&cid=bafkreiglnysron3h2je7nf6cmvtimuaxi7xe2c7rkxitmks3mzmajnc2ou https://pds.robocracy.org/xrpc/com.atproto.sync.getBlob?did=did:plc:44ybard66vv44zksje25o7dz&cid=bafkreiglnysron3h2je7nf6cmvtimuaxi7xe2c7rkxitmks3mzmajnc2ou https://njump.me/nevent1qqsdsgyp2ey78eezqpdneftscgwp320z3h03sg8s4tul9cfdukdjjgczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkxf6pr7 I&#39;m hoping to discuss the current &#34;repository&#34; and &#34;synchronization&#34; drafts, and small changes that might be made to them. and the AT URI scheme syntax situation.<br/><br/>I think other folks have proposed talking about floats, and AT-over-MoQ 2026-07-02T20:11:44Z https://njump.me/nevent1qqs80p6f2qy233pf89590m2ulsl02wjnstr5h3lw57wsmdpy4hcz77qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkfpvvl5 devs working with bsky image embeds:<br/><br/>we are planning to bump the image embed blob max size from 1 MByte to 2 MByte. first in the lexicon schema, and then in the Bluesky Social app.<br/><br/>more about impacts, and what this means in terms of lexicon evolution, in this github discussion<br/><br/>Increasing Bluesky Post Image ... 2026-04-07T05:37:32Z https://njump.me/nevent1qqs0mnjs9v4uhr2xg0j8adtj2spema8t2zvcqwa9axgsgwee0d7tl0qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkc3juu9 just one novel cryptographic primitive in permissioned data.<br/><br/>a tasteful treat, an amuse-bouche, some fan service. 2026-03-13T02:16:31Z https://njump.me/nevent1qqsg2qpgaumvk7mfqvwvk39f8dcpqwq59fuat9t9aj3e50ev5kr905gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk4t0spd atmosphere != bluesky<br/><br/>and that has always been the goal 2026-03-09T20:26:20Z https://njump.me/nevent1qqspal9tvj7hff7gyxyzrzae90a0wxdnxnaqdsglz5zz3klw2n9j2gszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkzf5a9e what does this mean for the atmosphere?<br/><br/>one the one hand, not much: Bluesky isn&#39;t making any near-term changes, and retains long-term alignment on protocol+ecosystem success.<br/><br/>but it is part of maturation and growing beyond single orgs or individuals 2026-03-09T20:25:45Z https://njump.me/nevent1qqsgerg70jeuu63uplls67uu5qwuhe55ym6lvcewvy67hnkra4gdutszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkeux3vk Jay seized a historical moment and speed-ran Bluesky through some wild times. I&#39;m glad she&#39;s found her own way back to the big picture.<br/><br/>have had great conversations with @toni.bsky.team@bsky.brid.gy about standards work and open source governance. very important to have continuity of clueful leadership!<br/><blockquote class="border-l-05rem border-l-strongpink border-solid"><div class="-ml-4 bg-gradient-to-r from-gray-100 dark:from-zinc-800 to-transparent mr-0 mt-0 mb-4 pl-4 pr-2 py-2">quoting <br/><span itemprop="mentions" itemscope itemtype="https://schema.org/Article"><a itemprop="url" href="/nevent1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqxd0d9kdwfjsf3ugqpa9e3mnxf83akjmuj2e865c7axgf09nww28q9fzua4" class="bg-lavender dark:prose:text-neutral-50 dark:text-neutral-50 dark:bg-garnet px-1">nevent1q…zua4</a></span> </div> Some personal news: I’m transitioning from CEO to a new role as Bluesky’s Chief Innovation Officer! I’m excited to welcome @toni.bsky.team@bsky.brid.gy as our interim CEO. <br/><br/>More here: bsky.social/about/blog/0...<br/><br/>A New Chapter for Bluesky - Bl... </blockquote> 2026-03-09T20:11:41Z https://njump.me/nevent1qqsphyvra48jrlq3ly5kc36us8ekzehy0tr20san08xj670wh80wf9czyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkgdmt7n I just pushed a new version of goat (v0.2.3).<br/><br/>it includes a new &#39;xrpc&#39; sub-command for calling API endpoints in various ways. give it a try!<br/><br/>XRPC sub-command for calling H... 2026-03-08T05:57:48Z https://njump.me/nevent1qqsyqus585n7hghv8cnwwcl8ylwgltvw6vcezdcufdgm8d65yw5xy2gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkxdca82 *tweak haha 2026-03-07T19:28:15Z https://njump.me/nevent1qqs2thsqxyejr7gg27g74khzysprzwr6dqf6y9snh59u0lg74efe6jqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkk7a8ns &#34;what prevalent LLMs mean for social media&#34; is a huge/top question for the whole bsky team (and has been for a while).<br/><br/>I think there is some urgency to address current stuff, but also want to be careful setting long-term precedent in March 2026. my individual 2c 2026-03-07T19:20:47Z https://njump.me/nevent1qqstyel9ydygnssqrarxv2rh4sk478lu5wz7cxw6zu0e8945gh6tj5szyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk4el5e3 this hits a couple things I wish collectiondir did; maybe we can switch over to this implementation...<br/><br/>two other things that would be great, but might be hard: reflect account status (eg, don&#39;t show inactive accounts); and have support for wildcard (and/or the getRepos endpoint) 2026-03-06T01:12:09Z https://njump.me/nevent1qqs0adamkwyq5clyarsvz939c6azt7xa9mskahvasg76u3futcw0f3gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkpqt5x2 in the long run, if a website or app had something sharable, and that thing has an at:// URI, then browser/OS could use your preferences to open to relevant app to interact with that content.<br/><br/>for sharing a URL, I think something like a browser share API (like Emelia linked) would be good 2026-03-06T01:04:15Z https://njump.me/nevent1qqsfaye0zkmelxu4xnlnqfvhsrgm0fmvgj247hxs3vt6xgjgp6djehqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkk68fp4 the @xblock.aendra.dev@bsky.brid.gy and @stechlab-labels.bsky.social@bsky.brid.gy labelers are good starters for composable moderation. 2026-03-05T20:01:17Z https://njump.me/nevent1qqsv4e48fty2f08x82phr3ydv2xtjx7aus53pf0r3sw4drk44wepttczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkwrr8uw welcome @doctorow.pluralistic.net@bsky.brid.gy!<br/><br/>on your own PDS host (bsky.pluralistic.net), using did:web (did:web:pluralistic.net), and self-controlled domain handle.<br/><br/>looks like things are working/interoperating; let us (the atmosphere ecosystem) know if you run in to any issues 2026-03-05T19:55:27Z https://njump.me/nevent1qqsz4xfcuwa3k9c67ja6nj69c8pe540vl9r0dmm0gfqqtw30ghlsczszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk2qn769 Pierre Menard, Author of the 4chan 2026-03-05T04:50:16Z https://njump.me/nevent1qqstsl9tkstw6ygpphvwsxr0k09qy9r0j4yr5gsxakf940s23hjknyqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkrlj8hp emailing users: I don&#39;t know about this specific situation, but a lot of folks originally signed up on the bsky PDS instances, and by default migration leaves the old account in &#34;deactivated&#34; status. mod team may have contact info from that, if the account hasn&#39;t been fully deleted. 2026-03-04T19:57:49Z https://njump.me/nevent1qqs09wfl9tjynuuxqnh7hzkna9tkajwm9gewld4hkqgjrlmukxygxsszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkwarmxr there is a delay before new accounts show up in search, which might be causing confusion. I don&#39;t know what the current status of a profile record being required or not is; I think it is at least requires for profile search / typeahead. 2026-03-04T19:55:50Z https://njump.me/nevent1qqs2k4yxms5d0gkj456vny033gyux0g0qxku5suyfkmhqgmt9e463sgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkugd6uu I don&#39;t know who is running pds.0.space, but it seems like all LLM spam bots. could be intentional, or could be open invites. blocked at our relays 2026-03-03T20:53:41Z https://njump.me/nevent1qqspee5ef5hdt5g06466mn47f4yssqgfectm2v3ptj5l7v4xh0fqd6qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkfppn8n cc: @iame.li@bsky.brid.gy who is interested in MoQ for streaming video and atproto; different from firehose transport, but this is definitely in your area of interest 2026-03-03T00:21:44Z https://njump.me/nevent1qqs8gcsgdpq530x4r2an8pe6lvjjuwac6asr0za48x28u7dq5s7anpqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkf8eqzc but for that matter, the current full AT-URI string syntax allows a whole bunch of weird stuff, like at://did:plc:abc123/com.example.record (just an NSID, no record key).<br/><br/>a change like the blob one could happen at the same time as record versioning. I think the IETF process could be an opportunity<br/><br/>Record Versioning - at:// pizz... 2026-03-02T23:45:25Z https://njump.me/nevent1qqszx9apwlu2e004ncx0appawk5e2f469ttavnc9lux9e6wd6sun7hszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmksqvq5r this is huge! it&#39;s a big lift just to do the backfill indexing, but even more to get it operational and chase down all the little product affordances.<br/><br/>blacksky ships.<br/><blockquote class="border-l-05rem border-l-strongpink border-solid"><div class="-ml-4 bg-gradient-to-r from-gray-100 dark:from-zinc-800 to-transparent mr-0 mt-0 mb-4 pl-4 pr-2 py-2">quoting <br/><span itemprop="mentions" itemscope itemtype="https://schema.org/Article"><a itemprop="url" href="/nevent1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqmamq42fnp4qeg9nq0whe6ggfy8yz9nuz8lajuec47eamk0084cwqhhts42" class="bg-lavender dark:prose:text-neutral-50 dark:text-neutral-50 dark:bg-garnet px-1">nevent1q…ts42</a></span> </div> The blacksky.community web application now uses our own API servers to load posts, timelines and profiles.<br/><br/>When data is missing (accounts on the edges of the network, outside of our primary community, etc) we leverage @microcosm.blue@bsky.brid.gy<br/><br/>Blacksky </blockquote> 2026-03-02T19:53:46Z https://njump.me/nevent1qqs9e6qn52fekh3p3y3g2acumw3ecd7uyp47r5ja6jpj0e0zmgd4ucqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkjquxgf we obviously can&#39;t guarantee any particular outcome (eg, changing URI syntax will probably be a hard sell even for a single scheme), but want to have the conversation. 2026-03-02T17:32:32Z https://njump.me/nevent1qqs9sj08kl94vehv94nlxj4t6vh76ucpejps3ykaumt38wuxfeasa5gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkjme50d I worry about &#34;oh PDS is just no-cost CDN / S3 bucket / mega.nz / Google Drive / Dropbox&#34; dynamics.<br/><br/>there are some out-of-box limits in place, but probably more needed. better to get those in early so expectations are clear, and things don&#39;t need to be broken by reactive measures 2026-03-02T03:08:00Z https://njump.me/nevent1qqszaefs8qn7d0g8rfvaactl5r4njmcxx9xt3hjgsa00x0hgpqyg5dczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk9lvst9 I think the benefit of doing RASL stuff would be that non-atproto folks could leverage PDS hosting as one possible way to retrieve a blob.<br/><br/>that&#39;s fun to experiment with, but does raise questions around how many requests, do PDS instances need quotas, etc 2026-03-02T03:05:28Z https://njump.me/nevent1qqsd7kt52nsrlkg0g76az2ztdywa23ng3asxnrajhx99748wvqfd09qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkct06mx I do like the URI schema better than &#39;ipfs://&#39; for account-agnostic references (eg, if you wanted to label a blob CID regardless of account). but that doesn&#39;t seem to be the use-case RASL is shooting for (it&#39;s about mechanism to retrieve) 2026-03-02T03:00:45Z https://njump.me/nevent1qqsryr8mrncevm6d4m42xj0vxm905hvpvcyqlsu6g8tlz6xv68mx23szyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmku6c0wh simple little blob URI proposal<br/><br/>Blob AT-URIs - at:// pizza tho... 2026-03-02T02:13:42Z https://njump.me/nevent1qqsde5aenu47e0xjhwy52vtnkxk3ymgldjyv8tg6l9zuyn0j0xgyjjgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmky2pv0u finally got the atproto.com specs updated for sync1.1, and some general copy-editing.<br/><br/>feels like a milestone to me! still some details to resolve+clarify, but all the bones of public atproto are there, as deployed and running in the network.<br/><br/>github.com/bluesky-soci...<br/>github.com/bluesky-soci... 2026-02-28T07:08:38Z https://njump.me/nevent1qqsdq8zasshpxjzlstvqhyvhyrndjd8zwpt4jenk6rjgl6cy2gwr6aqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkaevrjz (it&#39;s back) 2026-02-28T06:34:34Z https://njump.me/nevent1qqsdeg86ygm07tm5nqu82lqcmrhacpqvdhhp8lm6crwxy3g0mtnhngszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkz4jsjw discover feed is down for all users. we have a hypothesis for what the problem is and getting a fix out. 2026-02-28T06:09:46Z https://njump.me/nevent1qqsdzsvc08w938d4c8yxkmucr6xsewar2y8mm3e0emu0nfxssfu675gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk939m8k anthropic/DoD thing is a real CCP moment. who is the real top dog? will it be wall street? 2026-02-28T01:00:32Z https://njump.me/nevent1qqsv06zs286tgzn2x5x2src504al9xetcu6z3jmax7nx2p5pqa4a6jczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk4m2yfq that can be real friction for devs: we are pretty used to getting away with making impactful design decisions over in dev-oriented spaces, like git code review!<br/><br/>but in the case of wiki talk pages, there is at least some consensus process and you generally get to a decision 2026-02-27T22:19:08Z https://njump.me/nevent1qqsdzdp6eg6e3d6jl5lq3nhn50k5fu2zgklgzmex6ecfnu8amc52xkczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk58nacx wikipedia has an interesting dynamic where the editor community has the most clout and set direction/priorities.<br/><br/>one of the way that plays out is that devs are strongly supposed to have all editor-impacting design discussions in wiki talk pages, *not* in code-review tools. and use accessible lang 2026-02-27T22:16:40Z https://njump.me/nevent1qqs9dhs6tfp7ptyeuc5s43tqg6ehka0tqmfcrgx7ssargwgsaap67yczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkhsu780 the counter point is that there is a lot of important software which is very annoying/unfun for devs to work with: wordpress and mediawiki come to mind. has not held back those ecosystems at all.<br/><br/>(maybe those are fun for some to work with, but I think not for most devs) 2026-02-27T21:57:57Z https://njump.me/nevent1qqsta4t40u6szw7w9lcqfjegfv2feyd0266rk00wtsu8v377gusrnegzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkml4mp2 ... though having it as a process that manages everything in a language-ecosystem-agnostic way is also pretty effective 2026-02-27T20:53:13Z https://njump.me/nevent1qqsq47lxuyecm2ydezgmn3267kk8kk8723g52vl3zenvgg54pjwanqszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkyhku2y I do plan to come back to that, probably in april.<br/><br/>will first swap to new codegen impl, then do a bunch of additions like server handlers, full validation (eg, grapheme counts), maybe json/v2 annotations and alt CBOR marshalling support (eg, go-dasl) 2026-02-27T20:35:45Z https://njump.me/nevent1qqspe3xt3nxa367v72y0gnn5fnl3sjjvft2g3l63vpr2hkys5tzz9ngzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkyadhg0 the app/feature design space is also just huge: everything from preferences (kawaii mode!) to monetized content to closed groups. wide spectrum from &#34;minor affordances attached to a public app with public network effects&#34; to &#34;exclusive app modality with no public profiles or discoverability at all&#34; 2026-02-27T03:08:25Z https://njump.me/nevent1qqsynaam0m29tdpampsdjk8cm2fjzc8d5ct7qyrrfm8ymghkusw7asszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkcsu6pj one thing we&#39;ve discussed about that is deleting (or moving) the profile record, to indicate the account is inactive for a specific app type.<br/><br/>a related thing is that we&#39;d like to move account deletion (and change password, handle, etc) to a web UI on the PDS itself, instead of every app duplicating 2026-02-27T00:07:42Z https://njump.me/nevent1qqsytqcm6p7g607k9ehf9j7vt50j4lgn5dcjgcvwa2js8ephsy2uadczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkckgrkz as an example, it has come up a bit, but generally don&#39;t see a lot of folks talking about labeler declarations being an app.bsky.* thing. should those be per-app? or a common/generic lexicon?<br/><br/>I do think a &#34;full example&#34; app w/ all this would be very helpful. statusphere is very bare-bones 2026-02-26T22:25:40Z https://njump.me/nevent1qqstjkcf84fey0cage32m4q8kln5ecr8wyydcld67gslg3twsq899zgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkwuscnm I agree with your observation that other app modalities (lexicons) haven&#39;t built up ecosystems around feeds, or composable moderation yet.<br/><br/>(though I vaguely remember at least one other app has feeds/lexicons? maybe sprk does labeling also?) 2026-02-26T22:23:43Z https://njump.me/nevent1qqszzjtjlmwp96fytp8ujtsyy2ause6g6q04hs9lmhhy969seuzek6czyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk8uk3yw tap and jetstream are popular/important, but also protocol-adjacent. I think of tap as being a reference implementation of how to implement a full sync consumer.<br/><br/>feeds I think of as an app design pattern; similar to having a &#34;signifier&#34; record (like a profile record) 2026-02-26T22:21:15Z https://njump.me/nevent1qqszrnclynqef9lx8lq97ms5ktmkh2zjsvwlqceucf5q80yl3g5ppxczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk9z24tq and the jetstream issue should be clear now as well 2026-02-26T21:57:35Z https://njump.me/nevent1qqst7rltmzznn7wtvawqtlxdgdgmpz86v79zd60d0av0pftnyckr7wszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkfvrrnh this should be clear.<br/><br/>we are looking at a separate jetstream issue with jetstream1.us-east.bsky.network specifically. that will be down another hour or so. recommend switching to any of the 3x other instances for now. cursors are compatible across instances so this is an easy config change<br/><blockquote class="border-l-05rem border-l-strongpink border-solid"><div class="-ml-4 bg-gradient-to-r from-gray-100 dark:from-zinc-800 to-transparent mr-0 mt-0 mb-4 pl-4 pr-2 py-2">quoting <br/><span itemprop="mentions" itemscope itemtype="https://schema.org/Article"><a itemprop="url" href="/nevent1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqeu73dqlwedya9wfkq7s6zvzgf8jqz8fk9z95v4yu2xzk2hywsqzqvdvpmn" class="bg-lavender dark:prose:text-neutral-50 dark:text-neutral-50 dark:bg-garnet px-1">nevent1q…vpmn</a></span> </div> looking in to degraded bsky app content over the past hours for a large fraction of users. </blockquote> 2026-02-26T20:33:55Z https://njump.me/nevent1qqsp9wnpuhzunkj6xsth2qqfrsevl37mdeew8szrghuyhkvf8qd7hwgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkxdhdaa indexing! degraded indexing 2026-02-26T20:16:23Z https://njump.me/nevent1qqsv70gks0hvkjwjhymq0gdpxpyyneqpr5mz3z6x2jw9rpt9tj8gqpqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmktnt84h looking in to degraded bsky app content over the past hours for a large fraction of users. 2026-02-26T19:42:09Z https://njump.me/nevent1qqs04eqzyzn2pt233d5ptyrjelyju0pwmrs4em2vkxwj3wl38f7cmfqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk3lnuk8 I poked some buttons, please let me know if it is working now 2026-02-25T21:00:03Z https://njump.me/nevent1qqsfg428mzkyyeshzv2872tph3pgh5qyvvq2dksd7xgzlvtk4fhj6rgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk2wlsed and just to re-iterate the point: PDS implementations are expected to be &#34;app agnostic&#34;: they provide generic atproto support, but it should be possible to use any atproto app w/o special PDS support. eg, leaflet.pub, tangled.org. 2026-02-25T17:43:14Z https://njump.me/nevent1qqs9me5l2s3t9fpwxjgjqpkv5xzn5h585hpxy4kvqrundttscaqmhkgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkg9maja and the general philosophy is that the client app decides where it wants requests routed to.<br/><br/>for example, the bluesky social app (in app stores) is designed to use bluesky API servers. some forks of the app (web, not app stores) are configurable which API servers to use 2026-02-25T17:40:43Z https://njump.me/nevent1qqspwyuk0qvd6dfsp0eh4l2hleeah7ttr2q6pz656dqt0qcgqw79e7gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk7gfx6z you could also look at the size/complexity of @hailey.at@bsky.brid.gy &#39;s &#34;cocoon&#34; PDS implementation. there is a list of API endpoints in the README: github.com/haileyok/coc...<br/><br/>hailey uses that implementation for her main account on bsky<br/><br/>GitHub - haileyok/cocoon: An A... 2026-02-25T17:29:58Z https://njump.me/nevent1qqstj82g9py6d9j3eway7utvfg5pkzt6znpmlpzs3wgtecanvd8288szyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkd3lqy4 IMO if your implementation is emitting valid firehose events, and you have the server side of OAuth working, you are most of the way to a functional PDS.<br/><br/>this (old, 2024) doc summarizes what a PDS does<br/><br/>What does a PDS implementation... 2026-02-25T17:28:49Z https://njump.me/nevent1qqs0qxadprwefap866m5rtnk3j9lgk898eh0gn9dh06mzv9q8alayxqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk0c9rsy there is a bit of legacy cruft around this where the PDS reference implementation auto-proxies a handful of app.bsky.* endpoints, but that is off-protocol and (optimistically) should not be required any more. 2026-02-25T17:25:59Z https://njump.me/nevent1qqsg6lm2wjvffdw9jxa6y0jh0r52knujgwg6exgahy8vluwgl75y0aqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk9usjgl an example of where this arch does work great is the feed ecosystem. you don&#39;t need to do OAuth or login to any feeds, you just start sending them requests. the feed can decide whether the auth header (JWT) is required, optional, or ignored. 2026-02-25T04:29:47Z https://njump.me/nevent1qqsz96cdhptcm0j34hhv47n6jfzfsmjjedcj690vvpq3t2yh8zj2tpgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmklp48qx that auth mechanism is part of the protocol, but isn&#39;t the only way to architect atproto apps.<br/><br/>eg, logged-out public views go direct to those API servers, not via any PDS.<br/><br/>you can do a traditional web app (eg, rails/django) where browser/app talks just to server 2026-02-25T04:28:15Z https://njump.me/nevent1qqsrpkjsjuq9tm8z7n8ul7zx2ewtlcauxkqae4p3txmy6evn8yahrmczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkzdugql ... which your PDS resolves to an HTTP location, and then proxies the request through, adding an auth header w/ JWT signed by your identity key.<br/><br/>it&#39;s a weird architecture, but keeps auth (and keys) all in the PDS, so clients stay simple. 2026-02-25T04:26:28Z https://njump.me/nevent1qqs9g8qv7z9xp8jwk7jxdtxjaw62w0ua8ncf700k35xsfmrxg6awjzczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkug4jgv (this is all doable, it just isn&#39;t a quickie PR merge. it needs lots of planning, testing, gradual rollout, etc) 2026-02-25T04:23:04Z https://njump.me/nevent1qqsdqt2jqlk7qh3rc9wgxc0lnhhna46qdvs2f0ukjhaqhc4ytjejj5szyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmklgyf3d let&#39;s make &#34;app modality&#34; happen (as terminology) 2026-02-25T00:43:49Z https://njump.me/nevent1qqsq2s2lem4vy82agd37m4xhmmd4wl396mcdg32sl89h3py2x83ukxqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkuqrke9 cc: @thisismissem.social@bsky.brid.gy w/r/t @questionable.fyi@bsky.brid.gy 2026-02-24T19:39:49Z https://njump.me/nevent1qqsfr09r7y4u8hskfkdhetd38u8stm973qwpu6mlkv4mlq5p7peq8mgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkhpt7dp great real-world analysis of voice/exit power dynamics around the stack exchange community collapse.<br/><br/>very different narrative from gen AI inevitability: governance tensions had been growing for years, with active contributors excluded from high-level decision making.<br/><br/>AI Didn’t Start the Fire: How ... 2026-02-24T19:39:16Z https://njump.me/nevent1qqsycvasevmfcjql5aurtpuytmflgqv9atkeu6k6xcg9u3kwddk8j8qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkuqp3xl the code changes are pretty straight-forward, a number of forks have implemented it already.<br/><br/>a good first step would be merging optional support in to social-app repo.<br/><br/>then work through a transition that minimizes user confusion and account loss.<br/><br/>not rocket science, but we are still a small team 2026-02-24T18:53:41Z https://njump.me/nevent1qqs2atv94lrjj6hq78xe2hy5h587rtttfn5a9xxd4z9f09ck55gclgczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkgyqx2h it also doesn&#39;t *have* to all go in the same system similar to tangled/git, you could have an S3-compatible bulk storage service, or a HIPPA-compliant data vault.<br/><br/>these could be bundled or unbundled vs regular PDS service 2026-02-20T20:18:06Z https://njump.me/nevent1qqswwn4q6pnfz2zxnrexdakzh042fehgdv9qfz24jvhe8t0p0m3dgcczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk6tflrc I think we will probably want to go slow on this, mostly to not disrupt the user/PDS relationship too much. hosting a public social media account is pretty different from sensitive private data, or hundreds of GByte of stuff.<br/><br/>that&#39;s probably a combination of rate-limits, quotes, and docs 2026-02-20T20:08:02Z https://njump.me/nevent1qqsft7kmju752376c0xthwykv72vap5yw88d9curq8l8g40tj4mse6czyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkscx92l atproto *personal* data has a ton of potential.<br/><br/>this is stuff backed up and synchronized between devices, but not shared with others. personal notes and annotations, sure.<br/><br/>what about all your photos? health data? laptop backups? lots of questions for PDS ops. 2026-02-20T19:33:06Z https://njump.me/nevent1qqsx4re0ppwymxps504tnkwh4m0y9kfu908f9cmmx89ku37m373dtmczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkhjel07 you can query status with goat:<br/><br/>RELAY_HOST=<a href="https://relay1.us-east.bsky.network">https://relay1.us-east.bsky.network</a> goat relay host status pds.sgo.to --json<br/><br/>debug.hose.cam can also be helpful<br/><br/>atproto PDS &amp; account debugger 2026-02-20T06:27:11Z https://njump.me/nevent1qqs8a3uraskgzrzyhlyna0x9g7cmnrgqryqagg8ktv5kr308x8jugxszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkxf48dz to your original question: better operator tools for debugging PDS/relay issues are needed. it&#39;s been sorta surprising how reliable things have been that these haven&#39;t developed more given how many PDS instances there are... 2026-02-20T06:26:07Z https://njump.me/nevent1qqsf5dlyh4fdr2s9f8mmt5dqxnk0869gksy2jnvr4gpa93vewayzpcgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmktfm85d I think it would be a great sign for our docs/specs if an agent could build a functional PDS.<br/><br/>we have some improvements to the specs coming which might help (see PR)<br/><br/>there interop tests might be helpful for syntax stuff, repo MST, etc:<br/>github.com/bluesky-soci...<br/><br/>specs: update repo, sync, and ... 2026-02-20T06:23:22Z https://njump.me/nevent1qqs2f8ypr2evvhqpp5dtskeqaxnqqd92u3nutjccdwcmwyjhxfja0zqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk6e6w32 in some sense, impressive it output a firehose at all! a couple issues though:<br/><br/>- the output firehose events all have seq=0 (they should go up)<br/>- the DID in the messages don&#39;t match the DID in the repo commits 2026-02-20T06:21:19Z https://njump.me/nevent1qqs0lp9665ckyxzugatwecz8kwjzpur0z58xn2jrlcpqmvqarcxmjrqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkuk3ued Overall I think this is a promising direction to explore. Thanks for the write-up! 2026-02-19T09:42:15Z https://njump.me/nevent1qqs8v8kcvq537dk6l4qpdnpumxjjhjlppstar9q7tgtuqwwkrw0wlnqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkvfda40 we&#39;re also thinking about how to represent things like &#34;only people I follow can see this&#34; (in the context of the bsky app).<br/><br/>I think annotating per-content is an interesting alternative though. also that the same content (exact record) could be visible to multiple audiences/groups is interesting 2026-02-19T09:38:19Z https://njump.me/nevent1qqsfcf6r3j5u22ctuh66kmlaep4qepwwjljkteptctstta6dun3hxmszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkx5ye8q if I understand correctly, you propose annotating individual pieces of content (eg, posts) with access control metadata (eg, &#34;domain&#34;).<br/><br/>we have been thinking about having a shared group space (a &#34;bucket&#34;) that everybody scopes their content to; and then the overall bucket has access control info 2026-02-19T09:35:46Z https://njump.me/nevent1qqsw95ya4kjp434f2l80n4k80k3lgjrr4y0hpuuy9tfw6prettlfz3gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmktwu4ms (aka, two infra-rooted private spaces on separate infra; your account can only be in one or the other)<br/><br/>you mention avoiding accidental linkage between public/private data. agree! I think we should make it so URI syntax is unambiguously pointing at public vs permissioned records. even a new scheme? 2026-02-19T09:32:17Z https://njump.me/nevent1qqsxhwkd7rerfvms2g0ateypr9dwqphuh2wxwswx394e20k8sve53vszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk6yex47 you mention having flexibility in the definition of boundaries, which could enable defining groups which span PDS instances. that is a pretty important property to me: having community infra is great, but I don&#39;t want to put folks in positions where they need to decide between exclusive groups 2026-02-19T09:29:56Z https://njump.me/nevent1qqsr449hh4a6xcw8n8ff0a20t6g9s39al5vn6kjqwmm4p2sx577977qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkgjjpff you also touch on moderation! which definitely needs thinking as early as possible. I hope we can reuse as much as possible from public atproto systems (labeling? abuse notice proposal?), but it&#39;s going to be a bit of a different beast and involve different mechanisms, relationships, risks, etc. 2026-02-19T09:23:57Z https://njump.me/nevent1qqsxy0xqyncmqsrwvf9ndy2kj9h4e9mcjcduu58xvhkwkfzyzxn0dnqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkkrtwsu great ways to experiment and collab with the active local-first protocol scene 2026-02-19T09:14:11Z https://njump.me/nevent1qqsq5gsnckx4y8qcd7dlavmxs94re6yp9r5yv0tck3c53h7zlmegvpszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk23k2fa I phrased things a bit strongly in this thread, but I think it is an interesting proposal and good contribution to the conversation! 2026-02-19T09:02:34Z https://njump.me/nevent1qqs8lp2s89r3jh8v0utfl3wed4a5g7s7tt48vs285394dmkxla6endqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk0y34qs overall i&#39;m pretty confident that data storage and data transfer for non-public data should not use the same public repo and firehose mechanism. I&#39;m pretty sure we should also have a new or variant URI scheme so that it is impossible to confuse references to public and permissioned records 2026-02-19T09:00:23Z https://njump.me/nevent1qqs2w625z9lpnuylvgydtvjk6pczwwjcf5kmctayud6w7uzc5glx5nqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkks5hdt &gt; Nothing in this flow needs protocol changes<br/><br/>I think there would be several things here that break interoperability, or create easy footguns. eg, if you export full CAR from a PDS with these behaviors, and import into reference PDS, it would just make all records public 2026-02-19T08:56:03Z https://njump.me/nevent1qqsq3egjnp9t867a48e8nyfeyl6rasgcqgwfzwrwh0uu7hyme8x6dagzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkyput6k re: sync<br/><br/>&gt; metadata-only firehose event is the natural next step in this progression<br/><br/>I disagree, the core goal of sync 1.1 is to redistribute repos+records in a complete way. the changes in sync1.1 were a step in making that stronger (making it more certain that you didn&#39;t miss anything) 2026-02-19T08:53:11Z https://njump.me/nevent1qqszqn7r5cpcee2vv757egjfgh6vtw9dmmankqq2v2nph5yrkydn4sgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk408qld I&#39;d recommend/request not overriding the existing `strongRef` lexicon type, or built-in blob types directly, even in experiments. instead just define new lexicon object types directly with the fields you want. 2026-02-19T08:45:52Z https://njump.me/nevent1qqs9uuvlnknau97tr7l9paqt963qg2u3d585t863cdtmkw6gkjmg40szyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkwysf43 I think trying to actually have CARs/MSTs with fully dangling records would break stuff, and go against the intent of the repo structure. eg, the relay and validating sync consumers should reject commits or backfill CAR files with missing records. stub records don&#39;t have that problem 2026-02-19T08:42:06Z https://njump.me/nevent1qqsyrflgcuz8w72r5jns2ff6dgunycaxt2fefsn4p6fhgnqe52gkxzszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkut6uet I do appreciate that your proposal wouldn&#39;t require huge conceptual changes, and has variants that folks can just start building with today (like what Blacksky is doing with stub records and community-only posts). 2026-02-19T08:39:55Z https://njump.me/nevent1qqs98jvw8lqj8rq60kmlk9rsxl424tpt9a6d5hzwfz9fft9r2u59z3czyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkfpkcf0 your current proposal puts service endpoints directly in data records. that would make it very hard to switch providers. I think the credible exit story for permission data service providers is going to have different dynamics than public, but I still want it to be as low-friction as possible 2026-02-19T08:35:35Z https://njump.me/nevent1qqsrvnnkk4n7yjmnq5se7l08whdc3pp5jn3des8hvtl98lesl0fqjhgzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkz9xzad not everybody wants everything from the public firehose, but there is a degree of value to all the firehose data. doing metadata events is almost as much resource (evt/sec, MST churn) as a public record, but almost nobody gets any value from it, only the recipients 2026-02-19T08:33:09Z https://njump.me/nevent1qqswscpk04an5mnvv4qwumj5fvy30jtwenc6lewnnc3ctfjkfd3y73qzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk0xvt6u but I don&#39;t love the metadata CAR concept.<br/><br/>my main issues are metadata leakage, and using a public broadcast and redistribution system for non-public data. they are diametric design goals.<br/><br/>some secondary issues are around efficiency, both on the wire and in implementation... 2026-02-19T08:31:02Z https://njump.me/nevent1qqsqdqewulvl7477xd872cec2xmuen0v85qmy8e60sfn3f4kwzuqwvszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkswrv6n (though there are benefits to having ACL explicitly at protocol layer)<br/><br/>and I definitely agree with export and migration parity. 2026-02-19T08:28:32Z https://njump.me/nevent1qqsyzmsr5xkpxx45w4eevgt7fwfn87x8zdnd7snnqtezea46he5eggczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk7l3uyx in some sense having &#34;permitted identities&#34; (other users) represented as an ACL directly in the protocol is sort of optional. you might want to leave that to application layer; like &#34;folks I follow can read this&#34;. those relationships are in app-defined and app-specific schemas, not protocol-level 2026-02-19T08:25:07Z https://njump.me/nevent1qqs0lhj9y0wakmjvd8c7pvxv3zevvy6llxt77cmttm4l9u47wgrlglqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkt9x93x not sure if any of this is what caused the problem you saw, these are just some common related issues 2026-02-10T01:17:01Z https://njump.me/nevent1qqsqdjtpsa3gcf9e98a6cxkxc8wyyumhpyjh6zpuhn36pspp4ez8jpszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkagmkc6 a number of things can cause the websocket to get &#34;stuck&#34;: packet loss, network glitches, implementation bugs. we used to hit problems with &#34;stale&#34; websockets ourselves; we&#39;ve mostly resolved it by iterating on our client logic, including keepalive and heartbeat details 2026-02-10T01:16:23Z https://njump.me/nevent1qqspjq6g9nf8027vrltlnvt005e3xy5pyny07jvsy8lfmhatv55l9aczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkqupyk5 this is because the relay code has two modes: reading old messages from disk (backfill window) and receiving live/current messages. a client connection can go from backfill to live, but we don&#39;t have live-to-backfill implemented. simpler to just drop the client and have it reconnect 2026-02-10T01:15:11Z https://njump.me/nevent1qqstlwqwayzgnz6wk4uet5kw9e2hkwwjf5t39xf9cq7dpa7slj5zvgqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkys9rg7 (should be all resolved) 2026-02-09T23:07:58Z https://njump.me/nevent1qqsplsfdn9pg76tls8x8quk0fuzqd6xd6990v2p06gmwt0f2hejzpcqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkdx6rh4 every time I post these now I envision @rude1.blacksky.team@bsky.brid.gy doing a little smh 2026-02-09T22:50:13Z https://njump.me/nevent1qqsvrg3tvm6rjfdwr9kgq5arm9sgald3jzckrw7dsvacj2gtn8e58ygzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkd2pfs8 bsky API bump due to a bad deploy. rolling back and should resolve in a couple minutes... 2026-02-09T22:46:52Z https://njump.me/nevent1qqs296ju9e3j047gs54czl06le9u0qwuwect8wg2d3kaccldnw3nglszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmktwnqv0 nice 2026-02-09T03:23:59Z https://njump.me/nevent1qqswtv5za63plv9454umy5uff5sf5nuq2kz0c7mnsnxrmpwthfhlucqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk43svk7 touche budwiser 2026-02-09T02:09:17Z https://njump.me/nevent1qqsgllm35zd202pwrmsu08zwzcwc3ye2qp9xu9qg33e2h7qvevk9waczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkkt8e7y Twitter is also showing up on downdetector, likely contributing. 2026-02-09T01:53:23Z https://njump.me/nevent1qqs20m0spm4a5mvu3d25a9w2wsq3m006wwl4cqfqze0wmjd8p6g2snqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmku3e7gu what are the seven layers of dip 2026-02-08T05:49:45Z https://njump.me/nevent1qqsvx9j8kwaa8kxcs467gkmgatj69z6zgyqwfytuyap5tq3k3wshvyszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkjdj2dq will the whole things scale to billions of users? I think so. will need to do some boring distributed system stuff at the relay (sharding), but there are patterns for filtering firehose by data type and community. 2026-02-07T19:50:32Z https://njump.me/nevent1qqswpfhlcpjultkcdjyper23jrnqcruwm32kxnqkqvm2hyepc9kay4gzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk9zx4pl the:// 2026-02-06T23:35:05Z https://njump.me/nevent1qqs285dzqh72am3f4w3ykw3pg52y2nd7ekpq7gr89k2q3rv0xgwxlgqzyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk65tf3c the simple solution to the atproto capitalization debate is to not capitalize any other words in the sentence either 2026-02-06T23:23:10Z https://njump.me/nevent1qqsrjpdr34x4xm7h8jckxr3wcqnthysu2wzg5ha5je8wzxdr2ckxyqszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmk36nsq2 I think calling out bullshit or attempts at enclosure is good. social media has a history of rug pulls and skepticism is healthy.<br/><br/>but i&#39;m not seeing it here, at all 2026-02-06T23:09:22Z https://njump.me/nevent1qqs0lqnpe5dt3yhch6amrlh668hd6309wefglqr9j74x296w037eyxczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkvvu6ch cc: @mosh.bsky.social@bsky.brid.gy @bumblefudge.com@bsky.brid.gy @robin.berjon.com@bsky.brid.gy 2026-02-06T20:20:52Z https://njump.me/nevent1qqspsemw37wefg0epnz3g6hsrpqxc8rfr6664xw8nw3mpl4zf00excszyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkd5hawf you can have self-hosted community appviews, and larger appviews which contain all public communities, at the same time. users can chose which they want to use. 2026-02-05T04:29:54Z https://njump.me/nevent1qqstwclrv4a52332yjyhs67mnya6fj83auwckqn9kttdvxhf57q9zuczyp0nll6kef86s9mlzy53pg60fdnykmfq7j5jgm48nj5c5vpprzdmkcru5al bunch of bsky PDS hosts are struggling, though may already be recovering 2026-01-29T23:08:33Z