Landscape Weekly

[matildepark]

If you are a current or prospective Landscape contributor, feel free to attend our weekly meetings to discuss the ongoing work on Landscape.

• Every Monday, 6:30pm to 7:30pm Eastern Time (Toronto / New York)
• Video call link

Agenda

• Landscape team at Tlon checks in
• Tlon Design handoff
• Tlon Mobile handoff
• Community + Urbit Foundation handoff

External contributors are welcome to ask questions or report in during the final handoff. Recordings and notes will be posted weekly in the comments below.

Template

**Sprint + Date:**
**Recording:**

<details>
<summary>Meeting notes</summary>

### Tlon

@tacryt-socryp

@liam-fitzgerald

@arthyn

### Design handoff

@urcades

### Mobile handoff

@narner

### Community + Foundation handoff

</details>

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.4.19 Recording: https://drive.google.com/file/d/1gzWRg4Xcwd9O2nD-SVHAfzHj8b2Kr4o4/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Referential transparency work (https://github.com/urbit/urbit/pull/4764), also the group feed not loading issue (https://github.com/urbit/urbit/pull/4770) - Also bug work, including no longer caching session.js (https://github.com/urbit/urbit/pull/4776) - Changed tag queries within graph-store to be a jug of term to uid, instead of term to resource - Made group feed no longer appear in omnibox (https://github.com/urbit/urbit/pull/4786) - Disabled data URIs (https://github.com/urbit/urbit/pull/4787) - Fixed timestamp sorting in chat (https://github.com/urbit/urbit/pull/4785) - Added two arms to the ordered-map arm in zuse (https://github.com/urbit/urbit/pull/4788) - Testing should be getting ready for graph node deletion, back-end is done, will put the changes to the reducers to reflect the new way we do node deletion and put the other change to the reducer and add API actions to delete next - @tacryt-socryp to stage a branch - Threading next week https://github.com/urbit/landscape/issues/776 @liam-fitzgerald - Did new notifications redesign, part 1 of 2 (https://github.com/urbit/urbit/pull/4774) - Need to change the binning model, and the way we group notifications - Hoon changes will go out on next-userspace for storing unreads - Wrote a more performant ordered-map implementation for JS, to be PR'd; it's easier on structural sharing - This week will get to notifications part 2, and publish tokenisation (https://github.com/urbit/urbit/pull/4763) - Q: Design says there are questions for preference forms whether they're submission on change or submission on click - @urcades: After https://github.com/urbit/landscape/issues/179, should be immediate all the time - @g-a-v-i-n: The only exception is form that requires several pieces of information to be validated (eg. checkout flow with address + phone number) - Independently atomic pieces of data that can be updated, submit on change is the way to go @jamesacklin - Worked on group DM case (https://github.com/urbit/urbit/pull/4780) - Hard to detect if other participant in 1:1 DM has disappeared (if they breach) - It's now easy to invite someone to a group DM now, though - What happens if you want to invite someone into your 1:1 DM? It creates a new group DM with people who are already there - We might want an option to preserve history or not, depending on whether you want to bring a private conversation - Meeting with @jyng shortly on chat and reference spacing this week - @tacryt-socryp: Do we know in %jael if we have a date and if someone breaches? - @Fang-: We don't store a date, I think. - @tacryt-socryp: We should query the bowl then and make some way in the graph-push-hook, if you're the DM host, to get an easy path for who's actually included in your resource from the bowl, as an accurate representation. I think @liam-fitzgerald built an easy path, not surfaced outside the pull-hook data structure. - @liam-fitzgerald: The pull-hook knows whether it's got an open subscription or not, the issue is if you're the host checking that there's a subscriber for each user in the group. - @tacryt-socryp: This shouldn't be crazy, it should be doable in a short period of time depending if you know how to use that data. ### Design handoff @urcades - Due to changes in prioritising projects, @jyng will be working on Bridge in a more focused manner and it will be likely he will tie up loose ends here. I will be the point of contact for projects he was doing after this week. - @jyng: Collections is the only thing not presented to any engineer (https://github.com/urbit/landscape/issues/779), will work with Ed on timing. - @tacryt-socryp: I want to talk to design point this week about how the change to the tag queries in the graph-store could enable collections to be more interesting or more cool, because the tag queries in graph-store is a natural place to put like "bookmarks" or something similar - @liam-fitzgerald: The new notifications stuff requires swipe-to-reveal secondary menu to dismiss notification. Without getting into the weeds, it requires gesture or animation library to make it feel nice. Does anyone have strong opinions on animation libraries? - @urcades: The only guideline is that it should feel physically based, it should feel like a piece of paper. - @g-a-v-i-n: https://www.npmjs.com/package/react-swipeable ### Mobile handoff @narner - Big thing I have working now is S3 storage, which is cool because it means the ability to send images in a chat - Some regression on the profile — can't update anything in the user profile right now, L suggested getting on non-Tlon hosted - @liam-fitzgerald to try and replicate - Also stuck on pending messages. ### Community + Foundation handoff @jalehman - You may have heard of Urbit LFS (https://grants.urbit.org/proposals/1760204192-urbit-lfs), there's an external piece of software, that you run next to your ship, you connect to it like you do your Bitcoin full node, and you talk to it to get the file, so it acts like a drop-in replacement, using your FS and you can provide access to other ships. The win is that no one has to set up S3, and this is easier to set up. - The contributor is figuring out how to make this work with S3 file storage. We've had offers to document S3, but I think this is a better solution overall, so, how attached are we to S3 and is there interest in dropping S3 completely? - @matildepark: It doesn't necessarily have to be either or, since other storage implementations use the same code. - @tacryt-socryp: It would be best to have a little bit of dialogue about how these are set up, to be transparent. - @jalehman to set up meeting to scope Landscape implementation - Additionally, we'll have a demo of WebRTC ready. - @ixv: Is it possible to have an Urbit specific link to link outside Landscape? - @liam-fitzgerald: Links are Landscape-specific, because we can't do nice things with cages. We support a graph, a group, and an item on a graph. - @ixv: I could still do this with a relative URL path, right? - @liam-fitzgerald: You could piggyback and register a new protocol. Or use bitcoin:// - @ixv: There was a BTC icon made, do you know when we'll get that in Indigo? - @jyng: We can add that in indigo-react.

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.4.26 Recording: https://drive.google.com/file/d/1KvAERRIvnIR7dFBabeNVVubIShNRfTwW/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Last work on https://github.com/urbit/urbit/pull/4764 - Was thinking he'd make it backward compatible with graph-update-1 but decided against it - Because the update adds a condition to every node, if the post has been removed the node is still there but we keep the hash of the post there where the post was there before, it changes the noun structure of every node, and so if I want to support the add-graph and add-nodes updates I'd need to do some distinctly unpleasant work of putting the migrations into the mark itself as well as in the on-load of the store and support that - LF: What you could do as a gesture toward compatibility is, you can write a 1-2 adapter, don't change the min ver in the hooks, and the pokes don't check for compatibility range, so you retain the behaviour of pokes but subscriptions would still be broken - Implement the translation, crash on the remove-nodes case - LA: The remove-nodes case is easy, it's the migration that's hard - Nudged infra PRs - Reviewed webRTC - Up next is getting 4764 onto the network and add deleting to group feed and to chat after; and fix some p1 bugs @liam-fitzgerald - Publish work, VirtualScroller fixes, started work on the DMs flow - Might have to refactor ChatMessage to remove associations requirement - LA: If we want to do signature verification more or less we're going to need to implement some form of "when I send a msg through dm-hook, it signs two messages" - Vaguely concerned about how exactly the components look from the front-end - Otherwise sweeping - Deploying https://github.com/urbit/urbit/pull/4793 @jamesacklin - Jimmy's omnibus PR (https://github.com/urbit/urbit/pull/4801) is deployed - Meeting with @jyng on adjustments on it - Working on that and the transclusion work, the group DM invitation PR is out there and included the heuristic ### Design handoff @urcades - Alongside the general productisation of our efforts, the shuffling of general project reporting and what we're figuring out - @jyng and @tacryt-socryp to talk about https://github.com/urbit/landscape/issues/776 this week - LF: Can we get designs for permissioning on new DMs? (Yes, midweek.) ### Mobile handoff @nickarner - Got comments working on notebooks and collections - Doing TestFlight right now, banging on this a bit more @Fang- - Doing the last bits of work on webterm, there was an item to clean up subscription logic - Can I just connect on Landscape start? - (LF and LA both object.) - LF: The more updates we get the laggier Landscape gets, basically. - MP: Could we just deliver the past 100 lines or whatever on initial subscription? - MS: That's not trivial to do, basically.

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.5.3 Recording: https://drive.google.com/file/d/1nCWELQDXiVCiM6ogAXux-TeHBRpSA1Y9/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Finalised referential transparency PR and adding deletion into notes and links (https://github.com/urbit/urbit/pull/4847) - The actual logical content of the PR went through without a hitch, but the network pressure caused by the upgrade was catastrophic. - LF, LA, PM, JB learned a lot about the network. - Did a little bit of looking into hark-graph-hook slowness and did a small PR that likely has little affect (https://github.com/urbit/urbit/pull/4846) to try and learn it better - Have an approach, want to talk to LF about it, to drastically increase the performance; instead of scrying down tubes in clay and running the conversions, I want to make another update path on graph-store that issues marked graph updates (graph-validator-chat-update for example) so we can make assumptions about the contents of the data and then also do watch-as on that from hark-graph-hook so the mark conversion would be run by gall to convert all of our subscription updates into a format hark-graph-update is more amenable to. The slow thing is the scries, even if they are memoised. The improvement should be a perception of improved speed on graph-store itself, even if it's hark-graph-hook is the faster agent. - Working on group feed threading next. @liam-fitzgerald - Have basically finished DM stuff (https://github.com/urbit/urbit/pull/4813) with the exception of references and signatures. If we want to reference a message that's been sent, we have no mapping from our indices to their indices, so... - MP: Why not just blockquote it? - Seems fine. - We also don't know what path to sign on; if we're not rewriting the index for strange reference magic, then it's probably fine. - LA: Index isn't in the validated post. - LF: I meant if we had to make references work. - If you go to make a DM there's also no blocking or loading time, it throws you straight in. - LA: How is blacklist implemented? - LF: It's whitelist only. - We could probably do a blacklist store and store ships in that. @jamesacklin - Chat references are feeling really good. It's coming together. @matildepark to review https://github.com/urbit/urbit/pull/4780 ### Design handoff @urcades - @jyng moving to Bridge - Ed to take point on Collections after - Ed wants to put the bug icon into indigo-react - **Change it to a bug in 1.2.22** - @liam-fitzgerald may take over import/export after DM work https://github.com/urbit/urbit/pull/4457 - @liam-fitzgerald to review https://github.com/urbit/urbit/pull/4810 - @liam-fitzgerald to wrap https://github.com/urbit/urbit/pull/4809 - **Let's do a real dumb web notifications PR** @liam-fitzgerald - @ixv: anyone have better ideas for glob than hard-coding a hash? - @tacryt-socryp: I could see a situation where you register a sponsor, who sends you a hash and URL and then you'd download the file, see if it matches the hash - @philipcmonk: glob should be extended as a generic piece of infrastructure **@matildepark, @ixv, @liam-fitzgerald, @tacryt-socryp to book meeting on generic globbing.** - Finally, @liam-fitzgerald has end to end testing, performs automated actions, screenshotting visual diff testing, wanted to float how we could get CI working on it. **@matildepark and @liam-fitzgerald to book meeting talking about CI stuff.**

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.5.10 Recording: https://drive.google.com/file/d/1wEvYpk2zAQ6jn8slo58UOqMZabUoN_1F/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Blocked on the graph-threading improvements (somewhat) just on FE - Know there'll have to be some weird rebasing when the next-sys stuff gets pushed out - In the meantime basically been doing Hoony stuff, re-implemented deep subset functionality using tab method (https://github.com/urbit/urbit/pull/4852) - Added profiling to graph-push-hook (https://github.com/urbit/urbit/pull/4877) - Discovered was slow in the multitude of mark scries in the graph-push-hook (https://github.com/urbit/urbit/pull/4880), when @liam-fitzgerald added graph permissioning changes, more or less we incurred a performance hit of 100ms per message and another 50ms on other work - Figured out how to cache a lot of things, came to short-term solution that gains 50ms per message - Infra-wise, we should add nest-checks to compiler itself so we don't need caching layers into our agents - Need @liam-fitzgerald to review 4880 @liam-fitzgerald - Spent Friday and Monday doing cleanup and fixes - https://github.com/urbit/urbit/pull/4886 - https://github.com/urbit/urbit/pull/4887 - Did some indigo PRs so we could style the markdown for tables and lists - Five or six bugs assigned to me that have to do with VirtualScroller and backlog - I want to add a property to every graph called `live` and keep a window of chat messages, instead of adding chat message to our state for the end of time; if we go up so the backlog would never be at the bottom, then the `live` flag is `false` and `/updates` no longer apply - LA: As we move forward, we should use the actual react state as an LRU cache of the data we actually know about, and more of the data we know about should get stored in localstorage - We should do garbage collection of our state - Graph store referential transparency is useful for this because we could forgo doing this with localstorage and use the browser cache more aggressively and set longer TTLs on graph-json requests we make - But we'd need to change the approach we use for our graph-subsets to use the tabulate method instead of the subset (`+lot` in next-sys) - The younger and older siblings arms in graph-store would need to use absolute message start and use tabulate to give a count - @liam-fitzgerald: My concern about localstorage is that you need to serialise and deserialise state on the main thread @jamesacklin - After a frantic weekend of fixes, got references wrapped up - I feel like I need to make another pass on the group link - @jyng: Deleted link just shows text in a non-designed way. ### Mobile handoff @nickarner - Worked on adding everyone to TestFlight, if you haven't been added, I sent out an invite for joining the App Store Connect - Did fixes around logins, URL mangling, etc. - Everything works on arvo.network as well - Did work on interface and UX bugs that were filed - Did some work on a bug caching profile information for groups - Got chat working again - Then just working on fixes and crashes - Would love some pointers on JSON parsing code @liam-fitzgerald: Can I get a list on actions that should be optimistic?

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.5.17 Recording: https://drive.google.com/file/d/1aEzmKyubNVBuLYlklE8ObKXROuc8OOuL/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Expecting to finish https://github.com/urbit/urbit/pull/4904 this week - Have views this week that work for thread views, timeline view; you can navigate, except for going back - Need to make pending messages work properly - VirtualScroller bugs based on how large the scroller is - State management problems, still troubleshooting with @liam-fitzgerald, Immer seemingly doesn't play nice with vanilla JS objects - Hoon is done, blocked on next-sys, requesting a design pass in latter half of the week - next-sys should be sometime this week @liam-fitzgerald - Finished the optimistic store state - Fixed issues with graphcontent, have table support, fixes blockquotes (https://github.com/urbit/urbit/pull/4905), reimplements the blockquote parser - Working on the new boxing model for unreads - In the old timebox model, notifications are added to each timebox, the timebox is reset every 3-4 hours - The new box model, all your unreads get collapsed into one and marked as read when you open up notifications and there's some refinements to how marking a chat as read impacts notifications in the box - [Mockup](https://www.figma.com/file/ovD1mlsYDa0agyYTdvCmGr/Landscape?node-id=6245%3A136287) - LF: Should we crash if we receive an add-graph and after application it's the old graph? Same as what we do for add-nodes? - LA: We have an overwrite flag because we want to know if we want to overwrite an existing graph or not. If not, we should just turn off the overwrite flag @jamesacklin - Working on loading placeholders for transcluded graph nodes - Don't have it solved, was looking for existence of the node in the state because we do some async stuff to the get the node - The problem with that logic is that sometimes you can't get the node - LF: Can't you just catch the exception when the API fails? ### Design handoff @urcades - Bitcoin going out on Ropsten obviously - Reworks of Bridge are in the pipeline - There are explicit near-term changes in the existing Bridge, @jyng considering future scenarios - It's been vocalised we'll be tapering hosting and diving more into a client-dominated onboarding pipeline - A pretty minimal scaffolding was proposed within the design team to structure how a client-forward environment could go down; it has implications for what we know as Landscape and deconstructing it into client and application launching interfaces - Grid is already taking on the form of @philipcmonk distribution stories and bundling it into an interface expression, and there are thus far no serious flaws in the design plan from @jyng and @galenwp and expect to see more soon - @tacryt-socryp: as for onboarding and group integrations on Taisho, we could place another text file with the group name as an resource the app could access ### Mobile handoff @nickarner - We're looking on focusing the release more on Chats and DMs now - Working on performance improvements right now, message loading and bug fixes - Working on getting rid of a lot of duplicate code, breaking out of giant duplicate classes

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.5.24 Recording: https://drive.google.com/file/d/14bGcxhH3esM8ABF5aITwnkqtdiSgJFG0/view?usp=sharing

Meeting notes

### Tlon @liam-fitzgerald - Bug work mostly, mostly with edge cases - Working on Jest setup for edge cases - Deploying https://github.com/urbit/urbit/pull/4904 onto devnet with @matildepark @jamesacklin - https://github.com/urbit/urbit/pull/4931 PR'd - Default names for group DMs working — have a question, since it relies on nicknames, it relies on whether the host has seen that ship before, should it also depend on the host's CalmEngine? - @matildepark: Just name them patps and key it into nicknames ### Design handoff @urcades - Week or two ago onward, design's focused on 6+ weeks in the future - The idea is that @g-a-v-i-n and I are taking the idea that things will be gridified / made as modular applications as driving principle and he's done a great job of translating the editor into a standalone writing application - It's begging a lot of interesting questions — how do we provide a general API between separate applications to let them speak to one another? It's something we'll see happen over and over again as we decouple them into grid-oriented software - As a design team, for a general specification, I should be able to write a blog and send it to an arbitrary channel in Landscape ### Mobile handoff @narner - Working on getting new DM architecture implemented in iOS app (https://github.com/urbit/pocket/issues/24) - Sending a new message and updating the view - Question: For the sending a new message flow, where we add a node at index/ship/timestamp, that timestamp is whatever the most recent message is, right? - @liam-fitzgerald: It's right now. - After is the work Ed specified in https://github.com/urbit/pocket/issues/22 - @jalehman: What's the scope of the engineering sprint? What are we working on? - Unsure, but we usually decide during cooldown week. - @liam-fitzgerald: Modernising codebase with newer hoon style; graph-store scry paths need a refactor; we gotta get rid of location parsing crap in group feed; JSON mismatches for ships - @jalehman: What about testing? - @liam-fitzgerald: After this Jest stuff, we basically have industry standard suite of tests, barring end-to-end stuff on FE; on back-end, what I'd like to do is revive Aqua and write tests for Aqua and prevent regressions on Hoon stuff - Most of the Hoon errors occurs at the integration level - Asked @ixv to implement the push provider for mobile, so he may be tapped out on that and may not be able to do import/export work either - @jalehman and @matildepark to spec the system for that; maybe there is wiggle room to specify the import/export project

[matildepark]

Sprint + Date: 1.8 (Abellaite), ~2021.6.1 Recording: https://drive.google.com/file/d/1ca_p66uKVaOrhJ0OVO4PAPdgVDWPbFKd/view?usp=sharing

Meeting notes

- Deploy is tonight; all subscriptions are intact and we clean them up later ### Tlon @liam-fitzgerald - Couple of things on the trot — remove posts fix; new collections (had to re-implement imageinput) - Tested migrations, everything is fine, more or less - On the Janeway change made — may need to give fetch permissions - **@ixv to see if he can change perms** - Possibly changing CI to detect changes and glob specifically @jamesacklin - Group DM PR merged - Need help making the colour assignment for each of them random - Need helmet to respect nicknames ### Design handoff @urcades - Presented design spec for Landscape Grid work; directionally it's going well, dictating what should be abstracted into developer APIs - Need to take one more pass with product heads for a proper design specification ### Mobile handoff @narner - Finished the new DMs work on mobile-side - Working on more design-oriented stuff for new profile designs and help+privacy stuff, more to share on Wednesday

[urcades]

Sprint + Date: 1.9 (Abastumani), ~2021.6.7 Recording: https://drive.google.com/file/d/185ALeyXgQuFKGT7s-0dz74RNc6M7qqbV/view?usp=sharing

Meeting notes

- Engineering Sprint Begins ### Tlon @liam-fitzgerald - finished new collections - tested on mobile, looks good - remote content is heavily refactored - only thumbnails are fetched instead of iframes - shortcuts fixed - finishing API hooks to get started 1.9 @tacryt-socryp - perf improvements are going out in next 1-2 days - minor updates pushed out last week, major kernel updates forthcoming - faster type printer for debugging - cleanup work, addressed in-code notes for moving stuff elsewhere (to the kernel) - looking for review on cleanup PR: moving *-store code to scry code - have yet to work on scry-arm in graph store - wanted to pass over other works first - Liam Q: publicly flag changes to broader community, re: cleanup PR? @jamesacklin - filed low-hanging fruit PRs today - will review external contributor's PRs ### Design handoff @urcades - Latest updates to Landscape Grid, Design Projects Solidified - Tour of latest Grid interfaces - Liam Q: Grid security model? App Sandboxing? Performance - ~tacryt: We'll need to experiment with a variety of models before we land on something correct - Liam: Sandbox will need to incorporate store/connection management, etc. - R&D for multi-windowing aspects of Grid work — can be handled in Port? - ~palfun Q: dm sidebar sorting? - Liam Q: Collections — get rid of notification counts/display for more calmness? (yes) ### Mobile handoff @narner - Last week: Profile UI updates, bugfixes re: new DM migration, S3 credentials - Color system for sigil colors in DM text - Work with Liam to iron out remaining DM migration bugs - Questions re: accessing settings-store re: CalmEngine - Isaac: working on push notification stuff this week, will connect with Nick about this

[matildepark]

Sprint + Date: 1.9 (Abastumani), ~2021.6.14 Recording: https://drive.google.com/file/d/162eXr8ut6ESu2sqI0MrvIyc0s4V6JFAD/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Miscellaneous cleanup last week - Starting on #938 this week - Remove the actual cached in state validators in various places since it’s no longer necessary - https://github.com/urbit/landscape/issues/857 - https://github.com/urbit/landscape/issues/850 @liam-fitzgerald - Did #940, now in review - Paired with @narner on iOS app functionality over the line - Did performance work on transitions and stuff like that, getting render times down - Next up is #941 after wrapping up urbit/urbit#4998 and urbit/urbit#4996 @jamesacklin - Working on the unjoined channel experience — if you switch to a channel you haven’t joined, it will pop up a modal, not a blank page. (#868) - Need help — talked with @liam-fitzgerald on joining resources, setting a flag in the store state that says “don’t receive notifications” - Asked @urcades about whether we pop the modal on cross-group references, so that references to a group you’re in but not that channel, you could just open a modal. ### Design handoff @urcades - No news on Landscape-relevant news. Working with @g-a-v-i-n on Grid MVP. - As we move apps into standalone formats, will be opening it up on Figma for people to give feedback on. ### Mobile handoff @narner - When we were processing channels in a dm-inbox, we weren’t taking into account comets, which caused some strange issues. - Was using an instance of a BigInt in one place. - Currently, Pocket is in design review. - Did work on settings section, fixing up design changes on welcome screen (urbit/pocket#31), updated the assets to vectors (urbit/pocket#34), sidelining urbit/pocket#28, and fixed urbit/pocket#27. - Currently working on interface for creating DMs (urbit/pocket#26). Might need some guidance on structure for creating a new DM based on someone’s ship. - Worked with @ixv so that he could work on notifications. - @Fang-: @liam-fitzgerald can you review terminal PR? (Yes.)

[matildepark]

Sprint + Date: 1.9 (Abastumani), ~2021.6.21 Recording: https://drive.google.com/file/d/1H6bRIR5Wj24HpqHAPeVQwY4aFeNzNySA/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Cleaning up the scry interface, documenting it now (urbit/urbit#5030) - LF: Might be used to do this in a way that TypeScript documentation generators can handle for us, skimming over some of my comments in `release/next-userspace` in the hook apis? - Link incoming for @tacryt-socryp … @liam-fitzgerald - Touching up all the various PRs made so far … - Spent a bunch of time on collections (urbit/urbit#4996) and ensuring the content types look good - Spent some time playing around with setting up an embed provider, akin to how we do the BTC provider, which would let us get a lot more embed providers, which would be nice ### Design handoff @urcades - The windowing, multiplexing, cross-data referencing story and everything that informs it is the key constraint that’s missing in defining some of the design work done in the Grid project; what we actually go with gets radically altered based on this constraint - On the document shared for strategy, it’s been outlined to surface up things like “Pocket” since it’s what Urbit is providing and there’s an idea that rather than a web app, it might link to an apk download page or the like, or the app store or whatever; what constitutes Grid is less a thing meant to embody an operating system and more of a method for surfacing what’s connected to the Urbit - What we surface on Grid could even be terminal apps, and we could surface it up in something like webterm or native terminal ### Mobile handoff @narner - Shipped a few TestFlights, implementing fixes from Ed’s filed issues - Finished up some of the new behaviour for pinning and unpinning — @urcades urbit/pocket#39 ready for review with the current TestFlight build - Right now, working on behaviour around message ordering so if I’m in Landscape and I send someone a message it should populate it back to the top of the list in Pocket; whatever is handling timestamps wasn’t updating properly, and that would solve urbit/pocket#32 and urbit/pocket#33, and from there it’s fine tuning - @liam-fitzgerald: Running out of stuff to do … will take on @jamesacklin priority twos and re-sweep the project board - @Fang-: The group feed layout, when we started out, was a threaded model, and currently it’s a reverse chronological single feed; were we going to get a setting to let you configure that? (Maybe #939). - Adding option for the primordial modifiers, so to speak … filed as #988. - @liam-fitzgerald: We should flag the scry changes one or two weeks before they go out, so the community can amend for it. @tacryt-socryp

[matildepark]

Sprint + Date: 1.9 (Abastumani), ~2021.6.28 Recording: https://drive.google.com/file/d/11NFbBh0pGSjawhq-mKCqy7lKjg_oN5o5/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - More Bitcoin issues than previously thought, infinite loop and other issues (https://github.com/urbit/urbit/pull/5061), @ixv is a champion, should be good now, doing another deploy today or tomorrow - https://github.com/urbit/urbit/pull/5030 after - BTC took precedence because it completely wrecked his ship @liam-fitzgerald - Paired with @Fang- to get webterm compatible with new API hooks - Did the PR for the music player — https://github.com/urbit/urbit/pull/5058 - @urcades PR'd the pause button into indigo-react, waiting on that merge and release - Unflaking Storybook for LinkDetail - Maybe pair with @arthyn to peek at https://github.com/urbit/landscape/issues/559 - Two notifications issues related to optimistic state management - Discovered it with @urcades - LF: @urcades have you gotten progress on embed corruption artifacting? - https://github.com/urbit/landscape/issues/1000 still persists @arthyn - Started looking at https://github.com/urbit/landscape/issues/1002 - Reproduced, saw how the data is flowing, think we need to change some things on how we do the reply message injection - Otherwise starting R&D soon and onboarding ### Design handoff @urcades - Presentation milestone next Tuesday on an MVP of Grid - Encountering cases that will be nice fuel to the R&D pipeline - The work on urbit.org is a side story ### Mobile handoff @narner - Last week, did a lot of work on unread states between Landscape and Pocket, so that if something gets marked as unread in Landscape, it will be unread in Pocket; similarly, if you have an unread DM in Pocket and you click to open it, it will be marked as unread in Landscape as well (urbit/pocket#41, urbit/pocket#52 ready for review @urcades, urbit/pocket#51 waiting on review @matildepark) - Trying to get DM ordering issue fixed, and load avatars properly - Was working on it this morning, noticing that sometimes getting default timestamps from back when Hark was being migrated to a new format, want to do a sanity check with @liam-fitzgerald on the timestamp locations, and that's the last big behavioural problem - From there, design, QA changes - Working with @ixv on getting notification flow set up — we have baby notifications, so if a notification gets posted and app in background, you'll see a banner - @ixv: When can we expect a next-userspace release to go out? - Scrying for ship and Bitcoin deploy tonight - @ixv to talk to @narner about getting ship name from scry, not URL (urbit/pocket#54) - @Fang-: Bridge may already, it calls to who.json on the root of the address - It doesn't look there any longer, so reimplementing who.json might be better solution - @liam-fitzgerald: It seemingly got pulled out with Ford Fusion landed, maybe? @ixv: @liam-fitzgerald, want to pair on updating archive stuff after aqua lands? - LF: sure

[matildepark]

Sprint + Date: 1.9 (Abastumani), ~2021.7.6 Recording: https://drive.google.com/file/d/1lHR5FGvJgKtxBb73PGY__ac313cX2j5e/view?usp=sharing

Meeting notes

### Tlon @tacryt-socryp - Did work on graph scry stuff today - Have a PR ready for reviewed and looked over tomorrow - Pushed out a little PR today that lets you set the interval for BTC provider to ping its full node https://github.com/urbit/urbit/pull/5083 - Increased default ping time from 30s to 60s @liam-fitzgerald - Doing bug fixes and stuff, nothing that interesting - Migrated us to Lerna in a PR https://github.com/urbit/urbit/pull/5073 - Thinking about embed stuff — @urcades are we good for all reviewing all the ways we do remote content? - Merge bitcoin-wallet into next-userspace, manually merge the two Lerna PRs into next-userspace, and then deploy to master afterward, if everything works - Optimistic marking as read (#1030) and collections cruft breakpoint last remaining, then can deploy, alongside https://github.com/urbit/urbit/pull/5070 ### Design handoff @urcades - @g-a-v-i-n and I shared our work, drew a boundary on the maximum capability we want to embody on the Grid work and showing a more minimal experience next ### Mobile handoff @narner - Was working on maintaining the eventstream in the mobile app - Was also working on urbit/pocket#49 - urbit/pocket#56 needs to be tested @ixv - Looked a bit at urbit/pocket#57 for accepting new DMs, the data parsing is done, UI needs to be done - @urcades: I'm going to write up critical paths for testing tomorrow - @ixv: @liam-fitzgerald, when are you going to deploy? - Getting a coffee, then doing it.

[matildepark]

Sprint + Date: 1.9 (Abastumani), ~2021.7.12 Recording: https://drive.google.com/file/d/12hH8gHhhwfmVR9neeA53hkIpu45xg7Ww/view?usp=sharing

Meeting notes

- QA starts — put https://github.com/urbit/urbit/pull/5091 and https://github.com/urbit/urbit/pull/5030 into a dated release, test several times over the next two weeks for both action and migration - Another PR incoming for the unread stuff around notebooks so the permissions allow you to remove revisions from Publish that make it readable - Breaking changes: @liam-fitzgerald to provide breaking changes notes to urbit-dev by mid-week - @matildepark to make a dated release branch for the breaking change work ### Tlon @liam-fitzgerald - The aforementioned PR for notebook unreads - Another PR for a bug fix omnibus; wrote tests for HTTP library, found bugs stopping us from reconnecting sometimes - Got most of the bugs assigned in the PR - Was thinking of having a crack at graph restructuring stuff — https://github.com/urbit/landscape/issues/829 - Deploy tonight for the comments fix @arthyn - Wrapped up tab-based Grid prototype - https://grid-prototype.vercel.app/ - What happens with Port and Grid is an upcoming line of work — the M1 warning isn't working - Working on keyboard global Leap shortcut for Port - @liam-fitzgerald: Is the overheard of a new JS object large? Would we be better off storing every single node with its fully-qualified index instead of this recursive business in https://github.com/urbit/landscape/issues/829? ### Design handoff @urcades - No major news to report to Landscape, other than that feedback is being boiled down into immediate lines of work for the next two weeks, so that a more compact prototype can be built - Comparing idealised Grid maybe in Q1, and Landscape now, we should bring to bear some of the affordances we'll need to make software distribution more obvious in an incremental fashion - Compiling this into actionable work; if there's last minute feedback, let me know - @arthyn, can we get a Port UX meeting going? (Sure.) - @urcades to slate some higher priority polish ### Mobile handoff @narner - Got three issues tested and completed (urbit/pocket#44, urbit/pocket#45, urbit/pocket#46) - Got UI and reaction work finished for urbit/pocket#57 - Started working on poke logic for that, but it doesn't seem to be propagating the diff, talking to @liam-fitzgerald on that - Working on urbit/pocket#48, where if Pocket goes into background mode, it can lose its connection - The issue seemed to be that we need to subscribe to a new channel URL; but the issue now is that it'll go through the resubscription flow (urbit/urbit#5048 blocking) - Once that merges, it may be able to resume automatically, or alternatively you could resume it manually by passing the last event ID in the header - @liam-fitzgerald: On an actual app, you're not constrained by how many channels you can have, so you can open a channel for each subscription and clog less often - Push notifications working with Twilio, figured it out with @ixv - Was looking at group chats, ran into an issue where, if we open the eventsource, there's no guarantee we'll finish processing metadata updates before getting a response from group-update, so ideally what would be great is that we can do some sort of scry for group chats once we get this initial metadata-update - @liam-fitzgerald: In Landscape, we replicate the structure of the agents — we have a metadata-store and a group-store, and we join them up at render-time. I think what @narner is doing is grabbing the data and putting it into one group-chat object; but Landscape doesn't know what a group chat is, they're just chats in specific configurations we render as group chats - For the minimal use case, I think it makes sense to do these group chats as an object; but if we make one general purpose mobile client, it doesn't make sense - You should just be careful with the async stuff — when you receive that initial metadata or group update, _then_ fire off the subscription to the other - Alternatively, you could, while you're still receiving updates, validate that they're the ones that need to be there

[matildepark]

Sprint + Date: 1.9 (Abastumani), ~2021.7.19 Recording: https://drive.google.com/file/d/1s-pBLC1vU_BxdcRJyETl9v1J7JRfqV4p/view?usp=sharing

Meeting notes

### Tlon @liam-fitzgerald - Cleaned up form handling in https://github.com/urbit/urbit/pull/5115, there's a unified way to do it so that everything doesn't get wack in terms of desyncs - Thought about it and figured out why connections are mysteriously dying and it's likely to do with timers being throttled when the tab is inactive, so acks aren't being sent and we're clogging and not resubscribing after the kick - Want to change the Airlock interface to unconditionally ack when we have 20, 30 messages and rip out the debouncing logic and it makes everything really unpredictable - @ixv, what's up with import/export? - @ixv: Code's changed a bit, test each app one at a time, fix them; it doesn't do hark, so might want to do that, and what other apps do we have? group-view doesn't need a state migration - Migration testing tomorrow, fixing import/export today - @Fang- is there a reason the clog logic isn't per subscription not per channel? - @Fang-: No reason comes to mind; there may be a case like "multiple subscriptions, but only one is blasting out the updates" in which you don't want to kill the entire channel ### Mobile handoff @narner - Have some issues ready for review from TestFlight build by @urcades — urbit/pocket#57, urbit/pocket#58 - Pairing with @liam-fitzgerald after @ixv meeting @Fang-: @liam-fitzgerald, messaged you about graph validator behaviour, where it obliterates the index; why does it do that? - @liam-fitzgerald: It normalises the timestamp. - @Fang-: But it's on the sender side. - @liam-fitzgerald: No, it's push-hook on host side. - Poking @tacryt-socryp on the implementation here

[jalehman]

Date: ~2021.7.26 Recording: https://drive.google.com/file/d/1cRf1nC4oXRDQhKl8TV1akRFDZ6KrpyAX/view Sprint: https://github.com/orgs/urbit/projects/15

Weekly Goals

• Read the issues and look at the designs. Leave comments. Become familiar with what we're doing, and voice your opinions. Let's make sure we're all on the same page.
• Decide on names for things. We'll use this issue to come to consensus about what to call the various new things that we'll be building.

cc @urcades @g-a-v-i-n @philipcmonk @belisarius222 @liam-fitzgerald @ryjm @arthyn @Fang-

Meeting notes

- [Working design prototype](https://www.figma.com/proto/VxNYyFRnj8ZnqWG54VbVRM/Landscape-(Grid)?page-id=16%3A4544&node-id=68%3A12068&viewport=215%2C-1272%2C0.25&scaling=min-zoom&starting-point-node-id=68%3A12068&show-proto-sidebar=1) - Keeping Landscape as-is (tile in Grid) is contentious - Primary concern from @urcades is that keeping leap and profile as-is in Landscape would result in confusion, leading to support and comms issues, for end-users—it's better to "rip the bandaid off" - Reason for making minimal changes to existing Landscape is predominantly driven by scope, not experience, although ensuring that there's minimal confusion is obviously important - [More context in Figma](https://www.figma.com/file/VxNYyFRnj8ZnqWG54VbVRM/Landscape-(Grid)?node-id=6%3A10013) - Software installation could be surfaced more prominently - @jalehman Leap could prompt by default to *Install software*, which prompts you to search for a ship, and then shows software after choosing—this would be better than burying it as a key combination after looking - @liam-fitzgerald pinned applications to profile would be a ~10 minute addition - @liam-fitzgerald there's other critical infrastructure work to do that isn't directly relevant to grid - @arthyn updates on sharing state / connections across tabs - doing state management in apps, subscription management in the service worker - @liam-fitzgerald what happens if the channel breaks? - good question—@arthyn and @liam-fitzgerald to sync up on this - @arthyn it's worth writing a skeleton app manifest, should work with @belisarius222 on this - @jalehman what of the current landscape is specific to it and not general-purpose userspace infrastructure? - @liam-fitzgerald metadata store is largely landscape-specific - @ryjm as we decouple things we should revisit names - @liam-fitzgerald likes the name `treaty` more than `manifest`, and it doesn't carry any programming baggage with it

[jalehman]

Date: ~2021.8.2 Recording: https://drive.google.com/file/d/1aJQ6myh__mSByCceLxJm8DmS7vcxyTlh/view?usp=drive_web Sprint: https://github.com/orgs/urbit/projects/15

Meeting Notes

@belisarius222 - Been messing with kiln, drum, hood, gall in order to run |install - |install asks kiln to install a desk from another ship, read manifest file, keep up to date - Based on manifest, determines agents to start and connect to drum for CLI access - %base desk has hod and dojo, landscape desk has chat-cli connected to drum - changed the way gall is asked to start and stop agents, drum is no longer involved (responsibility moved to kiln) - as of tonight, two different desks: %base and %landscape - both run correctly, there are issues with %landscape—this can be turned over to someone (@liam-fitzgerald?) - you can run |install and |uninstall, works with desks on foreign ships, possibly with messed up error handling on local desks - has tested with a galaxy that can host both desks, and another ship that pulls from them - not done: - kernel kelvin checking stuff (bug); just don't change the kernel, and it can be handed over - gall doesn't use cache when rebuilding agents, so it's slow; fix has been written and may be correct - wants a %base-dev desk which base and landscape pull from; this would have shared files - right now it's easy to end up with conflicts when changing a file that both depend on - we'll probably have to iterate/experiment with workflows - can start working off of Ted's landscape branch - should write up instructions on how to work with this stuff - ted/dist branch — needs testing of landscape running across desks to ensure that everything still works - would be good to begin working on the version mismatch logic from kernel to interface ASAP since it's one of the most complicated pieces - we can only really simulate at this point @liam-fitzgerald - has been writing a new agent docket which is new userspace manifest - does glob fetching and maintenance of frontend state, e.g. serving globs - has pulled out a lot of the "serving from clay" stuff from fileserver - makes build process a lot easier - ported metamask as a demo - it's easier for now to serve one tile per desk, so they can be namespaced by desk - open question: how much should serving of tiles be to serving of packages? e.g. metadata - what metadata is even specific to |install? - list of agents to start, list of agents for drum to connect to, kelvin version - everything else should be grid-specific - @Fang- should we maybe include some kind of human-readable name, e.g. debug messages? - @liam-fitzgerald we should have a standardized form of package management outside of grid - mark agrees, general agreement from everybody really - WIP on this PR: https://github.com/urbit/urbit/pull/5136 - has merged file-server and glob into one agent - doesn't make sense to be decoupled if all it does is serve globs @arthyn - took @g-a-v-i-n 's grid prototype and put it into its own repo - this will become the actual grid interface - working on stress testing # of subscriptions we can handle by proxying event streams through a service-worker - so far so good - converting indigo to tailwind, since we'll be using that instead - why? - styled components is a performance hog and is hard to use - grid prototype was made with tailwind - @liam-fitzgerald landscape would benefit from this conversion, specifically virtual scroller - this will imply a major version upgrade to indigo @Fang- - allowing dojo to run generators from desks other than home - will get to this ASAP - @belisarius222 not super high priority, since there's a way to do this already with `=dir`, or `|cd` - good starting point for @Fang- and @ryjm - generator that creates a new, empty desk with manifest file and sys version (boilerplate generators) - one that's grid-specific and one that's not — should sync from `landscape-dev` and `base-dev` respectively - @belisarius222 clay can't create a new desk, it can only merge from other ones - maybe we should just change this? - @philipcmonk a way around this would be to create a new version of `|commit` - or, an "origin" desk that never changes that you merge from. - needs more thought @ryjm - just working on khan, has been reading through everything - will work with @liam-fitzgerald and @belisarius222 to get an overview - work with @Fang- on developer stories (#1079 and #1077) - @belisarius222 would also be good to start writing docs while this stuff is fresh - use %seer as an example @urcades - fleshing out some more details on the current MVP - e.g. software searching, app suspension, system preferences - @belisarius222 not sure we want the ability to pause kernel updates in reaction to a failed kernel update due to blocking desk(s) - "wait until they're compatible" is a better way to express this - suspension to avoid compatibility issues should automatically result in being reinstated - turning off kernel updates is a feature we want, but not necessarily in critical path - we might want to be able to turn off syncing from any desk independently of suspension/nuking - (this is covered in this issue #1092 ) - there are indigo concerns with work that's going on in Bridge, want to ensure we don't end up diverging @g-a-v-i-n - what Ed said - last week was moving, now back. yay.

[tacryt-socryp]

Does it make sense to try to update our whole styling framework in such a massive way when there's a time crunch for 3rd party software distribution? @arthyn

[arthyn]

@tacryt-socryp yeah I do, given the state of what was actionable vs what's already been done and what needs to happen. I'm already halfway through and since this isn't critical it can be dropped/picked up at any time. Contextually, a lot of the things I would have been working on last week were dependent upon us finishing out our discussions/decisions so I made a call to work on this once my subscription work got to a solid point.

This update is not really that massive if we aren't tackling the full conversion of current Landscape. I'm mostly trying to allow us to use our current Indigo components in the Grid interface while still leveraging the work that was already done with Tailwind and avoiding instant tech debt on a new project.

[jalehman]

Date: ~2021.8.4

Meeting Notes

@urcades - finishing some design stuff with gav @ryjm - covid brain - [this needs to be preserved] @Fang- - writing stray notes about developer facing process, still solidifying - had a dream, and in it Grid was real and good. @liam-fitzgerald - got previews working, as in, "here's a package to install", which generates a link that you can click to install. dope. - code organization is messy, wat do? - keep monorepo, split after functionality is achieved - how do we want to structure the manifest/docket/bill/whatever? - Here's @belisarius222 's format for Landscape: https://github.com/urbit/urbit/blob/4e2fd0dcd04d738cafd9eb7a376b97ca436e57d5/pkg/landscape/desk.bill - And here's what @liam-fitzgerald has been working on: https://github.com/urbit/urbit/blob/8e9a6dfdb53772d84814689bb5b8875b45375551/pkg/landscape/grid.docket - @philipcmonk There should probably be version numbers @arthyn - Finished Indigo -> Tailwind conversion, will get @liam-fitzgerald and @g-a-v-i-n 's take - Started working on Grid tile menus

[urcades]

Date: ~2021.8.9 Sprint: Landscape Baikal

Meeting Notes

@arthyn - issue splitting - everything is in user story requirements - Intent has been set to get issues broken down into actionable tasks - we should merge liam + ted's branches together asap to push progress — @belisarius222 : let's just use liam's desk for now as the default — both will pair tomorrow on work to get the branch completely up to date - finished baseline tile interaction stuff - added variables to tailwind conversion to indigo, will support theming - last week, spoke during roundtable on how to evolve indigo - will be porting real indigo over to the product @urcades + @g-a-v-i-n will pair to figure out how to mitigate performance, implementation, etc. for the MVP @belisarius222 - tried getting gall to build apps faster, it's complex -> tabled for now - listed in a new branch - tried setting up %base dev desk, isn't up yet, kiln needs to be modified in order to proceed - will be booting up another pill to continue this work - Infra Digest located [here](web+urbitgraph://group/~dabben-larbet/tlon/graph/~bitpyx-dildus/infrastructure-digests/170141184505192082140875664807444873216) — Please read this - advice (via digest) is almost ready to be used - ~~will be getting a pill for people to use tonight — @liam-fitzgerald can start using it first re: landscape splitting~~ - @liam-fitzgerald will be sending @belisarius222 a pill - may be turning over some of his ongoing work to @Fang- , @liam-fitzgerald , or @ryjm - fix |sync merge stratagem - %base desk updating work - kiln datatypes are available for cross-app notifications, pokes, etc. — @belisarius222 is asking people to react to them to push this work further. This work is crucial to get right to mitigate client updates, OTAs, etc. - need to explicitly break down issue #1092 for the sub-tasks contained within @liam-fitzgerald - is writing discovery + request stuff w/ clay - can't continue as clay currently stands - working on desk compatibility checking - running into weird SPA stuff w/ foreign desk name discovery - @belisarius222 : we might be able to enforce some things re: ^this in grid - testing out developer experience w/ real code/projects @ryjm - first task re: Landscape Grid could be porting over SRRS to grid (?) - finishing up khan @Fang- - may find some time to work on Landscape this week - may be pairing with @ryjm @urcades - designs for critical paths are largely complete from a mechanics perspective - will be thinking through the mobile iterations - will be thinking a few projects ahead, on the standalone products

[Fang-]

~2021.08.12 meeting notes

@belisarius222 - Merged master into `ted/dist`, @liam-fitzgerald wants that for docket work. Merge not fully functional yet, maybe? Almost done. - Along the way changed how hood tracks versions for its sub-apps: tl;dr the sub-apps do it themselves now. - Probably want to split those into their own agents soon. - Talked to @Fang- about base-dev desk. - @Fang- and @liam-fitzgerald now working on top of `ted/dist`. This is good, we need to iterate on developer ux for it to become comfy. - @liam-fitzgerald started working on grid agent & fetching manifests from foreign desks. - Talked to @philipcmonk about asking clay which desks it has, this is hard to do referentially transparently. - For now, probably just local scry & do the remote in userspace. - Should figure out how to fix `-A`. @arthyn - Been implementing more of the UI with mock data. Got most of the "leap" done. - Can browse through devs -> apps -> app manifests. - Posted progress to landscape channel. - Hope to finish the rest of system preferences, help, and support by end of week. Maybe do install flow as well. @Fang- - Figured out how to run @belisarius222's branch at the start of the meeting, going to work on base-dev desk flow. - Talking with @ryjm about developer ux later this week. @ryjm - Trying to port srrs to use normal html instead of sail. - Still trying to get kahn pr ready. - Understand @liam-fitzgerald's PR. - Working on making kahn support threads as a side-project. - @belisarius222: we should talk about booting urbit with landscape. Couple approaches here, might have overlap with kahn.

[urcades]

Date: ~2021.8.16 Sprint: Landscape Baikal

Meeting Notes

@belisarius222 - came with agenda - to @Fang- think through options for merge strategies for base desk work - @Fang- : symlinks seem nice for this work - last week: bogged down trying to merge work, encountered issues - spoke with @ryjm about deleting files from clay @liam-fitzgerald - rough prototype of grid functionality - currently writing bindings for kiln - converting mark data to json - what's next?: a ton of PRs to merge + refine - after this, was going to test kelvin versioning in clay - app icon stuff isn't done just yet @arthyn - merged everything that liam had done together (minus latest code) - everything looks good, is contained in `grid/initial-ui` branch/PR - still working out "installing an app" on his branch, needs @liam-fitzgerald to check out - once today's work is merged, it should "just work" @ryjm - understand the necessity of the js glob for larger apps (landscape), but small apps don't need it — what's its purpose? - @liam-fitzgerald : js commits nightmare - issue isn't with glob itself, it's just opaque and not part of build process, don't want to force devs to use it - @philipcmonk globs will cause issues further down the line @philipcmonk - re: entire glob conversation: khan - we need to document how to use/make globs - for most people, unless you're doing something incredibly image-intensive, you'll be able to host apps off your ship @urcades - no designs are blocking implementation - we still have some detail to substantiate guest - meeting with @belisarius222 to chat about software distro structure/approach

[tacryt-socryp]

Tombstoning for Clay becomes far more important in a world where you are storing version histories for many desks. Worth noting, not a blocker. @philipcmonk @ryjm

[ryjm]

Took some notes on it here: https://roamresearch.com/#/app/the-forge/page/AoO4WLZpH

Ted had me convinced it wouldn't be that hard, but probably better to think carefully about this. At the very least we should write a proposal.

On Mon, Aug 16, 2021 at 8:37 PM, L @.***> wrote:

Tombstoning for Clay becomes far more important in a world where you are storing version histories for many desks. Worth noting, not a blocker. @ philipcmonk https://github.com/philipcmonk @ryjm https://github.com/ryjm

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/urbit/landscape/issues/792#issuecomment-899908075, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAPYHEZQM5PMQ2AATO2RDN3T5GVNJANCNFSM43GEQXAQ .

[jalehman]

Date: ~2021.8.18 Sprint: Landscape Baikal Designs UI Prototype

Meeting Notes

@liam-fitzgerald - work on discovery functionality per the new leap - to meet with @arthyn re: state management @arthyn - found and fixed a bug with search - working on recent apps and developers portion of search - use of localstorage for search is good - next: buildout of system preferences and help & support screens - will need help on notifications - @liam-fitzgerald will be best to wait on this until notification backend is made more generic @Fang- - Just Bridge work this week, completion of which should free up more time - Next is base-dev desk with symlinks @g-a-v-i-n - Working with @urcades to think about future projects - e.g. Leap API, Publish, Messages, Profile, Contacts, Distribution Centers, more universal system settings - Still lots of detail work to do on the critical path @ryjm - Porting ~%srrs~ %seer - Working on docket

[urcades]

Date: ~2021.8.23 Sprint: Landscape Baikal Designs UI Prototype

Meeting Notes

@belisarius222 - agenda: - put link unfurling for apps somewhere in the critical path? - this has been implemented as part of the app's "Info" view - currently shows up as a raw URL, does it need styling - @liam-fitzgerald and @urcades had been collabing on universal references, can surface up the latest work for the requisite issue: #1101 - ask Liam what he meant by needing to make notifications generic to interact with frontend - liam has been thinking about this, group is unsure whether any work has been done on it - designs needed: generic notifications row data - todo for @belisarius222 : I need to add something to kiln to turn off desk updates, or we need to remove that button from the frontend - @Fang- - talk about state transitions for app installations and updates, and pending states -- both frontend and backend (sequencing glob download carefully and publishing that state to the client) - app installs might take many seconds to minutes to tens of minutes - @urcades designs have been completed for a single on-tile indeterminate loader for an MVP, could evolve to be determinate, or send hark notifications that an install/update has been completed - @belisarius222 : this will largely need to be adapted depending on how it actually feels to download an app — depending on how slow/fast it is, this item will need to be prioritized differently - @ryjm should keep the sequencing of glob state in mind wrt this item + his work - if a ship is disconnected, we should treat the 'waiting' state as a persistent one, could help to alleviate the perceived annoyance with app install/update states. For prior art, see this issue (especially the last few comments + interface designs): #179 - Liam: what's the difference between "site" and "glob"? - what remains to be done to separate the bitcoin wallet into its own desk - Liam and Hunter: what's the status of update notifications from kiln? - migration of work to `master` - coming close to being able to run a testnet for QA @ryjm - globbing updates: just started on it, basement has been flooding @liam-fitzgerald - is being questioned by the gov @arthyn - https://github.com/urbit/urbit/pull/5158 - "graft api to prototype" @urcades - MVP of "App Share" reference - MVP of more sane/urbity installation/update state - MVP of async 'jobs' queue interface - MVP of critical paths QA assessment doc (see this [example](https://docs.google.com/document/d/15dWyuJ4mUOONJvdVsOnFb6QLFKVkX13AdlBurTDfq8w/edit?usp=sharing)) @jalehman - Neil David is demonstrating the UI/development/software distro, which should hew closely to our "developer user stories", so he'll need to learn how to do everything perfectly for demo purposes - @ryjm + @Fang- publishing + development story status - @belisarius222 we're blocked by unknowns re: globbing work — @ryjm "globbing is largest attack vector for bad dev experience" + "dev desks" - @belisarius222 we should make a video of the development/migration/publishing processes - @Fang- + @ryjm will set up the testnet, ideally wed/thurs afternoon @Fang- - App Uninstalling: How do we feel about `+on-fade` or gall agents?

[liam-fitzgerald]

Apologies from my absence:

ask Liam what he meant by needing to make notifications generic to interact with frontend liam has been thinking about this, group is unsure whether any work has been done on it designs needed: generic notifications row data

Notifications needs to have a generic datastructure so that we can render them in grid without having to dynamically load custom JS components (both a security nightmare and out of scope). There's a first pass at nu-notifications (here)[https://gist.github.com/liam-fitzgerald/b763a144b57fc98f79816cdbeafbd2b8], however it relies on scrying to render the actual contents of the notifications, which is a fairly significant UX issue. Alternatively we could replace the scry path with something like this

|%
+$  header 
  $@  title=cord
  [author=ship action=cord subtitle=cord]
::
+$  action
  [title=cord status=?(%primary %destructive %$) link=cord]
::
+$  notification
  $:  =header
       body=cord
      =time
     actions=(list action)
    from=[app=term =path] :: Partially qualified scry path, to fetch any more data

  ==
--

This is a quick sketch, largely based on iOS notifications. We lose any custom formatting of the body of the notification, which I could be convinced is actually a good thing.

talk about state transitions for app installations and updates, and pending states -- both frontend and backend (sequencing glob download carefully and publishing that state to the client) app installs might take many seconds to minutes to tens of minutes @urcades designs have been completed for a single on-tile indeterminate loader for an MVP, could evolve to be determinate, or send hark notifications that an install/update has been completed @belisarius222 : this will largely need to be adapted depending on how it actually feels to download an app — depending on how slow/fast it is, this item will need to be prioritized differently @ryjm should keep the sequencing of glob state in mind wrt this item + his work if a ship is disconnected, we should treat the 'waiting' state as a persistent one, could help to alleviate the perceived annoyance with app install/update states. For prior art, see this issue (especially the last few comments + interface designs): optimistic store updating #179

This has all been largely done with the latest two PRs, loading is treating the same as a suspension or error state. Fwiw downloading landscape over ames w/ local ships takes quite a while, especially given the build time.

Liam: what's the difference between "site" and "glob"?

%site is just a bare link to the ship, bypassing globs and whatnot. Desks that declare a %site path should bind to %eyre at that path, and handle all HTTP requests themselves. This is designed to make server-rendered sites viable (cc: @Fang- )

what remains to be done to separate the bitcoin wallet into its own desk

Basically nothing, the FE build process is done and the only other work is moving files into their own folder and writing a docket and bill file

Liam and Hunter: what's the status of update notifications from kiln? Getting close. My latest FE PR add bindings from kiln, but at the time they were written, the update notifications were still buggy.

my agenda for week:

• Address/merge outstanding PRs, shepherd everything into a workable state
• Integrate kiln further, test versioning machinery etc.
• Refactor settings-store and contact-store FE state management out into their own library.
• Drop settings-store and contact-store into grid, via aforementioned lib

[tacryt-socryp]

@liam-fitzgerald As you can imagine, I still advocate the more limited iOS-inspired API you specify above. I think it isn't worth opening the can of worms of "custom component rendering based on data fetched from a scry" yet, and don't think there's a pressing reason to tackle such a large problem at this juncture.

[urcades]

Date: ~2021.9.1 Sprint: Landscape Baikal Designs UI Prototype

Meeting Notes

- **Agenda** - check on testnet - status update from Jake on glob - evaluate next glob steps: integrate Mark's file upload - front-end: Hunter has an open PR - I'm working on getting the dist branch in good shape, including *-dev desks - migration has had some progress - check on kiln kelvin upgrade bugfix - I found a problem with the developer update pattern we've been thinking, where they can't host future desks - Liam general status report - @belisarius222 - pairing with Jake to build glob stuff - next steps: jake integrate file upload code that mark wrote - will enable upload of glob files into dev ship - Working on `dist` branch, can be targeted for PRs. Has latest kiln stuff. - Did latest symlink work on `dist` , so expect those changes - Needs to speak to phil about issue w/ developers running bleeding edge kernels in order to host a future desk - @liam-fitzgerald - Wrote blob distribution work, trying to alleviate nodes being hammered by - How do we ensure sponsors get blobs before planets do? Can we prioritize receiving by stars/galaxies? Ted: No, this isn't available in Ames yet - hark store work going smoothly, needs input from Ed - @arthyn - Working with Gav + Ed on small interaction moments within the main Grid interface - Working with Liam on "merge into dist" strategy, getting work streams unified - @ryjm - Trying to get a working ship on testnet, still working this issue out - Currently on liams branch, may need to rebase - Ted: tl;dr, regarding changes, merge into dist asap - Ted: Integrate mark's file upload work into glob stuff, and merge into dist - Liam: can pair to bring it all up to date - @Fang- - Test Network: Anticipating changes w/ ames protocol version + ropsten - Blockers? - None - Misc. - Team needs world clocks - We can just build a world clock

[Fang-]

date: ~2021.09.10 two's a team edition

meeting notes

@liam-fitzgerald: - finished hark store stuff for grid and landscape - fixed bump and update-flow stuff in kiln - brought back grid permalinks - we gotta get funky with componentizing - currently in lieu of npm pkgs, we just copy-pasting state management code from landscape to grid - all that should be componentized - landscape bugs basically done - pushed & merged to @arthyn's pr - reimplemented lag notifications - kind of a poweruser thing, so being able to dismiss (instead of it persisting forever) is fair game - just dirty-polling rn, auto-dismisses if lag flag gets unset - made setting-store settings scoped per desk - treaty can now look up arbitrary treaties outside of its gossip mechanism - can give it a ship and a desk to manually discover now - today: - do bunch of deploying, maybe merge party, get everything into `dist` - need to link up with @urcades - need to start work on migration logic soon - for notification migrations, options are - nuke everyone's notifications - pull groups etc sur files into garden - hardcode types into migration code, self-contained `+on-load` - poke notification origins with the old notifications, asking them to send new-style ones @fang-: - was out most of the week - will set up migration/migrated testnet once code for that is usable

[Fang-]

date: ~2021.09.16

meeting notes

@fang-: - improved the globulator upload page. instructions, desks selector - jake: yes custom glob is important - jake: i tested the uploaded glob, and it works - worked on multi-desk pills and installing software from them on-boot - liam: developer experience side of this can also be soved for by a better `-A`. you generally end up copying in the state of repo anyway - ed: wrt ux, we should have discussion about how we market this thing @liam-fitzgerald: - close to finishing invites/notification, just need a couple more hours - reworked the boxing logic. new one feels much better, doesn't jump, clutter up, etc - tldr: seen | unseen | archived, clicking the notification and reading its source both auto-archive it - archive is left half-open, notifications get merged appropriately @ryjm: - got `%seer` migrated, just need to work on eyre bindings, really want to get it working with a sail `index.hoon` - wrote up the entire process of migrating: lots of annoying things, but will be mostly solved by multi-desk stuff - did realize: `%seer` uses a lot of libraries from landscape, was wondering how we were gonna handle that - currently just merged everything from landscape into seer - mark: landscape just needs a landscape-dev package - jake: for outsiders, it's non-obvious. not sure what right approach for clarifying is, probably just lots of docs? or maybe developer-facing scripts - mark: what about a script that parses for ford runes, recursing on imports? (^: - neal: once we have some things live, this will clear up - jake: instructions for getting frontend app are non-trivial (by urbit standards, still better than earth) - liam: [vite](https://vitejs.dev/) seems to have lots of support for various js frameworks, getting the proxy setup isn't hard, we should just adapt create-landscape-app to use vite, then you can choose whatever - liam: could update create-landscape-app to use my little dependency manager thing. then have a wizard for initial setup, asks for desks you depend on etc. - jake: migration vs starting from scratch is actually really similar - will share write-up on the process. - gonna try to migrate `%picky`, which doesn't have a frontend - liam: we don't really support (in the softdist spec) frontendless apps. each desk has a tile - jake: it actually works, you can just have a tile that effectively links to the tile screen - jake: being able to hide ui-less tiles for non-powerusers could go a long way - liam: not that much effort to add another property and/or filter if site & glob are both absent - jake: want to implement this & touch frontend, hunter let's pair - probably gonna work with joshin on kahn for a bit - bug: have to put glob mark into base desk for it to get caught, even though it's only in garden-desk, because ames can't find it otherwise - liam: in theory, probably a bug. gall should look up the relevant desk. there is a similar known issue with remote clay scries... @arthyn: - fixed a bunch of little UI things, PR'd them - finished hooking up system preferences - added support for pausing/enabling updates for individual apps in system preferences - in notifications, we still have to hook up the pause button - liam: pausing is implied by dismissing the notification - started looking at permalink embeds in old-landscape, might need liam's help misc: - hunter: what about qa - liam: other than hark stuff, should be mostly ready to go - liam: just need to tie everything up and then we can do qa. by latest we can do it by friday - liam: mark, status of webterm package pr? - mark: just needs glob - liam: to glob: boot fakezod, commit garden desk, =dir /=garden, -make-glob %... - liam: there was the fronted locking up post-install issue - liam: mark, having trouble booting ropsten ships for testnet - mark: use testnet binaries from `ops/softdist-testnet`

[urcades]

date: ~2021.09.20

meeting notes

@belisarius222 - Everyone read @jalehman 's email: we're deploying monday - CC everyone on his email and get them to test out the network - We want to have everything setup on the testnet today so we can get rest of tlon on tomorrow - Check for dangling PRs - @Fang- : Is building solid pills w/ working desks on it working? - "Yes", PR is waiting around - Bug with brass pills that needs to get worked out - working out limitations with spider/ames/pills - Q: How has the app dev process been? - Neal: Not nuts about multiple tabs - By end of week will have tutorial done, probably - Need to check gall apps and set one up w/ docket file - Feels as responsive as any other webapp - inspectability of webterm isn't the most reassuring @Fang- - nothing is impossible - working on compilation issues, which need to be figured out asap/now - state of ames x spider x dist: not using threads, using a generator, getting it out tomorrow - Q: Do we want to replace solid pill in repo with multi-desk solid pill? - A: @liam-fitzgerald we could retain lite boot functionality if we keep both @ryjm - working on documentation for app migration with ~tinnus-napbus - cleaning up Seer for migration to grid landscape Neal - example of a desk that can be used as a basis for migrating apps? - @ryjm "just looked at garden or bitcoin desk, might not be the best examples" - once seer migration is done with, it'll be a good example too @liam-fitzgerald - did deploy to testnet, so everything from `dist` is online - status of migration testing: - will be doing this on friday - on sunday/monday: we launch - none of the state migrations seem messy @urcades - assisting @arthyn with remaining UI issues - system preferences icons, real favicons, copy, onboarding/notifications, small visual bugs @arthyn - made PR for system preferences improvements on mobile - meet with @urcades about the above - naming is still messy, we gotta figure it out (cc @marisarow ) - lingering app moving issues (standalone webterm, etc.)

[urcades]

date: ~2021.09.22

Midweek Meeting Notes

@belisarius222 - Agenda - spinning agent install bug - when ted's tried to install landscape + uniswap, the install spinner lasts forever - clay/ford cache issue - migration status - is everyone on the testnet? - documentation status - open PRs? - other issues? - there may be an issue in the `dist` version of clay - this could be related to clay not killing the cache on non-base desks when kernel reloads - migration status: @Fang- and @belisarius222 have been working on it — has been a ping-pong process that will conclude either tonight or tomorrow - testnet QA process: - nearly all participants are active, still need to bug a few people to get on - waiting on receiving a large issue drop from @marisarow - get @baudtack to check out #1173 and recommend which zendesk email to forward to - remaining notification finessing for #1172 - @urcades still needs to QA mobile interfaces all around - #5238 changes to glob stuff, adjust data structure - documentation status - [technical summary draft](https://gist.github.com/belisarius222/fe93df670b776ebd276f9bd0d42f0b12) - @liam-fitzgerald working on some documentation for gall - @liam-fitzgerald might work on a 0-1 guide on setting up an app from scratch - @ryjm is working on something along these lines too ^ ^ ^ @urcades - slowness of following notification from homescreen into landscape - clicking through a notification and declining the resulting notification and clicking through the notification again - need an "app distro for dummies" guide, so we can test out non-zod app distribution - will provide @arthyn with some mobile design stuff for sysprefs - will check out strange viewport sizes @ryjm - was working on Seer, but is currently working out some %spider bugs - writing up docs on app distribution/publishing - annoying part of app publishing is symlink stuff @liam-fitzgerald - re: eternal spinner on install: may have pushed a weird blob - will need everyone on landscape team to send jamfiles for real/mainnet ships - will check out the ford bug

[urcades]

date: ~2021.09.27

Weekly Meeting Notes

@jalehman - official naming of the release still underway, by the way of marcom @belisarius222 - Deploy Checklist: https://docs.google.com/document/d/1SBhGUSmFrS9SM1AhEc3TVSuX8UPbhnCMwH-7m3qZsSo/edit - testing @Fang- 's "mark + kiln stuff" - kiln was listening to merges and commits, this double listening was messing things up - pulled in @liam-fitzgerald 's merges into `dist` - blocker: merging master into dist is proving to be troublesome, re: bitcoin work - we might want to write up instructions for merging real ships into the new ota testnet(?) - try and get master merge done asap, after call — @arthyn working on this - we should get ready to rev testnet after kiln fixes, jakes updates, and master merge @philipcmonk - would hesitate to commit directly to stars because then we'd have two flows, more points for breakage @arthyn - done with his piece of work fixing up BTC to be merged into `dist` - 'keep incoming' re: merging of work - only thing remaining to be worked on is showing error states for stalled installations @urcades - revisit critical paths walkthrough doc and update for any bugs/new edges we've witnessed @ryjm - tested pr for %spider work @liam-fitzgerald - testing out app link updates - how are we handling non %base + %kids desks? Neal - working on gall agent examples - is waiting on updates for how to deploy a docket step-by-step

[urcades]

date: ~2021.10.4

Weekly Meeting Notes

@belisarius222 - agenda: - dabben is private got slow, do we know why? - any more urgent fixes? - any other work items that got triaged from deployment? - add notifications back to groups - %garden-dev should re-export %base-dev - we should make a %landscape-dev desk - /mar/kelvin should not be required to be present in a desk, should be hard-coded in clay - +on-init crash should kill arvo event - make sure kiln doesn't desync from +take-download crashing - kiln has bug with local reinstall, should adjust dudes in +abet:vats or at least in +install-local - should namespace agents (and http endpoints) by desk soon - should have clay keep all marks warm (might address some slowness issues), then take mark warming out of :observe-hook - figure out which desk webrtc agents and graph-store should live in - get /tests working and passing again - make leap work for searching new ships for published apps - plan out next work items - if there's anything urgent + on fire, shout 'em out - "Auto Treaty" bug - @liam-fitzgerald has done this work, just needs to clean up and PR - Not having to switch between garden tab + groups tab to view notifications - related to tabbing issue - groups writ large will probably change quite a bit ... short term solves are needed, but will need to be considered a part of a broader spectrum of work. - |uninstall should actually remove globs - namespace agents by desk - working on the kiln cleanup work - needs to get ship online @Fang- - shipped OTA @liam-fitzgerald - working through re-instating notifications in groups, patching up the grid notifications issues - eject beginner's island code

[arthyn]

date: ~2021.10.12

Weekly Meeting Notes

- liam - profiling - hark/graph-hook taking a lot of CPU - ed wanted group title in notifications - not missing much by removing graph-hook, you can write your own - adapter from graph-store to hark-store - remove posts properly changes unread count - not retroactively - ted - josh said get https://github.com/urbit/urbit/pull/5311 merged asap - don't scry into other agents - scrying into kernel should be safe, be careful about order - mark: struggling setting up btc wallet into right state for repro - shouldn't rely on updates not happening in single event - lots of discussion about how to handle scries in on-loads safely - asking liam about hark store - space leak in hark store - ship inoperable from hark store updates - nuke hark store recommended routine - liam thinks he knows what it is, should have fix soon - mark says don't sweep problems under the rug lol - talk on technical roadmap - what kind of applications we can support - need bigger loom - joe has some small changes to double/quadruple loom? - could drop some graph store signatures - setting to delete old graphs - channel mode to delete msgs older than 1000 msgs - new mars from amsden may help - been working on far out designs for namespace maximalism - mark - mostly small things last week, btc-wallet pr - paired with ted for hotfixes of clay and kiln around on-load behavior when installing desks - kiln failed some scries on-load - agents weren't updated properly - fixed and merged - reinstated thread from desk syntax - this week - |new-desk - tried making a bark? for clay, couldn't get it on plane - looking into upgrade failures - should fix issues with glob uploading - uploading strange mime types - reports of crashing on writing docket - hunter - last week - fixing big issues in groups - a few bugs in grid - didn't get to uninstall/install issues - this week - fix ci because of broken types - grid fixes - urchatfm - pr of typed out icepond/switchboard got approved, should fix a lot of issues - more fixes to come after pairing with amsden - getting landscape faster - code splitting routes that aren't necessarily going to load with react.lazy and suspense - needs fresh hark store always - everything else stale - if redirecting delay all other subs - big discussion on install issues - could be install locking event stream and it disconnects so we don't get updated state - issues with heartbeat/connection logic dropping connections too early - are we down for meetings and discussions about internal stuff during assembly? - everyone seems to agree this would be good, loose and casual on timing though

[Fang-]

~2021.11.01

weekly meeting notes

@ryjm: - docket pr for combining site & glob - remaining work is addressing pr feedback & reworking types, all small stuff, probably done today - getting messages from people for dist ota aid & support - failure to stay up to date/retry is probably a (kiln?) bug, should file an issue - kiln errors could be getting swallowed by gall - going on vacation from nov 17th until dec 4th - task to fit in before that is doing rage-driven development - clay tombstoning could maybe be done within that time, if paired on with someone who already knows clay, but availability is low-ish rn, so better to do later - been writing an app that lets you move graphs between hosts, and restore old graphs. would like to get that done this week @belisarius222: - in the process of big a move, will be in europe nov 20th to dec 18th, then back in the states - probably too busy with that to write code, but feel free to ping for questions/talking - on standby as resource for pairing @fang-: - fixed `+cite` bug, pr awaiting review - looked into glob upload issues from safari - this is unsalvageable because safari flattens uploaded directories into a list of files - worked on building any number of arbitrary pills, pr awaiting review - can discuss deploying on ~zod or moon afterwards - galaxy ops - backport new webterm @liam-fitzgerald: - attempting to find group joining bug, it's misery - not sure _exactly_ what's going on, but do have a fix that addresses most of the issues - have a refactoring of the group code that makes it easier to debug and fixes some issues/usability - built an app that works with pals: a world clock - debugged weird race conditions in the ios app - put a pr in to make it easier to fetch data from urbit - notify: need to think about it. - notify.hoon should be contained in a desk that's pocket-specific - grid tiles should probably support arbitrary `%site` urls, so we can link to the app store or wherever - doing group invites work that relates to @urcades's group overhaul @arthyn: - did a bunch of port things - seeing many blank screens, so released direct terminal access to allow for workarounds - did meeting about port's future - looked into fixing group invites, but handed over to @liam-fitzgerald - got onboarded to hosting team, will be doing hosting things for a while - still need to get the service worker thing out the door, picking that up in between hosting work - may help us cut down on total amount of subscriptions - that's important because subscriptions are expensive in eyre - have questions about some of the requirements for this, might post in landscape channel - was trying to track down our npm situation - there _is_ an organization for urbit, @jalehman is the owner, just added @arthyn as a publisher - will update our packages @jalehman: - https://gist.github.com/tinnus-napbus/8f53d14da63d81d47178d66e7cbfc63b - we would need dev desks. we already have autopill though! - qualms about pill situation fix from joe and brendan - do not remember their qualms - we came to consensus that we should do the big naming shuffle. but: how? - we did come up with a reasonable plan. we should write it down (or just do it) - @ryjm is taking this - we should definitely do it sooner than later - need to announce beforehand, & communicate to support team - old commands won't mean the same thing etc - @tinnus-napbus should know about this for updating the docs - beginners island removal has happened, pr just needs to get merged misc: - people are seeing lag spikes in groups ui - send @liam-fitzgerald perf traces when you see this! - also seeing long ames spinning - some are also seeing groups ui just refresh itself suddenly, even when in the middle of typing a message - could be a memory leak - might explain (part of) the lag spikes too

[Fang-]

~2021.11.15

some discussion happened about the `+on-rift` pr for which i did not take notes. tldr: it's not urgent, so just going in with the next breaking release. we might want to do improvements to the kelvin upgrade flow in the mean time, should start speccing that out until we get time to do it. @fang-: - made vere support `--long-form` cli arguments, in preparation for specify-your-own-boot-sequence stuff - working on that this week @liam-fitzgerald: - worked on groups reliability, also within context of breaches etc, - non-breaking change, can roll out by hand to a moon for qa @arthyn: - got js builds for urbit/api things set up neutrally: can include them in either es6 style or old import - awaiting pr appproval, then bumping versions and publishing - should do the same for azimuth-rpc-client, but confused about contexts in which that gets used - might start this or next week on service worker stuff @urcades: - worked with @liam-fitzgerald on group joining ux - getting ducks in a row for broader ux update @belisarius222: - moving, not much work done - traveling until eow, back in action (and earlier timezone) next week

[jalehman]

~2021.11.29

@liam-fitzgerald - OTA - Getting push notifications actually working properly - Issues with threading model - This is in the %landscape desk - Adding a bunch of features to group joining, big cleanup - Group joining stuff is done, has addressed @arthyn and @Fang- 's review comments, needs another pass - On support rotation this week - Go through issues, tackle things that are forwarded @belisarius222 - Infrastructure issues that need to get resolved - need to get out gall request queue fix - ames flow killing - can possibly involve @yosoyubik in this - will also ping ~nortex - another thing for this year: project beachhead - scale graph-store using remote scry - goal is to make it quick and incisive @Fang- - Took a detour from improving the pill situation to make parsers in hoon.hoon 240x slower - WSTR strike force - Did some app development with ~rabsef-bicrym, made 4 hours of video content - backported new webterm but got stuck on js build shenanigans, liam's gonna assist with that @arthyn - Got 2.1 version of npm packages released - Tackled colored input bug in profiles, found a bonus bug - May get to service-worker this week @jalehman - What's going on with markdown parsing in notebooks? - Liam wrote last revision - Mentions in bulleted lists are janky, has to do with the way we persist markdown. Would be better to store entire AST - This is an issue for next year

[Fang-]

Closing. 2022 coming up. New year, new org, seems nice to start with a clean meeting log.