<re_irc>
<@ryan-summers:matrix.org> Dang, I actually liked using bors to trigger CI for long running tasks that I only wanted to do right before merge (i.e. hardware-in-the-loop testing)
<re_irc>
<@ryan-summers:matrix.org> Guess you can still manually run a workflow, but that doesn't feel the same
starblue1 has quit [Ping timeout: 248 seconds]
starblue1 has joined #rust-embedded
<re_irc>
<@dirbaio:matrix.org> I'll miss "bors d+" too. does anyone know if there's another way to get that?
<re_irc>
<@thejpster:matrix.org> Maybe we should write a bors-ds9 (which is a follow up to bors-ng) which lets you pretend you're talking to bors, but does github things instead.
<re_irc>
<@therealprof:matrix.org> Aww, random science fiction reference of the day... 🤣
<re_irc>
<@joelsa:fehler-in-der-matrix.de> : That sounds like a decent idea, though some of the features might be hard to do with GHA
vulfe has joined #rust-embedded
vulfe has quit [Remote host closed the connection]
vulfe has joined #rust-embedded
<re_irc>
<@dngrs:matrix.org> "We are the Bors. Lower your shields and surrender your PRs."
vulfe has quit [Ping timeout: 246 seconds]
vulfe has joined #rust-embedded
emerent has quit [Ping timeout: 260 seconds]
emerent has joined #rust-embedded
vulfe has quit [Remote host closed the connection]
<re_irc>
<@jamesmunns:beeper.com> It's possible we could (in no particular order):
<re_irc>
- Set up a hosted instance of bors
<re_irc>
- Ask the rust project if we can use the official homu (https://github.com/rust-lang/homu) (rust's actual version of bors, not actually bors-ng)
<re_irc>
- Host our own version of homu (not sure if this is practical)
<re_irc>
- Switch to GH's Merge Queues
<re_irc>
- Abandon bors-flow
<re_irc>
<@jamesmunns:beeper.com> (homu also supports "d"/"delegate" commands)
<re_irc>
<@jamesmunns:beeper.com> unsure if homu will/should work outside of the "rust-lang", at least using the "main project infra".
<re_irc>
<@jamesmunns:beeper.com> Might be interesting to ping Pietro, or deciding whether we should ping him at today's meeting
<re_irc>
<@jamesmunns:beeper.com> Also maybe good to figure out who is running the meeting tonight, or let me know if both of y'all are busy :)
<re_irc>
<@dirbaio:matrix.org> github merge queues work very well
<re_irc>
<@dirbaio:matrix.org> can't see a reason for using something custom anymore
vulfe has joined #rust-embedded
<re_irc>
<@adamgreig:matrix.org> I'm back today 👋
<re_irc>
<@christophe:yatrix.org> Does this Rust Embedded community cover eBPF (as in aya-rs)? 🙂
<re_irc>
<@ryan-summers:matrix.org> Generally, rust-embedded is usually just "no_std" applications in my view, so I've not seen anyone even mention aya-rs here before
<re_irc>
<@ryan-summers:matrix.org> I.e. stuff that doesn't have any OS running at all, not really within the kernel etc.
<re_irc>
<@therealprof:matrix.org> : It feels like we have to touch all repos regarding CI every other year. 😕 This time it'll also affect standard procedures.
IlPalazzo-ojiisa has joined #rust-embedded
<re_irc>
<@adamgreig:matrix.org> @room i'm back, and it's meeting time again! agenda is https://hackmd.io/KPDoZaixT3SCJ4R8hu2c0w, please add anything you'd like to announce/discuss and we'll kick off in 5min
<re_irc>
<@adamgreig:matrix.org> back and very jetlagged, hopefully I can stay awake long enough to have dinner after the meeting 💪
<cr1901>
adamgreig: I'm deep into client work rn. Yes, even right now. This isn't a hibernation, but I will be not be active for at least this week and prob next.
<re_irc>
<@adamgreig:matrix.org> cr1901: got it, hope it goes well!
<cr1901>
I cannot afford to break my concentration :'D
<cr1901>
Tyvm
<re_irc>
<@adamgreig:matrix.org> ok, let's begin! first a quite old announcement that the embedded-hal crates got a new alpha release about a month ago now 😅
<re_irc>
<@adamgreig:matrix.org> I haven't noticed anything else announcement-wise since then but I probably missed some things, does anyone have anything else to announce?
<re_irc>
<@therealprof:matrix.org> Wasn't there a new probe-rs release?
<re_irc>
<@adamgreig:matrix.org> hmm, I think the v0.18 release was just before the last meeting
<re_irc>
<@jannic:matrix.org> Not really an announcement, but bugs/changes in defmt and probe-run caused some confusion.
<re_irc>
<@adamgreig:matrix.org> ah yea, i guess probe-run had the new release
<re_irc>
<@adamgreig:matrix.org> is there an issue thread or anything about the confusion?
<re_irc>
<@jamesmunns:beeper.com> There's an issue, but it's closed, and some people still bump into version mismatch issues
<re_irc>
<@dirbaio:matrix.org> what's funny is that issue was reported BEFORE the release
<re_irc>
<@dirbaio:matrix.org> and it makes all of rp2040, nrf+softdevice, any chip with a bootloader not work
<re_irc>
<@jamesmunns:beeper.com> Oh that one is separate from the "wire format" issue I was thinking of.
<re_irc>
<@adamgreig:matrix.org> ah, that would be their fix to us changing the c-m-rt startup code huh
<re_irc>
<@jannic:matrix.org> And if one downgraded because of the reset-handler issue, it wouldn't work because of the wire format.
<re_irc>
<@jannic:matrix.org> Aaand than half of ferrous was on vacation :-)
<re_irc>
<@adamgreig:matrix.org> popular time for it apparently!
<re_irc>
<@adamgreig:matrix.org> ok, thanks for mentioning, I guess they know about it at least
<re_irc>
<@jannic:matrix.org> * then
<re_irc>
<@dirbaio:matrix.org> just wanted to shamelessly plug "probe-rs-cli run"
<re_irc>
<@dirbaio:matrix.org> it's not now, but the goal is to get it to feature-parity with "probe-run"
<re_irc>
<@adamgreig:matrix.org> renaming "probe-rs-cli" to "probe-rs" when :P
<re_irc>
<@dirbaio:matrix.org> going forward I think it's valuable to have that feature built into "probe-rs", so it's always up-to-date
<re_irc>
<@dirbaio:matrix.org> vs probe-run usually takes time to update to newer probe-rs's
<re_irc>
<@dirbaio:matrix.org> : Soon™
vulfe has quit [Remote host closed the connection]
vulfe has joined #rust-embedded
<re_irc>
<@adamgreig:matrix.org> : next up, as mentioned, bors-ng (which we use on almost all wg repos) is being retired at some point
<re_irc>
<@adamgreig:matrix.org> : some options here, though one more is "leave it for now, it's not clear when the hosted bors-ng is going away yet"
<re_irc>
<@adamgreig:matrix.org> I'd be tempted to suggest trying gh merge queues on one or two repos soon, and use that experience to inform a later decision
<re_irc>
<@emilgardis:matrix.org> I like the idea of switching to rust-lang homu
<re_irc>
<@emilgardis:matrix.org> (if even possible)
<re_irc>
<@adamgreig:matrix.org> does anyone know who we'd ask to find out if that is possible?
<re_irc>
<@therealprof:matrix.org> Is homu even used anywhere? I thought people moved from homu to bors?
<re_irc>
<@adamgreig:matrix.org> homu is the name of the rust-lang bot
<re_irc>
<@adamgreig:matrix.org> that looks like bors/is called bors
<re_irc>
<@dirbaio:matrix.org> do we need something that the GH merge queue doesn't provide?
<re_irc>
<@dirbaio:matrix.org> if not, then I'd suggest using the merge queue
<re_irc>
<@emilgardis:matrix.org> the one thing I like is bors try
<re_irc>
<@therealprof:matrix.org> Huh. Must be confusing with something else then... or the name was recycled.
<re_irc>
<@dirbaio:matrix.org> seems the lowest maintenance option
<re_irc>
<@adamgreig:matrix.org> homura does have the advantage of a scary mascot too
<re_irc>
<@adamgreig:matrix.org> we'd more than triple the number of repos it's on if we used it for all the wg things we use bors for, so they might be a bit hesitant
vulfe has quit [Ping timeout: 268 seconds]
<re_irc>
<@adamgreig:matrix.org> I think generally our repos don't have the same level of sophisticated requirements and long CIs as like rust-lang/rust, so my guess is the gh merge queue will be fine
<re_irc>
<@jamesmunns:beeper.com> Imo is worth pinging Pietro for a "vibe check" on the idea
<re_irc>
<@jamesmunns:beeper.com> We're already in all the permissions lists, and just made an update on that recently
<re_irc>
<@therealprof:matrix.org> Oh yes, re forgotten announcements... I did my best to get the WG team updated... 😅
<re_irc>
<@jamesmunns:beeper.com> (bors-ng and rust-bors/homu I think are just forks/reimpls of the same ideas, tuned for different cases/deployments)
<re_irc>
<@dirbaio:matrix.org> what features does bors have that the GH merge queue doesn't?
<re_irc>
are they worth the trouble of using another bors, or can we go without it?
<re_irc>
can we get them somehow else?
<re_irc>
<@dirbaio:matrix.org> * them?
<re_irc>
<@thejpster:matrix.org> I asked Pietro about this today and Ferrocene doesn’t work with Merge Queues for some reason. I can find out details.
<re_irc>
<@adamgreig:matrix.org> I think most of the features are around the needs of big repos with a lot of PRs and a long CI process, where rollups are really important, which just isn't the case for any/many of the wg repos
<re_irc>
<@dirbaio:matrix.org> like, some github app that does just the thing, on top of github merge queues..?
<re_irc>
<@therealprof:matrix.org> I'm a bit wary about changing all repos from bors to GH and rewriting the instructions. That might be a considerable chunk of work (similarly to switching from Travis CI to GHA).
<re_irc>
<@emilgardis:matrix.org> I think it's "bors try" and "bors d+" that are missing
<re_irc>
<@adamgreig:matrix.org> I think it would be _much_ easier to swap to gh merge queue than it was to swap travis to gha
<re_irc>
<@dirbaio:matrix.org> "bors d+" I'll miss in Embassy too, yes. I use it a lot
<re_irc>
<@emilgardis:matrix.org> (and we don't use "bors d+" really)
<re_irc>
<@adamgreig:matrix.org> it's basically just delete the bors.toml and enable merge queue in repo settings
<re_irc>
<@adamgreig:matrix.org> but still, that's why we'd want to test it on one or two repos first before deciding, if we went that route
<re_irc>
<@emilgardis:matrix.org> * really, in embedded-wg due to the review requirement)
<re_irc>
<@emilgardis:matrix.org> no, you need to set the event up in the workflow also
<re_irc>
<@therealprof:matrix.org> : Yes, that's what I read earlier today, too.
<re_irc>
<@adamgreig:matrix.org> my guess is github eventually adds a delegate option since they copied the rest of bors' design :P
<re_irc>
<@adamgreig:matrix.org> for the wg repos, the checks run by "try" already run on any new PR, unless it's by a new contributor in which case gh has the 'run ci' button
<re_irc>
<@therealprof:matrix.org> I haven't quite figured out how that feature works TBH.
<re_irc>
<@adamgreig:matrix.org> I don't think we have any set up to run different checks on "try"?
<re_irc>
<@adamgreig:matrix.org> even for the HIL CI tests
<re_irc>
<@dirbaio:matrix.org> i've already used it in a few repos (nrf-softdevice, cyw43). Can confirm migrating is easy, it works fine, and no issues :)
<re_irc>
<@thejpster:matrix.org> (I added an agenda item, sorry it’s late)
<re_irc>
<@adamgreig:matrix.org> is there an example PR for migrating to it?
<re_irc>
<@therealprof:matrix.org> : How does the workflow work though? It seems all it does is give you a button that says: add to queue rather than manual merging.
<re_irc>
<@dirbaio:matrix.org> okay I misremembered, I didn't migrate these repos from bors, they didn't have bors before
<re_irc>
<@adamgreig:matrix.org> from looking at rust-lang's tomu config it looks like in principle it could operate on out-of-org repos, but it needs a secret set up for each repo too, my gut feeling is we wouldn't want to enable this for ~33 repos we have bors on (https://github.com/rust-lang/homu/blob/master/cfg.production.toml)
<re_irc>
<@emilgardis:matrix.org> github recommends setting the task also, but pretty easy
<re_irc>
merge_group:
<re_irc>
types: [checks_requested]
<re_irc>
on:
<re_irc>
<@dirbaio:matrix.org> : you configure in github's UI which checks are mandatory to pass, then enable merge queue
<re_irc>
<@dirbaio:matrix.org> the "queue merge" button enqueues it, then creates a "ghost" branch with the PR merged into master, and then merges if the mandatory checks pass
<re_irc>
<@adamgreig:matrix.org> using the GH UI for merge queue is a nicer user experience than mysterious "bors r+" comments too, I think
<re_irc>
<@dirbaio:matrix.org> and there's some settings in the UI for controlling batching, batch size etc
<re_irc>
<@dirbaio:matrix.org> you can enforce tests to pass in the PR individually before being added to a batch too, to reduce the rate of batch failures
<re_irc>
<@dirbaio:matrix.org> seems well designed for high-traffic repos
<re_irc>
<@thejpster:matrix.org> Is it worth doing this in an issue and bringing blockers back for discussion?
<re_irc>
<@dirbaio:matrix.org> same as bors
<re_irc>
<@dirbaio:matrix.org> * bors, really
<re_irc>
<@therealprof:matrix.org> : I do like the snarky comments when you mistype, though. ;l)
<re_irc>
<@therealprof:matrix.org> * 😉
<re_irc>
<@dirbaio:matrix.org> I think "bors d+" can be substituted with "automerge" to some extent, if you can configure it to not disengage when the contributor pushes changes. not sure
<re_irc>
<@dirbaio:matrix.org> not sure about "bors try"
<re_irc>
<@adamgreig:matrix.org> hah true, github queue doesn't have "bros"
<re_irc>
<@emilgardis:matrix.org> you could also, if you're desparate for delegate, make a tag and allow queueing with a comment via workflow event issue_commented
vulfe has joined #rust-embedded
<re_irc>
<@adamgreig:matrix.org> OK yea, I'll open an issue on the wg repo for further discussion
<re_irc>
<@adamgreig:matrix.org> I'll leave the e-h uart thing til last so we can get through the other points first
<re_irc>
<@therealprof:matrix.org> (lol, element also call the fistbump emoji facepunch... thought only Slack did that)
<re_irc>
<@emilgardis:matrix.org> I added the discussion point, just saw it while looking through the products page after the bors deprecation message, we should probably preserve what little we have there
<re_irc>
<@adamgreig:matrix.org> thanks
<re_irc>
<@adamgreig:matrix.org> the nuisance is when you transfer, "Posts will be seen by everyone who has permissions for this repository which may be different than members in this team. "
<re_irc>
<@adamgreig:matrix.org> and then you have to pick a repo it seems? I thought you could also make it an org-wide discussion but nto sure
<re_irc>
<@jamesmunns:beeper.com> (time check: not sure how vital the leadership council bit is, 15ish mins left in the meeting)
vulfe has quit [Ping timeout: 246 seconds]
<re_irc>
<@adamgreig:matrix.org> it seems like github don't immediately plan to delete the team discussions text, just remove the abiltiy to add/edit it
<re_irc>
<@adamgreig:matrix.org> for now anyway, heh
<re_irc>
<@adamgreig:matrix.org> it's not clear if you can still transfer it later too, but hopefully?
<re_irc>
<@adamgreig:matrix.org> I don't think there's been a huge amount of chat in the team discussion bits so I'm sort of inclined to leave it for now rather than transfer it all to a particular repo which obliterates the current permissions?
<re_irc>
<@adamgreig:matrix.org> not that there's much personal or anything but there are some email addresses and such
<re_irc>
<@adamgreig:matrix.org> in any event it seems like we shouldn't need to rush for the may 8 deadline
<re_irc>
<@adamgreig:matrix.org> so yea, let's move on to the last point for now, about the leadership council stuff
<re_irc>
<@thejpster:matrix.org> Jan Erik said on Zulip:
<re_irc>
As a reminder we're looking for one Council Representative for the Launching Pad team, representing the list of teams and working groups listed below.
<re_irc>
As I wrote last week I want to start the representative selection for the Launching Pad Team.
<re_irc>
<@thejpster:matrix.org> I have been nominated (thank you), as has Nell, Jan Erik (who declined), Ed Page, Forrest Anderson and Tyler Mandry.
<re_irc>
<@thejpster:matrix.org> Also some people noted they didn’t get the email. We should remind ourselves how the WG emails work and get that synced up.
<re_irc>
<@jamesmunns:beeper.com> : I think therealprof's work should have sorted that
<re_irc>
<@jamesmunns:beeper.com> It's based on the membership lists in the Rust project metadata, which was a bit out of date.
<re_irc>
<@therealprof:matrix.org> : That has been taken care of (unless I forgot to update someone who wanted to register their email too).
<re_irc>
<@thejpster:matrix.org> Can you remind me how it works?
<re_irc>
<@emilgardis:matrix.org> Was this supposed to go out as a mail? I didn't get one
<re_irc>
<@emilgardis:matrix.org> my mail is correct
<re_irc>
<@therealprof:matrix.org> Need to checkout the rust team repo. Then add people there and PR the change.
<re_irc>
<@adamgreig:matrix.org> there's a toml file for the whole embedded wg, another for each wg team, and then for each person in https://github.com/rust-lang/team/tree/master/people which can have an email address
vulfe has joined #rust-embedded
<re_irc>
<@emilgardis:matrix.org> my understaing is that all@ goes to wg-leads only
<re_irc>
<@thejpster:matrix.org> Argh. Is that repo public?
<re_irc>
<@adamgreig:matrix.org> hmm, the email was to embedded-wg@rust-lang.org, though that's not the same spelling as wg-embedded.toml, heh
<re_irc>
<@adamgreig:matrix.org> yea, the repo is public
<re_irc>
<@thejpster:matrix.org> /people/thejpster.toml has a private email address in it. I don’t give that one out publicly as I can’t spam block it.
<re_irc>
<@emilgardis:matrix.org> there's a way to encrypt the mail, instructions in the readme
<re_irc>
<@therealprof:matrix.org> : Guess there's little point in changing it now.
<re_irc>
<@thejpster:matrix.org> Well the important thing is people are aware the process is happening, and we’ll update you with the name of the selected Leadership Council Representative for the District of “Launching Pad” at a future meeting.
<re_irc>
<@adamgreig:matrix.org> who _did_ get the email?
<re_irc>
<@thejpster:matrix.org> I did.
<re_irc>
<@adamgreig:matrix.org> so presumably that was the source, but then I don't know why didn't get the email
<re_irc>
<@adamgreig:matrix.org> to the email in that toml file I guess?
<re_irc>
<@thejpster:matrix.org> Hard to say on my phone. It says to: embedded-wg@rust-lang.org
<re_irc>
<@adamgreig:matrix.org> anyway yea, the text and zulip above sum up what's going on, if anyone wants to weigh in or nominate anyone else the zulip thread is the place
<re_irc>
<@adamgreig:matrix.org> we don't yet know how the nominees will be selected from, I don't think
<re_irc>
<@emilgardis:matrix.org> yup, I have the info now
<re_irc>
<@emilgardis:matrix.org> I'll contact someone on the zulip about it I guess after some digging
<re_irc>
<@adamgreig:matrix.org> and I suspect once that happens there will then be further discussions about what the launching-pad team does and how the embedded-wg emerges from it, but so far that's all in the future
<re_irc>
<@therealprof:matrix.org> : That is mentioned in the team repo as the mailing list address.
<re_irc>
<@emilgardis:matrix.org> we'd be under launching-pad yes?
<re_irc>
<@adamgreig:matrix.org> yea
<re_irc>
<@adamgreig:matrix.org> along with the other wgs
<re_irc>
<@therealprof:matrix.org> Can you confirm your email in the repo is correct?
<re_irc>
<@emilgardis:matrix.org> I can
<re_irc>
<@jamesmunns:beeper.com> : T-infra is probably the place to start :D
<re_irc>
<@emilgardis:matrix.org> I'm all for not getting mails though :D
<re_irc>
<@therealprof:matrix.org> : Don't worry. Traffic is like 1 mail/a on that list...
<re_irc>
<@adamgreig:matrix.org> confusingly we also have the rust-embedded.org mailing lists which do receive a bit more spam :P
<re_irc>
<@jamesmunns:beeper.com> : Might make sense to ask @nastevens to deprecate those, if they aren't tied to auth for anyone?
<re_irc>
<@thejpster:matrix.org> Ugh, yeah. I get a lot to cortex-m@ my domain.
<re_irc>
<@adamgreig:matrix.org> : probably, though perhaps let's wait for the launching-pad stuff to shake out first and see what mailing lists remain
<re_irc>
<@thejpster:matrix.org> I think it’s been better recently but then I ran for six months with both an MX record and a CNAME, so my mail was a mess.
<re_irc>
<@adamgreig:matrix.org> good way to get less spam I guess :P
<re_irc>
<@adamgreig:matrix.org> right, that's our time for today, thanks everyone! regular meetings will resume so seeya same time next week
<re_irc>
I have some availability for consulting starting in late May, if anyone here needs any help with embedded rust (or general systems topics).
<re_irc>
Working on my website (https://onevariable.com/) to be a better landing page, but feel free to DM or email me if y'all (or someone you know) could use a hand :)
<re_irc>
<@bitslinger:matrix.org> what was the e-h uart issue?
<re_irc>
<@emilgardis:matrix.org> I've contacted infra now and they did some adjustment, next time there's a mail to embedded-wg, if someone could remember to ping me to see if I received it as well, would be appreciated!
vulfe has quit [Remote host closed the connection]
vulfe has joined #rust-embedded
vulfe has quit [Ping timeout: 268 seconds]
vulfe has joined #rust-embedded
m5zs7k has quit [Ping timeout: 250 seconds]
m5zs7k has joined #rust-embedded
<vulfe>
I just spent a while examining the concurrency examples from embedded_hal only to realize that they aren't no_std legal (which seems weird given the context)
<re_irc>
<@walstib-alex:matrix.org> Any specific examples?
<vulfe>
unless I'm misunderstanding something, I don't think Async and future::Loop are even exposed by futures when default-features is set to false
<vulfe>
(which is what futures tells you to set in no_std environments)