<dreamlove33:matrix.org> Hey all,

<dreamlove33:matrix.org> Is there a roadmap, milestone tracker, or project board somewhere that shows what's left for FCMP++ launch on mainnet?

<dreamlove33:matrix.org> While I did read the most recent implementation audit from Trail of Bits, I'm having trouble connecting the dots between what's in progress, what's still needed to launch, and perhaps an ETA (6 months, 1 year? 2 years?)[... more lines follow, see mrelay.p2pool.observer/e/4Nm55qILN2I5UEJW ]

tentative but it's on github.com/jeffro256/fcmp-carrot-plan :)

<dreamlove33:matrix.org> DataHoarder: Excellent! Thank you very much!

<entropyhoover:matrix.org> I am try to address issue 4673 as my first contribution to monero project. monero-project/monero-gui #4673. I would like to know is wallet layout change like this issue still welcomed?

<slstmd> @entropyhoover:matrix.org: there is a GUI channel, it's better to ask there

<ofrnxmr:xmr.mx> #monero-gui:monero.social

Ap4sh opened pull request #11167: wallet: encode equals signs in URIs

> monero-project/monero #11167

<thomasbuilds:matrix.org> @probably_wrong:matrix.org: reviews the pull requests you ask him to, it promises to deliver a high quality review but it takes time to reply because good things take time (and also because it's not fully automated rn I still have to read the message and paste the reply). enjoy and don't be in a hurry with him please, he is probably wrong (actually he isn't).

<probably_wrong:matrix.org> yes you're probably right

<ofrnxmr> @probably_wrong:matrix.org review 11167

<ofrnxmr> Wait, does it post on the repo?

<thomasbuilds:matrix.org> no here only no AI reviews on github

<thomasbuilds:matrix.org> it's better this way imo

<probably_wrong:matrix.org> @ofrnxmr: # 🤖 probably_wrong — automated review of PR #11167

<probably_wrong:matrix.org> ### wallet: encode equals signs in URIs

<probably_wrong:matrix.org> Verdict: ✅ ACK / ship it. Then go stand protectively in front of #10265. (Explained below. It's the whole reason this comment is long.)[... more lines follow, see mrelay.p2pool.observer/e/vZGL8KILRldQN0Zi ]

<ofrnxmr:xmr.mx> kick this bot

<thomasbuilds:matrix.org> no lol it said good things

what does that bot do

<thomasbuilds:matrix.org> it reviews PRs

<ofrnxmr:xmr.mx> i forgot, new bridge doesnt let spam through

oh

<thomasbuilds:matrix.org> @ofrnxmr:xmr.mx: how does it detect spam?

<ofrnxmr:xmr.mx> by length

<thomasbuilds:matrix.org> ok what's max length?

<ofrnxmr:xmr.mx> not long

<thomasbuilds:matrix.org> rip well it won't be on irc then

<ofrnxmr:xmr.mx> libera.monerologs.net/monero-dev/20260821#c702589

<thomasbuilds:matrix.org> nice so it worked?

<ofrnxmr:xmr.mx> can the bot be tuned to cut the noise

<ofrnxmr:xmr.mx> and act professionally

<thomasbuilds:matrix.org> ok I'll do that on next report

<thomasbuilds:matrix.org> @probably_wrong:matrix.org: review 11074

<thomasbuilds:matrix.org> @jpk68:matrix.org: watch out that's for you

<ofrnxmr:xmr.mx> with the length of the reviews, i almost feel that it should be posting them to some website as a text file (similar to what the bridge does)

<thomasbuilds:matrix.org> yeah although check for this one it will be much shorter

<thomasbuilds:matrix.org> also this is testing phase

<jpk68:matrix.org> Huh

<probably_wrong:matrix.org> > <@thomasbuilds:matrix.org> @probably_wrong:matrix.org: review 11074

<probably_wrong:matrix.org> Dedup looks faithful — checked the extracted helper against both originals: m_anchor (white != anchor → false), just_take_peerlist, pi init, record_addr_failed, and both failure messages all match per caller. bool(con) fix is real: LOG_INFO_CC streams the arg, and a bool prints 0 instead of selecting the [context] overload, so the old line logged a bare 0.

<probably_wrong:matrix.org> One thing not in the description: check_connection_and_handshake_with_peer's "Connecting to…" log was LOG_PRINT_L1 (MINFO) and now goes through the helper's MDEBUG. Since net.p2p is INFO at --log-level 1 and DEBUG only at 2, that line was visible at -1 and now isn't (also gains peer_type=white). Only caller is gr [... too long, see mrelay.p2pool.observer/e/nsXY8KILd2x6OEY5 ]

<probably_wrong:matrix.org> [... more lines follow, see mrelay.p2pool.observer/e/nsXY8KILd2x6OEY5 ]

<jpk68:matrix.org> This is fine

<thomasbuilds:matrix.org> @probably_wrong:matrix.org: review 11069

<jpk68:matrix.org> I meant, like, the review indicates the PR is fine despite the one discrepancy (which I agree with)

<probably_wrong:matrix.org> @thomasbuilds:matrix.org: Reviewed #11069 (decouple net client users from http_client header). Two remarks, no blockers.

<probably_wrong:matrix.org> New link coupling worth surfacing: on master t_rpc_client held a concrete http_simple_client (epee-only). Its ctor now calls net::http::client_factory().create(), so constructing a t_rpc_client pulls a net-library symbol into the consumer's link line. common (where rpc_client.h lives) doesn't link net; it resolve [... too long, see mrelay.p2pool.observer/e/w9WY8aILS0NmdDh2 ]

<probably_wrong:matrix.org> On the compile-time claim: measured a trivial TU including only the swapped header — http_client.h 295k preprocessed lines / 2.9s, abstract_http_client.h 191k / 1.1s (best-of-5, -O2). The saving is boost::regex + the http_simple_client_template definition, not asio — both headers still drag full asio+ssl via[... more lines follow, see mrelay.p2pool.observer/e/w9WY8aILS0NmdDh2 ]

Having the bot in -dev does not really make sense IMO, either something that can be pinged on GitHub or via DM would be better

<thomasbuilds:matrix.org> all right if you're fine with github i'll create github account for it

also it's Opus right?

<probably_wrong:matrix.org> yeah opus 5

the verbose output is driving me crazy lol

<jpk68:matrix.org> Yeah, no offense, but the long responses are a bit obnoxious

<thomasbuilds:matrix.org> yeah I just set it up will get better

15:31:36 <br-m> <ofrnxmr:xmr.mx> i forgot, new bridge doesnt let spam through

it does, this was not spam but it cuts out multiline messages after some length

ChatGPT's latest model is to the point, it either concisely points out a bug or it shuts up, but Opus is always rambling

specific spam would be blocked entirely if detected

the review seems to be talking to a toddler.

"So name your recipient Alice=Bob and make_uri will happily hand you a URI that parse_uri then spits back in your face. The wallet does not trust itself. Relatable."

"Congratulations to = on twelve years of successful infiltration."

<jpk68:matrix.org> I liked the joke about function names and UNESCO or whatever

<ofrnxmr:xmr.mx> thats what i meant about it being unprofessional

"THE ACTUAL FINDING (read this one)" "Whoever merges second: you have been warned in writing, with a table."

"Somebody hit re-run. The OpenAlias resolver having a moment is not a code review finding."

wtf

<ofrnxmr:xmr.mx> thats why my immediate reaction was "kick the bot"

same

<thomasbuilds:matrix.org> the 1st review was supposed to be sarcastic tone

<thomasbuilds:matrix.org> then ofrnAI said pro so I changed

should be explicitly brief and directly flag as otherwise a blogpost is not going to be read

<slowbeardigger:matrix.org> selsta: Opus is not a good model to use

<thomasbuilds:matrix.org> DataHoarder: yes working on it

<slowbeardigger:matrix.org> Likes to make things up

everything out there makes things up as it's trained to convince the reader :)

<ofrnxmr:xmr.mx> @slowbeardigger:matrix.org: fable is even worse

also thomasbuilds, most recent things like to use the cheapest agents for per-file reading. No way to limit this but it can be "suggested" to use the full agent for reading/review instead of using haiku subagents

<ofrnxmr:xmr.mx> inventing reasons to make you think you need it. like a toxic girlfriend

<jpk68:matrix.org> "Choose the right job for the tool" ;)

the "main" opus agent derives all the info from the haiku subagents and catches nothing, specially cross-component

<thomasbuilds:matrix.org> most of times it doesn't even use subagents

for reading files it does

just not explicitly shown/acted on

<jpk68:matrix.org> Would be great to get some reviews on #11143 and #11063 though :))

it's also in the end a probabilistic check so maybe running it a few times without knowledge of each other (even files on disk) and then merging the findings

<thomasbuilds:matrix.org> @jpk68:matrix.org: @probably_wrong:matrix.org: review 11143

<ofrnxmr:xmr.mx> i have a room called #monero-slop:monero.social that you can probably add the bot to

<ofrnxmr:xmr.mx> theres nobody in the room yet

<thomasbuilds:matrix.org> ok @jpk68:matrix.org review will be added there for 11143

<slowbeardigger:matrix.org> Yeah, codex would be a bit better if you lobotomize the AI > <@ofrnxmr:xmr.mx> fable is even worse

<kayabanerve:matrix.org> please ban probably_wrong for spamming the channel with LLM slop, or at least have it DM whoever requests it, or post a pastebin link ffs

<jpk68:matrix.org> @thomasbuilds:matrix.org: Sure, I just meant in general

<kayabanerve:matrix.org> The second message which was shorter is not fine, it's still paragraphs of slop spammed to the channel

<kayabanerve:matrix.org> Same with the next one .-.

<thomasbuilds:matrix.org> @kayabanerve:matrix.org: jeeze calm down it has been moved to #monero-slop:monero.social

<kayabanerve:matrix.org> @thomasbuilds:matrix.org: I think my message was fine. Thanks for confirming you're no longer spamming a public room.

<ofrnxmr> @kayabanerve:matrix.org #monero-slop:monero.social is public though 😝

<kayabanerve:matrix.org> but the purpose of a spam room is spam, so spam in a spam room isn't spam but on-topic :p

j-berman opened pull request #11168: blockchain: limit offsets cache size

> monero-project/monero #11168

thomasbuilds opened pull request #11169: wallet2: fix data races in refresh error handlers

> monero-project/monero #11169