is the monero codebase mirrored/self-hosted anywhere?

i see there was a move to gitlab back when MS bought github but i just the the CCS on there

s/the/see/

bbqcore[m]: we have backhub.co setup

Hey guys, somebody knows where in the codebase can I find the function that encodes the seed to get the mnemonic?

The equivalent of this

s/Hey/Hi/

jrandom[m]: Check this for seed to key and vica versa: src/mnemonics/electrum-words.h

Do you know already my page about seeds? Maybe it will be useful to cross-check and help debugging: monerotech.info/Home/Seed

<selsta> "bbqcore: we have backhub.co..." <- wouldnt a self-hosted solution be better in case shtf?

endogenic: i sent you DM

self hosted solution for what?

the project code base

it was decided against it

yeah i read that, but arent the concerns still valid?

it's a bit of an endless discussion, we have a backup and we can move somewhere else in case the repos get taken down

i think a public self-hosted mirror would remove the trust aspect of uploading a backup in case github gets taken down without notice (hypothetical worse case)

users should be able to verify the codebase on their own in such a situation

git is decentralized, a lot of people have a copy of the repo

commits are signed

yes but lets say it goes down and i dont have a copy before

how can i verify its the same code?

if you don't have a copy of the code and if you don't trust anyone else then you can't

but that's kinda a pointless what if, the code could also change now and if everyone is in on it you wouldn't notice

right - a public self-hosted mirror would allow me to verify on my own

selsta: yeah i know, this is all hypothetical and assuming some onewith commit access has been compromised

What makes you able to verify authenticity of the source depending on how you obtain it ?

moneromooo: if there was a public mirror that would survive a github takedown, id be able to download the latest code and verify it against whatever backup comes on line after

Why is not the same as "if there was any other person having the source, id be able to download the latest code and verify it against whatever backup comes on line after" ?

what do you mean? id have to trust whatever code comes back online after such an attack

im sure loads of people will be able to call out any shady commits, but id have to trust them

Are you implying you would not have to trust it if it where that self-hosted mirror ?

you can't trust no one and then trust a random mirror :D

moneromooo: id at least know that whatever was re-upped is the same as what was taken down

apart from fully auditing the code myself theres already an obvious element of trust

How would you do that, and at the same time not be able to do that if you got it elsewhere ?

because id be trusting some one elses copy of the code

So you're basing this on "I trust what comes from this source", correct ?

no, im basing this on not being able to verify for myself if there was any fuckery between take-down and re-up

(I mean it's a fair default, but you started off with not having to trust)

how do you know that no one tempered with the mirror?

i guess something like uploading the code to a torrent would solve this

selsta: id be able to check, before every back up id download both sources and see if they match

of course this is assuming the monero project isnt already compromised before anything like this happens

thats not the approach im taking

I think you need to work the steps out in detail.

just because they matched in the past doesn't mean they don't diverge at some point

And I mean *detail*.

selsta: of course, so the whole point is to see where that happens for myself

the easiest is to make your own mirror

(btw, I pushed a PR to decrease the output export data size, which you'd expressed interest in, still chunky though)

selsta: yeah thats true, just thought a public one would be good for the project

It'd be good for availability.

moneromooo: link? thanks for taking that up. what kind of reduction we looking at?

cheers. can the same be done for key images?

that's a public mirror

Took about 2/3 off for the cold_signing functional test, but thse are pretty much only coinbase outs.

I think it'll save a little more in most cases.

Key images are mostly monero-project/monero #8179incompressible really.

And that data is way smaller already.

Now I'm wondering whether this should be a wallet flag... export all data, or just what's needed for cold signing.

thanks moneromooo, ill pass this on to airgap and get this QR standard rolling

Thanks moneromooo

<moneromooo> "Now I'm wondering whether this..." <- you mean like seed, view keys, addresses etc?

I meant whether the "send subset of data for size optimization" should be an option. To avoid the cold wallet having "dummy" data for outputs.

ie, run incoming_transfers and the txids will show up as 0000 I expect.

oh ok i thought you meant if there was a way to reduce all data in the wallet for export

cause there was talk of restoring wallets via QR also

hyc: if I want to lookup N arbitrary records in a table, is there a faster way to do it ? ie, sort them first ? MDB_MULTIPLE seems to be only for writing.

Ideally I'd want to minimize the number of uncached disk reads.

So I'm not sure sorting matters enough to bother with, root and level 1 are going to be likely cached, the rest probably uncached.

(the table is large)

Any other tip ?

It's DUPFIXED|DUPSORT btw.

Naively, it seems lmdb could implement MDB_MULTIPLE for lookups, locate the last non-leaf page for all requests, and use iovec for the last step, which will be full of cache misses. And hope the I/O scheduler can work with that intelligently.

MDB_GET_MULTIPLE / MDB_NEXT_MULTIPLE

I remember seeing this here: github.com/monero-project/monero/bl…lockchain_db/lmdb/db_lmdb.cpp#L2565

Thanks. That's not really what I want here.

It could be used for a manual scan for small tables (or large amounts of lookups) but that's not my case.

Ah I see

I guess we won't really prevent seeks, unless really RAM starved. They'll just happen in another order :/

Except if we query neighbours but that's not very likely in this case.

Can someone please explain to me what the point of contrib/epee/include/copyable_atomic.h is? The class extends std::atomic<uint32_t> but doesn't appear to add any functionality. Thanks!

Especially since std::atomic<uint32_t> is used in other places throughout the project