i wish to use the stagenet but i cant find a remote node or a working faucet

azy: community.xmr.to/faucet/stagenet community.xmr.to/nodes.html

worked yesterday

oh... a m inute ago that faucet threw an error

thank you

azy: Start a XMRig on stagenet and you will have coins in no time. You can't use a remote node for that. Run your own.

on master with a fresh install, I'm seeing wallet2 return balance and account information for account 0 only (e.g. wallet2.balance_all()). the wallet has balances in other accounts. I also see a test failure unit_tests. Wondering if anyone is encountering the same or knows of a possible cause?

Is balance_all supposed to look in all accounts ? wallet2 has a "current account" concept, though a couple RPC annoyingly ignore it.

Looks like it, looking at the source.

the unit_tests fail on block_reward_and_current_block_weight.fails_on_huge_median_size: paste.debian.net/1142781

CI passes all tests

same locally here

Is that related to javascript somehow ?

Compiled with... forget the name.

no monero-wallet-rpc shows the same behavior

emscripten is used to compile core c++ to webassembly, but that should be separate

OK. Any modifications ?

no modifications in the compiled core code, fresh download and install from master

378cdeaeaeb1dadc2dd7a324927f1af226611254 ?

yes

I'll give it a go later on/

the problem doesn't appear when compiling a branch of mine from working on the webassembly, so it seems related to updating to master

What commit hash did you fork that one from ?

ecb8cc9df40e466573df1c1956e374f17c960b8e Merge pull request #6397

woodser: that’s from 20 days ago

can you rebase with master?

The first commit is today's.

Unless I messed up a pull.

moneromooo: your commit hash is correct.

woodser posted one from 20 days ago

ah I misunderstood what he wrote :)

woodser: Unit tests on 378cdeaeaeb1dadc2dd7a324927f1af226611254 worked 2/2 here. Are you running this on a 32 bit arch ?

no 64 bit

Are you sure you have no modifications ?

I don't see how I could after trying git reset --hard upstream/master and downloading to fresh directory

Are you sure you've running the newly built binaries ?

yeah the tests were run against the new build directory, and I observed the problem running monero-wallet-rpc within the built bin directory

Then you get to bisect.

I'll continue to do difference tests between the branch that runs

yeah

the issue is caused by the if() conditionals added before expand_subaddresses() in commit bce050ee7827b630f5fce6922854c0d7d3a3896d

Oh. You were on about your balance issue ? I was on about the block weights one. That's the wallet. It has nothing to do with block weights.

Did you create a one off subaddress on that particular wallet ?

the wallet uses subaddresses but I don't call anything specifically to make it "one off"

Then it's not.

I think something like paste.debian.net/hidden/6b5f0cf9. Building.

It built for me. But since tests passed before, I can't tell whether it fixes your bug.

The consensus one is the one I'd really like to fix first though :)

no problem I'll test it shortly

seems like my functional tests also fail now

Which part ?

running again with `--output-on-failure`

I'm having problems applying the patch

git apply patch.txt says the patch does not apply

I probably did that on my work branch, let me make if off master now...

Pff. That's the rejected flat strucutre patch again.

5370=

?

running with --output-on-failure and now it passed again

will try one more time, maybe some timing stuff again :/

5370= looks like some valgrind message prefix.

Are you running with valgrind ?

(partial prefix)

no, I meant if the flat structure patch is PR #5370 :D

Ah, yes it is this one.

because it still seems open, maybe we can get a third opinion

(after the release)

woodser_: if you didn't see it when you timed out: paste.debian.net/hidden/083e8332

got it, thanks

accessing out of bounds memory with this :/

okay, tried 5 times, can’t reproduce the functional test fail

I get transient errors here paste.debian.net/hidden/63455963

(On Windows)

transfer sometimes works, haven't been able to reproduce the txpool one yet.

Just reproduced the txpool one.

Transient and txpool sounds like dandelion.

vtnerd: ^

it look like a bug in the tests. if the node "rolls" fluff, then the behavior can change

crap, I don't think the dandelion++ restricted mode can easily be tested through the functional tests

I'll revert most of the recent changes then, because there isn't a way around it given all of the constraints

revert the test changes, the other code was necessary for the existing tests to work

Anyone wants to maintain the "snap" build system ? If not, it gets axed. Comment in 6475 if you want to maintain it.