needbrrrrrrr90: it can be used with a remote node too. doesn't have to be local.

it can even be used with remote public nodes (just like many do for their wallets)

hence one doesn't need to run their own local node to use self-select.

ah fantastic, xmrpow was implying that it required a full node

shilling your work jtgrassie :)

What are the major drawbacks, in your opinion?

I feel like I should be actively attempting to get this popular, unless there's some big drawback I'm missing

well the only real difference for a pool is they must validate the templates miners use, this is very fast though so not really a drawback.

for a miner, they need access to a node (local or remote).

which again, is really a non-issue.

the bigger reason uptake has not been that high I put down to the majority of the pools use that crappy nodejs pool, and nobody has bothered to implement it for that pool.

if one of the nodejs pool devs bothered to implement, we'd probably see much higher usage.

but then they can't add fast-mode randomx for share validation, I'm not holding my breath.

Is there any reason for me to upgrade the gui monero client to 15.04? I use opensuse leap 15.1 and there is a glibc version problem with the latest build of monero, opensuse leap 15.1 has version 2.26 and monero 15.04 wants version 2.27.

What glibc version problem exactly ?

Is this a gitian build ?

IIRC gitian (or depends) wants a libc compatible with some particular verison, but later ones should be accepted too.

glibc.so.6 wants version 2.27, has version 2.26

You've built the binary on the same box, right ?

checked and opensuse does not offer an update unless I change to tumbleweed.

And did you update glibc after building ?

no update for glibc is offered for opensuse leap 15.1

I used the straight downloaded linux 64 bit version of monero. monero 15.02 runs fine

Ah, there's your problem then. If you're going to develop for monero, build it yourself :)

I suck at Cmake, LOL. Ok, will build from source if I want the latest

I *think* the GUI binaries will need whatever was on the machine where it was built since it does not use depends (which does some things to ensure backward compat with some versions of glibc).

I also have the issue with AMD ryzen 1600+ not liking randomX

saving for a newer processor

GUI binaries were compiled on Ubuntu 18.04

ok, everyone loves Ubuntu. I am old Novel admin person so am partial to opensuse

thanks for the suggestions.

bye

vtnerd: how can I compile the supercop benchmark?

which cmake var?

`MONERO_WALLET_CRYPTO_BENCH=cn` ?

if you don't specify anything, then do `make monero-wallet-crypto-bench`

probably need a readme for this

if you specify no special parameters on the command-line, it chooses amd64-64-24k for the wallet backend, and does `cn amd64-51-30k` for the benchmark

or sorry `cn amd64-51-30k amd64-64-24k` for the benchmark. but only on amd64 platforms

if you want to do it manually then a cmake option `-DMONERO_WALLET_CRYPTO_BENCH="cn amd64-64-24k amd64-51-30k` is what to specify for amd64 (there are no other backends currently)

the crypto bench is NOT in the default cmake target group, so doing `make` will never compile it by default

the benchmark by default "depends on" every crypto backend available for the platform, so I removed it from the default make target so that only the backend in use (if any) gets built

bloody hell, i still got this on the release tagged thingy

2020-05-11 03:12:22.194 E Failed to load hashes - unexpected data size

huzzah the 6514/15 PR worked

Any idea why two or more blanks lead to an empty element in the list returned by this method: wipeable_string::split(std::vector<wipeable_string> &fields)

No idea what this could be good for. I just now what is not good: Include 2 blanks anywhere in a seed to restore and it already fails. Bad UX, if you ask me

At least in core it seems that spliting method is only used for seeds; did not check the GUI

rbrunner7: it should not, what version are you using?

HEAD

oh wait, I read the code wrong

I think the intention is implemented correctly (2 or more blanks -> empty element in list), I just don't have any idea in what situations I would want that method to do that :)

And as I said, the result for seed is not very good

I think dEBRUYNE even has that in their standard arsenal of helpful questions if somebody reports that a restore by seed did not work: "Any double blanks?"

If my quick glance is correct the GUI seems to use JavaScript split(" ") which also would have the problem to fail with multiple blanks. For being robust a regular expression would probably be needed, like split(" +")

I call it redundant spaces :-P

^ rbrunner7

Does this fail ?

ASSERT_TRUE(check_split("foo bar baz ", {"foo", "bar", "baz"}));

tests/unit_tests/wipeable_string.cpp

Well, did not try, but I would say so, after my attempt to understand said method

Then paste an example input giving bad results and I'll check.

empty element gets added here: github.com/monero-project/monero/bl…b/epee/src/wipeable_string.cpp#L196

Yeah, but I just tried and could *not* get it to fail (again). Maybe I had some invisible characters in the input string, from editing. But then I start to doubt my capacity to understand C++ methods that are a mere 10 lines long :(

Alright, will debug. That's too much.

GUI might be completely different story, will check that afterwards.

Adding a seed with spaces works here.

rbrunner7: GUI uses the same split function

Alright, then my cursory analysis was wrong that it uses JavaScript for splitting seeds

can't reproduce here, just restored with many spaces in the string

I think the problem is when entering arrow keys and then deleting them again.

Did you have a test case selsta ?

ah yeah I've had issues with backspaces on password entry, but a loooong time ago

might be an interaction with readline

With readline or without ? If without, then it's expected.

I don’t know how to do arrow keys in testcases.

moneromooo: yeah was old builds pre-readline, iirc

WIth readline. But readline does not seem to work during seed entry.

It does work afterwards.

oh hmmmm

Ah, good point. It's raw entry to avoid leaving traces in memory.

Sorry then for making a fuss, blanks do work, I just worry about my mind not understanding a few lines of C++, but debugging and watching will surely help finally

rbrunner7: I got confused too :D

And you know why blanks do not disturb? Did not see any logic higher up that filters out those empty array entries again ...

Before I make again a complete laughing stock out of myself: Can somebody please check whether the most recent GUI wallet accepts seeds with extra blanks in them?

With this seed: abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey

And then please try this: abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey abbey

(single difference should be a blank)

And of course that split method does not add empty entries. I was simply reading the code wrong, probably under the unconscious assumption that there *must* be something wrong with it. Sigh.

rbrunner7: Can you explain?

Which? The latest sentence?

`fields.push_back({});` looks like an empty element to me.

But maybe I don’t understand it :)

Yes. It's the place where elements are created in the first place. They have to be created somehow, right? And after creating it letters are added to it. The "if (!blanks)" around that whole statement group filters out multiple blans

Had to actually debug and single-step through it until I could convince myself that there is no bug :)

Talk about the power of prejudice

But I still wait anxiously somebody checks multiple blanks on the GUI seed entry field ...

Ok I get it now too. The second push_back is not the vector it’s for the wipeable_string.

Yes :)

Both functions being called push_back confused me but it makes sense.

rbrunner7: see here: github.com/monero-project/monero-gu…ad941bd8a4a3986fb/js/Wizard.js#L111

double space means you can’t enter the last word

?

I mean if you enter two spaces you can’t click on next

Yes.

What do you think about this behaviour?

We tried the automatic stripping of double spaces but it did not work well

I see. That 'split' function on the line you give, is that JavaScript?

split(" ")

yes but that is just the frontend, it uses wipeable_string afterwards

Ah ok. But if you don't get beyond this point that's for naught, right?

Please let me try split(" +") instead of split(" "). That could work. That might be my redemption for the fuss from me today here

rbrunner7: split(" +")? AFAIK split requires one separator and not regex.

I will check ... after my old creaky notebook finally finishes to compile the embedded Monero core ...

It's split(/\s+/)

Works like a charm

Will make a PR

Wow, cool new animation when checking for daemon still running :)

I checked that regular expression on this interactive playground first: developer.mozilla.org/en-US/docs/We…ference/Global_Objects/String/split

Sooo... I've been looking every now and then just in case and... google/oss-fuzz #2451

It's an account with the ability to reopen a bug, so likely talking from a position of being able to back it up.

Looks like fuzz tests are about to receive more attention ^_^

Nice one :D

curious - how does fuzz testing work with vulnerability reporting?

is there a human between that fuzz thing and reporting a bug, or does some machine just crawl github and fuzz everything and report bugs

Coder runs fuzz tests. Fuzz tests find a crash or other. Coder tries to find an exploit for it. If found, coder reports to H1 os similar.

You mean how google actually runs them themselves ? I think they have a doc explaining this somewhere, I saw that long ago. I think it's partly automated.

ah ok. so there's a human.

"As of January 2020, OSS-Fuzz has found over 16,000 bugs in 250 open source projects." From their project page.

At least an AI that can recognize bloody cars and traffic lights.

i think that the guy that created AFL is working on an automated exploit thingy too :D

Anyone's worked with oss-fuzz before ?

selsta do you want 6515 merged now?

Docker downloads like 1 GB every build ? My god it can't be that stupid...

most everything gets cached.. but if you blow it away, yup (though i never looked at how much it actually was)

Well, if it errors out after downloading, it definitely slurps it again.

I assume the commenter might be the same person as unseddd?

Yes, they said they asked at the time.

I thought it was pointless and a bit spammy, but looks like I was wrong ^_^

luigi1111w: I think he wanted to have it merged, yes

Jtgrassie, are you there?

luigi1111w: yes please

Nice. It stopped downloading a gig after 5 times. It just has to fail a bit later.