-
jeffro256[m]
In that PR, jberman "added logic to re-attempt the [select_outputs.gamma] test with a 10x larger sample size if the test fails on first try."
-
jeffro256[m]
Isn't that data dredging?
-
UkoeHB
Rucknium[m]:
-
jeffro256[m]
Assuming the events are independent, you square the probability of the test failing b/c it has to fail twice in a row
-
jberman[m]
the point of that test in particular is to calculate a median. 10x'ing the sample size gets a more accurate median
-
jberman[m]
it's not 10x higher by default because it takes longer to run IIRC
-
one-horse-wagon[
Seraphis meeting today at 18:00 UTC in Matrix room--No Wallet Left Behind. Everyone is cordially invited.
-
Rucknium[m]
This issue came up before. I'll look into it.
-
Guest4
Where can I get some reading/discussions on the growing size of Monero nodes? This is a big issue for other blockchains like BSC, and while Monero looks like it is not so far down the line, the issue of node size should eventually come up, right?
-
selsta
blockchain size?
-
moneromooo
Assuming you mean the size of the data on the chain, you can look at posts from ArticMine on bitcointalk and monero's meta repo issues.
-
moneromooo
You can also experiment by syncing a new node and taking file size measurements every month or so. If you do that, beware the monero block time went from 1 minute to 2 minutes a few years ago.
-
moneromooo
You can see the effect of various forks with that (ie, rct, bp...).
-
Guest4
I got thrown out for a minute. Did you write something after the message about ArticMine and bitcointalk?
-
moneromooo
< moneromooo> You can also experiment by syncing a new node and taking file size measurements every month or so. If you do that, beware the monero block time went from 1 minute to 2 minutes a few years ago.
-
Guest4
Got it, thanks. Will take a look at the stuff you've suggested.
-
moneromooo
"every month" here means when the sync reaches the block height corresponding to one extra month, ofc.
-
moneromooo
You can lookup seraphis too, it'll likely be added to monero within a couple years, and will change tx size, and unprunable tx size.
-
moneromooo
You can also look at the pruning patch (b750fb27b0f20e9443827732b69a504a76036430), which prunes 7/8 of the signature data.
-
moneromooo
Which approximates to about 2/3 of the database size.
-
moneromooo
This ratio will change based on the ratio of prunable vs unprunable size in a tx.
-
moneromooo
Also with the total tx throughput.
-
Rucknium[m]
tevador: Is select_outputs.density in master or now just in PR #8024?
-
binaryFate
-
Rucknium[m]
jeffro256: It is possible to do multiple or sequential hypothesis tests in a statistically valid way. I don't think the current software tests in the codebase are formal hypothesis tests. To do it in a statistically valid way you would first need to have p-values or something similar.
-
Rucknium[m]
These are just "sanity checks"
-
luigi1111w
.merges
-
xmr-pr
8691 8692 8693 8697 8699 8700 8705 8706 8712 8713 8714
-
moneromooo
Hey Sysiphus1111w.
-
luigi1111w
wat did I do
-
moneromooo
Not enough, clearly.
-
luigi1111w
agreed
-
luigi1111w
oh I see I'm laborious and futile
-
moneromooo
lol, it just felt like whatever you do, you always end up with a non empty list. Like Sisyphus. Or wherever the y is.
-
xfedex[m]
I've noticed that monerod gets the number of already_generated_coins by doing a database call (`m_db->get_block_already_generated_coins`). Do you think this could slow down syncing on hard drives (where random reads take a relatively long time)?
-
luigi1111w
:)
-
xfedex[m]
This function is called quite often, without checking if tail emission has already started
-
moneromooo
Is this a random read ? I'd think it's pretty much always the last block, no ?
-
moneromooo
Also, this is in the block data, which is fixed length and there's a function to batch read the last N IIRC.
-
moneromooo
You'd need to log the pages that get hit to see how random it is.
-
moneromooo
It's also likely cached from one call to the next.
-
moneromooo
(so not hitting disk)
-
moneromooo
Now, if some of these calls are duplicative, removing is good.
-
moneromooo
ie, two reads from the same txn with no write in between.
-
someoneelse49549
Hi, why in the BlockchainDB class, sometimes functions return a bool and modify a mutable reference as the final data. And sometimes it just return the data?
-
someoneelse49549
why is it mixed like that?
-
rbrunner
Well, because a lot of people with a lot of different coding styles, and also level of experience, worked on the code.
-
rbrunner
There are things much, much worse than that in the codebase.
-
UkoeHB
someoneelse49549: sometimes you need a 'try_dosomething()', and sometimes you need a 'dosomething()'.
-
someoneelse49549
alright. As an heretic i'm going to convert all this try & do into 'deal with Result<>' then.
-
UkoeHB
are you writing a fork or something?
-
-
someoneelse49549
UkoeHB: yes a naive attempt
-
someoneelse49549
trying to rewrite it make me learn the codebase.
-
ghostway[m]
wdym hidden btw?
-
ghostway[m]
you cant "hide" trait methods
-
-
rbrunner
That looks pretty experimental
-
someoneelse49549
i didn't invent it
-
someoneelse49549
rbrunner: Yes, but hopefully with enough effort it could turn into a working thing
-
rbrunner
In any case, you will certainly learn a lot
-
ghostway[m]
<someoneelse49549> "yes.png" <- _yes_.png and _yes1_.png are the best
-
shalit[m]
rust🤮🤮🤮
-
someoneelse49549
<shalit[m]> "rust🤮🤮🤮" <- I don't understand why people hate Rust so much. And at this point i'm too afraid to ask
-
rbrunner
It's a while ago I vomitted looking at some code. Seen too much :)
-
someoneelse49549
hope my code doesn't bring u back a trauma
-
someoneelse49549
what's the difference between tx, pruned_tx and prunable_tx?
-
jeffro256[m]
The names are sort of confusing. A pruned_tx is basically only the tx_prefix part of the transaction. So input mixrings, outputs, amounts. Prunable_tx is the part of the transaction which is pruned: RCT sigs, bulletproofs, etc. A normal tx is both things
-
jeffro256[m]
pruned = what is left after pruning. Prunable = what is dropped from a full transaction
-
tevador
-
selsta
will try to get it fixed
-
jeffro256[m]
-
UkoeHB
yeah merged with failing test luigi1111w ?