i dont think you need to know all the internals in zero to monero to create an exchange web app :D

thats a bit of an overkill isnt it?

selsta: There has been quite a jump of the compilation time since the last 2-3 merge sets, but I just wanted you to know that it was NOT due to your C++11 -> C++14 change, because this jump happened earlier.

something ~20 commits before HEAD?

release or master branch?

<sech1> "something ~20 commits before..." <- This is all master.

and the incrementation is not really per each commit, but a set of commits. See the 1st two leftmost columns:

Reporting back on the gamma test issue: the addition of the "recent spend window" (where the gamma picker redraws a recent output for picks <10 blocks old) causes the test's median output age to drop from ~116,500s to ~115,000s, which is just close enough to the test's lower bound of 112,320s to trigger the occasional failure with a sample size of 100k picks (exactly 10 in 1000 tests failed when I ran it locally, so ~1% failure rate)

When I run the test with a gamma picker that matches Miller et al exactly (removed the recent spend window, allowed outputs <10 blocks old to be picked by the gamma, applied distr from chain tip), I get a median of around ~115,000s as well. Which is good, it means the current implementation is now more in line with the paper

The reason the paper's expected median is ~129,000 when plugging into the formula, and this test's median is way less than that, is because of the long tail where draws that are too old get thrown away. A very large proportion of draws get thrown away in this test because there just aren't that many outputs in the sample chain. This skews this test's median to be much younger than the gamma formula's expected median

Will put a fix out for the test soon probably in line with what moneromooo said in mrl. Basically nothing unexpected is happening here, all is well :) I'll add a deterministic test as well

<sech1> "something ~20 commits before..." <- And I'm not trying to pinpoint the exact commit here. All these commits were necessary for improvements. The regressions, that I show, are just side effects (like waste) of something better. Simply from time to time the waste has to be cleaned up, independently of the production.

jberman[m]: thanks for looking into it :)

mj-xmr[m]: at least there doesn't seem to be a huge increase due to C++14

selsta: Quite the opposite.

And this brings even more potential of reduction, by ditching bits of Boost.

Somehow compilation fails when building release branch under C++14

and I can't figure out why

some boost iterator error

also it only happens on Mac as can be seen here monero-project/monero #7997

selsta: my gut feeling tells me, that it expects `cend` instead of `end`

`/usr/local/include/boost/range/concepts.hpp:302:34: error: no matching function for call to 'end'

const_iterator ci2 = boost::end(const_range);

^~~~~~~~~~`

but since it's inside boost, all you can do is to bump the Boost version.

`-- Found Boost Version: 107600`

Try 1.77.0 ?

yep will try

but still don't understand why master compiles fine

Sounds like you're not including the header where that is.

Hello. What is causing this error?

```

2021-10-19 11:21:11.941 E failed to parse version number unknown

```

Break where it prints this (abort() would do), bt in gdb.

The shortest, yet correct debugging tutorial ever.

BigmenPixel[m]: GUI?

selsta: yes

does it cause any issues,

?

selsta: Monero gui version is not shown in Info section

* Info section (unknown (Qt 5.15.3))

* Info section (unknown (Qt 5.15.3)

* Info section (unknown (Qt 5.15.3))

is that with flatpak?

selsta: yes

git / git repository is required during build to get the version number and commit hash

though that can be improved at some point

selsta: Perhaps this is due to the limitation of the Internet after downloading the source files?

no, you have to be inside a git repository

download source files doesn't contain it

.git folder

it would require some cmake changes to support it

selsta: OK, thank you for answer

BigmenPixel[m]: will take a look later

BigmenPixel[m]: monero-project/monero-gui #3715 should solve it

selsta: ok, I'll try to apply this patch later

* patch later(or now...))

BigmenPixel[m]: updated it again, it should display "v0.17.2.3-unknown" now

you can overwrite the "unknown" with the `-D GUI_VERSION_TAG="release"` cmake option

but I don't know how the flatpak script works exactly

selsta: Is there a mistake? I just collected monero-gui with the patch and the version is still unknowт

hmm wait

Well, let's build again :)

selsta: without it

with the patch that I updated 5 mins ago?

selsta: no

<selsta> "you can overwrite the "unknown..." <- ok

I tested it locally and it works here

selsta: with an old or new patch?

with 3715 (before and after force push)

selsta: I'll try to build with a new patch.

Maybe I was wrong somewhere...

s/unknowт/unknown/

selsta: :)

nice :D