00:50 <gonidelis[m]> hkaiser:

00:50 <gonidelis[m]> https://github.com/gonidelis/hpx/blob/master/libs/full/performance_counters/include/hpx/performance_counters/counters_fwd.hpp#L33-L144

00:51 <gonidelis[m]> https://github.com/gonidelis/hpx/blob/master/libs/full/performance_counters/include/hpx/performance_counters/counters.hpp#L63-L173

00:51 <gonidelis[m]> any reason we define the documentation twice?

00:57 <gonidelis[m]> why ^^

01:24 <gonidelis[m]> <K-ballo> "some things have to be in..." <- seems more like headers are the go-to solution and then you basically retain a couple small sources to "run". at the end of the day we are building a libarary, I wouldn't consider headers either forced or lazy

01:24 <gonidelis[m]> maintain^^

01:24 <K-ballo> no

01:25 <K-ballo> both static and shared libraries are composed only of source files after all

01:26 <K-ballo> header files, on the other hand, are copy-pasted into each source file in which they are #included

01:27 <K-ballo> in old style C headers weren't even necessary, they were more of a luxury convenience

01:38 <gonidelis[m]> K-ballo: so headers are mainly for name conflicts avoidance

01:38 <gonidelis[m]> libraries are composed ONLY of sources? wow

01:39 <K-ballo> name conflicts? what do you have in mind?

01:40 <K-ballo> library *files* (.a/.lib, .so/.dll) are composed only of sources

01:40 <gonidelis[m]> actually the oposite

01:40 <gonidelis[m]> reusable code i meant

01:41 <K-ballo> headers are for interfaces

01:41 <gonidelis[m]> yes

01:41 <gonidelis[m]> interface <- reusable code

01:41 <gonidelis[m]> not the other way around necessary

01:41 <gonidelis[m]> but why the sources?

01:41 <K-ballo> I don't understand the question

01:41 <gonidelis[m]> why can i not write a library that is just comrised of .hpp's?

01:42 <gonidelis[m]> and just be included in the user's main

01:42 <gonidelis[m]> source file

01:42 <K-ballo> you can, it's just a terribly poor practice

01:42 <gonidelis[m]> alright

01:43 <gonidelis[m]> excellent

01:43 <K-ballo> unless you are forced to it, like it's the case of libraries entirely composed of macros, or (pre modules) templates

01:44 <K-ballo> each time your header is included the code gets copy/pasted into the user sources

01:44 <K-ballo> and it has to be parsed, and compiled (and optimized)

01:44 <K-ballo> and later the linker has to come in and de-duplicate all those copies

01:44 <K-ballo> each time the user compiles their program

01:45 <gonidelis[m]> :E

01:45 <gonidelis[m]> i get why it's awful

01:45 <gonidelis[m]> nice explanation

01:45 <gonidelis[m]> thanks!

01:45 <K-ballo> whereas a library source file gets compiled (and optimized) once, no duplicates, and gets linked just as the user asks for it

01:47 <K-ballo> and modules are some sort of parsed source representation (possibly compiled too in some cases), so at least you don't have to keep parsing the same templates over and over and over and over

01:49 <gonidelis[m]> modules?

01:49 <K-ballo> search online for c++ modules for some basic introduction

01:50 <K-ballo> they serve the same purpose of headers, but don't have (all) the downsides

01:51 <gonidelis[m]> wtf

01:51 <gonidelis[m]> i have never ever in my life seen modules

01:51 <gonidelis[m]> like ever

01:52 <gonidelis[m]> like in use i mean

01:54 <K-ballo> you were surprised by the existence of source files in a library mere hours ago

01:56 <gonidelis[m]> lol

01:56 <gonidelis[m]> hahaahhahahahahahahhahahahhaahhahahahahaha

01:56 <gonidelis[m]> no! i get it for hpx wth 😅

01:56 <gonidelis[m]> we have a runtime after all

01:57 <gonidelis[m]> but i was thinking maybe static librarires would be header only

01:57 <gonidelis[m]> actually range-v3 is header only....

01:57 <gonidelis[m]> i mean 'header only' IS a thing

01:58 <gonidelis[m]> i need to disambiguate static from header only at this point

02:00 <K-ballo> range-v3 is pure templates, it's forced to be header only

02:02 <gonidelis[m]> amazing

02:03 <gonidelis[m]> speaking off

02:03 <gonidelis[m]> https://www.cppstories.com/2022/ranges-composition/

02:03 <gonidelis[m]> why is the view "printable" immediately in the first example while in the second one a new string has to be constructed from scratch

02:09 <gonidelis[m]> 🙃

11:36 <srinivasyadav227> hkaiser: yt ?

11:37 <hkaiser> srinivasyadav227: here

11:37 <hkaiser> g'morning ;-)

11:37 <srinivasyadav227> hkaiser: is the SIMD meeting cancelled ?

11:37 <srinivasyadav227> good morning :)

11:37 <hkaiser> yes, we forgot that today is Good Friday

11:37 <srinivasyadav227> oh okay, gsoc meeting is cancelled aswell ?

11:38 <hkaiser> I'd think so, yes - can we meet Tuesday at the usual time?

11:39 <srinivasyadav227> okay sure :) sorry again for missing out on this Tuesday :)

11:39 <hkaiser> nah, no worries

11:39 <srinivasyadav227> thanks! ;)

13:06 <zao> gonidelis[m]: There's two aspects to what goes in "source files" (.cpp) and headers (.hpp) - "mechanical" and "cultural" and it largely boils down to the one-definition-rule (ODR).

13:06 <zao> Mechanically some constructs when compiled in a translation unit (TU, source file with all includes expanded) generate distinct entities each time, these are things like free functions, variables of global scope. Some generate the same thing across different TUs and are interchangeable by the linker like inline functions and expanded templates. Some don't generate anything like extern variables. Member functions defined inside a class definition are

13:06 <zao> implicitly inline. These are hard rules and you have to design around them.

13:06 <zao> Culturally a project may pick among multiple ways to do things and design their libraries accordingly - you can go "header-only" by only using mechanisms that don't need separate source files and can be expressed (with effort and cost) in a set of headers. You have projects with very traditional headers that put as much as possible in source files and keep the header interface as slim as possible, and there's the ones inbetween that may define smaller

13:06 <zao> things inline in headers but also have some of the costlier or bulkier things in source files. For things like HPX that go hard on templates, a lot of code is kind of forced into headers whether one likes it or not.

13:06 <zao> (hah, really screwed up my line length estimates there :D )

13:07 <hkaiser> zao: nice summary, thanks!

13:14 <satacker[m]> Are we missing on the sender adaptor closure object? http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p2300r4.html#spec-execution.senders.adaptor.objects

13:15 <hkaiser> satacker[m]: yep, that's missing

13:17 <satacker[m]> If it's alright with you, could you please mention it [here](https://github.com/STEllAR-GROUP/hpx/issues/5045#issuecomment-868810753) as well?

13:17 <satacker[m]> Or was it implied automatically ?

13:18 <hkaiser> done

13:18 <satacker[m]> Thank you!

13:18 <hkaiser> good catch!

13:44 <ruchipakhle> hkaiser stellar has been selected into the GSOD'22 so we can officially start sending the proposal submission so is there any google forms sent for the same as I am not able to see any communication or previous messages on libera and there is no chat going on in the matrix so am I left behind?

13:45 <hkaiser> ruchipakhle: will publish this by early next week

13:46 <hkaiser> sorry for being slow :/

13:46 <hkaiser> gonidelis[m]: has it almost ready, I believe

13:47 <hkaiser> ruchipakhle: you can definitely start thinking on what you would like to work on

13:47 <ruchipakhle> ohhh no worriess i was just concerned as I was not able to see any messages also is there any way by which I could remain in sync with the community here=L

13:48 <hkaiser> ruchipakhle: this channel is being logged: irclog.cct.lsu.edu

13:50 <ruchipakhle> hkaiser: ohh cool, thanks for sharing

13:50 <ruchipakhle> you can definitely start thinking on what you would like to work on :  yes definitely i have started to go through the stellar documentation thoroughly

13:51 <hkaiser> cool

14:05 <gnikunj[m]> hkaiser: do we have HPX + LCI meeting rn?

14:05 <gnikunj[m]> tried joining the room. Looks like not.

14:26 <gonidelis[m]> hkaiser: ruchipakhle will take care to publish the questionnaire in a bit. Apologies for that

16:50 <hkaiser> gnikunj[m]: Eastern...

16:52 <gnikunj[m]> Aah yeah

16:52 <gnikunj[m]> For some reason, I don't have a holiday today ;_;

16:54 <hkaiser> gnikunj[m]: you're paying to be punished ;-)

16:55 <gnikunj[m]> Fortunately not paying but unfortunately still being punished 🥲

17:14 <ruchipakhle> gonidelis[m] : no worries at all, i just asked coz I did not have logs of whats happening over here and hence asked

17:15 <ruchipakhle> but now that i have I would be patiently seeing that before :)

22:10 <pansysk75[m]> Hey, fellow newcomer here.

22:10 <pansysk75[m]> Are there any specific tools commonly used for profiling and performance testing? I only know of those included in Visual Studio

22:12 <hkaiser> pansysk75[m]: that's a good one for high-level assessments, more detailed information can be collected using vtune (Intel Amplifier)

22:12 <hkaiser> depends on what you're planning to do, really

22:12 <pansysk75[m]> Also what is "Traveler-integrated" for? Some sort of profiling tool as well?

22:13 <pansysk75[m]> Thanks so much!

22:13 <hkaiser> pansysk75[m]: Traveller is a set of web-based visualization tools for trace data collected by other tools (mainly otf2 traces, but I think they support other formats as well)

22:52 <gonidelis[m]> hkaiser: created the GSoD form and published it here https://github.com/STEllAR-GROUP/hpx/wiki/GSoD-2022-Project-Ideas

22:52 <hkaiser> gonidelis[m]: great! many thanks! - should we publicize that somehow?

22:54 <gonidelis[m]> not to the GSoD guys... i think putting it on a top header at our GSoD2022 page ( to where prospect are being redirected from here https://developers.google.com/season-of-docs/docs/participants)

22:54 <gonidelis[m]> is the best way to attract people

22:54 <hkaiser> yah, that's good

22:54 <hkaiser> somebody was asking this morning...

22:54 <gonidelis[m]> yes

22:55 <gonidelis[m]> unfortunately they are out

22:56 <gonidelis[m]> you should have also received an email with the spreadsheet that collects the responses along with privillages to modify the google form if you want

22:56 <gonidelis[m]> gnikunj: ms and diehlpk_work too

22:56 <hkaiser> ok, thanks

22:56 <hkaiser> yep, got it

22:56 <gnikunj[m]> gonidelis: thanks for working on this!

22:57 <gonidelis[m]> np