วีดีโอ

23:30 this is a fantastic trick by the way, I was running into this same issue myself recently of needing to deal with function overload sets and templates, and I really appreciate you sharing this

Very interesting talk. I didn't know declval always returned a reference type, that's quite a big corner to be wary of. I wonder why it was done that way initially?

If I may ask, what was "super hot previous talk"?

Great video. Thank you for sharing.

Clearly having a functioning package manager would make developing and learning C++ easier, which seems to be the opposite of what is wanted.

Very good overview. Thanks a lot

Great talk!

Yeah, only disasters in C++. No "feature" without it. Time to switch to something better.

I'm a simple man. I see A. Stepanov's (collaborative) books mentioned in a talk (4:54), I upvote and watch it till the end carefully.

flat_set and flat_map are useless, just look at the benchmarks. It's one of those cases when you think Y will be faster than X, but then you measure and it's the other way around.

Very interesting, however i would like for the questions to come after rather than during the talk.

Great talk. Thanks Joaquín and José Daniel.

I believe that this is the best introductory talk to coroutines. Not only you get the concept clearly in this hour, but also you practice it. Thank you Dietmar! TO ADMINS: Please add to the title of this video something like: [Hands-on Introduction to C++ Coroutines]

Happy that I made it to that conference, it was very well organized and fruitful one!

What a terrific leap forward along so many different dimensions! (No pun intended.) Breaking changes shouldn't be done lightly, but this talk makes a compelling case for doing so, and I'm relieved these new APIs were invented before the library made it into the standard (at which point we could never have taken advantage of them).

hi, hola, busco curro de desarrollador c++

6j6fyhhj

Mónada (del griego μονάς monas, "unidad" de μόνος monos, "uno", "solo", "único"),​ la fuente, o el Uno, de acuerdo con los pitagóricos, fue un término para Dios o el primer ser o la unidad originaria, o para la totalidad de todos los seres, con el significado de «sin división».

Buen video sigue así! 💃💃💃💃💃💃💃

Thanks for the referral I just purchased a good dL lookup site from *thomas davy* on telegram he is so nice.

Thanks for the referral I just purchased a good dL lookup site from *thomas davy* on telegram he is so nice.

Is there a version of this delivered in English somewhere?

Hasta ahora nadie ha sido capaz de explicar bien claro que significa realmente una monada, es decir que es realmente y como se construye, si es que se construye, aquí solo veo matemática y nada de claridad sobre realmente que es lo que es eso que llaman monada, crei que lo explicarían de manera física no matemática.. En fin seguiré buscando hasta comprender eso..

BIEN , ESTO TIPO DE FOROS DEBERIA SER DE INTERES DE TODO INVESTIGADOR

Muy buen video gracias por compartir

NO LOGRO VER EL COMETIDO DE QT. COMO CREAR UNA INTERFACES, MUY POBRE LA EXPLICACIÓN,

es lo mismo que dice Leibniz?

cree una maquina del tiempo y acabo de volver desde la monada inicial, estamos inmersos en su función y la sumatoria seguirá hasta la siguiente ecuación así sucesivamente. 0 = 1-1= 1+1-2=0 = nada = todo infinitamente con todos los números en todas las direcciones, como una esfera, como un teseracto y todas la dimensiones hasta regresar a la monada. deconstruyendo la realidad siendo todo ,al mismo tiempo nada. todo = nada (sin un sentido posmodernista) si no con un sentido a lo Baruch Spinoza tenia razón, dios es la naturaleza el todo. antes de nacer, todas la piezas estaban en diferentes ecuaciones y asi seguiran hasta la monada

Me alegro mucho de escucharte, te conocí, cuando eras pequeño, felicidades, y saludos a tus padres, desde Galicia Luisa

Muchas gracias, una charla interesante :)

Aprendan la pronunciación y acentuación de las palabras en Ingles Learn the pronunciation and stress of English words

Se está meando?

Is the auto problem still present - littering/putting the auto keyword between template-concept-type and identifier?

Talk begins 6:00

Monadas la fuente o el UNO de acuerdo a lo pitaforico o el primer SER o la UNIDAD originaria o para la totalidad de nuestra TODOS los SERES con el significado de SIN DIVISION.....universo sin division, individuo sin division?vacio .....?digo pitagórico arriba.......chau

El patron Singleton me ha sacado de muchos apuros...

El Ermitaño, te estoy buscando.

Muy buen Video pero me puedes resulver una duda. Como hago para que un "for" se me autocomplete como por ejemplo en java se pone un "fori" tabulas y te lo escribe solo

Muito bom!

🤔Colijo en mi entender, que una Optiónal"T" (opción alternativa) sirve para aislar, continente de contenido o al revés, contenido de continente. Si no estoy en lo cierto, por favor, espero que el profesor me corrija. Gracias.

The slides in english: github.com/Juanmihd/talks/blob/master/CppCoreGuidelines%2C%20part%201.pdf

Muy buen video recien me entero que siempre usaba coroutines en javascript, se nota que a muchos desarrolladores de software comercial nos falta la parte conceptual, y estos videos mitigan estas deficiencias. Gracias.

Muy interesante

github.com/Microsoft/GSL github.com/isocpp/CppCoreGuidelines

3' we can not have different implementation of any class, standard or not in a program without going against the One Definition Rule. I don't understand what you are talking about the library side and the user side that are using different memory management. This is ill formed for me, but It may be very well wrong. Whether a given library is ensuring you an ABI compatibility between different version is maybe different, but I believe it fall on the same ODR issue. One part of your application will be using a version of the library and other part using a newer version.

1:12:00 Why is there a picture of Douglas Adams' Starship Titanic on this slide? Is there some reasoning behind it?

TL;DR: You are basically right about "reserve() is evil". Just stick with that. 25:00 Actually, you shouldn't call reserve() here *at all* unless a) You want the code to only work with strings, vectors and some third-party data structures (yes, not even std::deque has reserve() ). b) You can prove that this *actually* improves performance with a *benchmark* . If I had to give a breakdown of reasons why people try to outsmart their C++ implementation - which is already *required by the standard* to do exponential scaling or whichever other way they have of achieving amortized constant vector::push_back(), string operator += etc. -, it would be this: - They are experts at designing data structures and have discovered a genuine way of speeding up the program (probability <= 1%). - They are too clever by half and actually have no clue how their C++ implementation works in the first place (probability >= 99%). I would also immediately flag code like this in code review unless it is backed up by solid performance data (documented in comments, ideally even with a link to the benchmark code that was used to obtain it). There is one common use case where calling reserve() can make sense: vector<X> vec; vec.reserve(known required size); vec.insert(…); // rsp. loop of push_backs rsp. algorithm with back_inserter etc. // and then vec is not modified in a long time This can save memory and does not carry any risk of the reserve() pessimization that you mentioned. Other than that, reserve() should be treated with some *reservation* (yes, I am sorry for that).

"We're hiring!" "Unless you're male and white!"

Very very interesting, but not really sure about the fun... :D

For fairness (and I like to mention this when the topic of iostream is brought up), *most* of the slowness in iostream is due to implementation laziness / stupidity, plus a questionable design choice in the standard that is however configurable. In particular, output does *not* require virtual stream buffer calls if the stream has sufficient space in its put area (i.e. buffer). This was definitely considered in the design. As for implementation laziness, I have e.g. discovered that one reason why iostream implementations are slow is that they usually overload the bare minimum of virtual methods in their stream buffer classes. The best example is that most implementations do *not* overload xsputn() but only overflow(). (Yes xsputn is a horrible name, but that's another discussion - think "excess put n (characters)". Even back then, RAM and harddisks were no longer small enough to justify this highly frugal naming approach.) So, what xsputn() is designed for is to write additional character data in case the put area is full (or non-existent), it receives a buffer pointer and a size. Ideally, it would write as much data as possible directly to the sink - e.g. the OS file descriptor - with very high efficiency. However, as I said, it is usually defaulted. Guess what the default implementation does: call overflow() for every character (overflow() accepts a single character). The default does make sense in some way since overflow() is required and this method should always work assuming the other methods are implemented correctly. However, this of course entails one vcall for *every* character. Obviously, this is highly inefficient, and most streams could easily provide an xsputn() that makes more sense. As for the questionable design issue: that's simply that standard iostream instances (cin, cout, cerr, clog) are by default thread safe and synchronized with the C standard functions printf(), fprintf(), puts(), fputs() and so on. The good news is if this is not needed, the performance can be improved by calling some_predefined_stream.sync_with_stdio(false) *before* doing any I/O through it. (I can partially understand the "correctness over performance" approach of making synchronicity the default - OTOH, "you shouldn't pay for what you don't need" has long been, and still is, a popular philosophy pertaining to C++.) Another widespread cause of slowness in iostream is the use of locales and facets (use_facet() is typically implemented in terms of dynamic_cast, which itself is usually terribly slow). E.g. outputting an int requires a lookup of the num_put facet within the imbued locale. While this allows using decimal group separators and other fancy things like this, is places a heavy burden on the default case that almost always suffices. If you want to know many more details, see Langer, Angelika & Kreft, Klaus: "Standard C++ IOStreams and Locales".