Articles & Books

Sometimes you get things wrong--Marshall Clow

What is the best thing to return?

Sometimes you get things wrong

by Marshall Clow

From the article:

A few years ago, Sean Parent challenged me to provide an implementation of Boyer-Moore searching in C++. I did that, first in boost and then, later as part of the proposed Library Fundamentals Technical Specification.

The idea here is that you have a searcher object, which encapsulates the actual searching. You construct it with the pattern that you want to search for, and then you call the searchers operator() with the corpus that you want to search, and it will return to you the start of the pattern in the corpus, if it exists, and the end of the corpus, if it does not (this is the same set of rules that std::search follows).

But this weekend I realized that this is not the right thing to return. The searcher (and std::search for that matter) should return a “range” (ok, a pair of iterators) denoting the beginning and end of the pattern in the corpus. (Yes, you can get the end of the pattern by incrementing the returned iterator by the length of the pattern, but that’s an O(N) operation if you only have forward iterators...

Raw loops vs STL algorithms

A new post on the Meeting C++ blog, this time on <algorithm>

Raw loops vs. STL algorithms

by Jens Weller

From the article:

Since last week I am working on my CMS for static HTML pages again, and so the series about Building applications with Qt and boost continues. Today its about using STL algorithms, or how Sean Parent once said "no raw loops!". Now, I am not Sean Parent, and not event the implementers of the STL are perfect. Most code which I write is application code, which then powers Meeting C++. Also, I don't know all STL algorithms, and some times its just to tempting to write a little loop instead of searching the STL for the specific algorithm. Yesterday I had such a case.

Emulating C++17 Structured Bindings in C++14--John Bandela

How to use tuple return values with ease?

Emulating C++17 Structured Bindings in C++14

by John Bandela

From the article:

Bjarne Stroustrup back in Novemeber wrote a nice progress report, available here, of the Kona meeting. One of the proposals considered is called structured binding. The proposal addresses one of the inconveniences of returning multiple values from a function using tuples. While, it is very easy for a function to return multiple values, it is harder for the caller to use them. Here is an example from the write up.

consider the following function

tuple<T1,T2,T3> f() { /*...*/ return make_tuple(a,b,c); }

If we want to split the tuple into variables without specifying the type, we have to do this;

auto t = f();
auto x = get<1>(t);
auto y = get<2>(t);
auto z = get<3>(t);

The proposal puts forth the following syntax instead

auto {x,y,z} = f();               // x has type T1, y has type T2, z has type T3

I am excited for this feature, and for C++17 in general. While waiting for C++17, I decided to see how close I could get with C++14. Here is the result.

auto r = AUTO_TIE(x,y,z) = f();               // x has type T1, y has type T2, z has type T3

// Unlike the C++17 feature, you need to use r.x instead of just x
std::cout << r.x << "," << r.y << "," << r.z << "\n";

C++ User Group Meetings in February

The monthly listing of upcoming C++ User Group meetings at Meeting C++:

C++ User Group Meetings in February

by Jens Weller

From the article

The monthly overview on the upcoming C++ User Group meetings! In the shortest month of the year there are still 21 C++ User Groups which are meeting!

There are 7 new C++ User Groups: Sofia, Ho Chi Minh, Iasi, Noida, Macedonia, Buenos Aires, Stuttgart / Ludwigsburg (Qt).

Constructor Failures--Arne Mertz

Constructing or not constructing, that is the question.

Constructor Failures

by Arne Mertz

From the article:

Sometimes we fail to acquire a needed resource or responsibility during the construction of an object. Sometimes the construction of a subobject fails. How can we deal with an incompletely constructed object?

 

Experiments in partition and remove -- Brent Friedman

An analysis and exploration of partition and remove algorithms. By combining aspects of these two, we rediscover useful algorithms that the Standard Library doesn't have.

Experiments in partition and remove

by Brent Friedman

From the article:

The C++ Standard Library exposes dozens of powerful algorithms for use in everyday software. Two of these algorithms, partition and remove_if, provide similar functionality for segregating data. If we dig carefully into implementations of these algorithms, a certain symmetry is exposed. This exploration will lead us to the discovery of useful algorithms that we can employ in our daily engineering work.
This article demonstrates particular algorithm implementations — but it is important to remember that the source code will vary between different implementations of the standard library, and can differ based on what sort of data you pass in.

Using parallelism with boost::future

A new blog entry about parallelism and boost::future:

Using parallelism with boost::future

by Jens Weller

From the article:

... While I'm fine with that the application locks up kind of hard during writing a GB zip file (its only job), I'd like to be as fast as possible. Thats why I decided to parallelize the part of the application that reads the file paths via boost::filesystem...

C++ in 2016

A short overview on what is to expect from C++ in 2016:

C++ in 2016

by Jens Weller

From the article:

Like in the previous years, a short outlook into the fresh year regarding C++...

A customizable framework -- Andrzej KrzemieĊ„ski

How to write customizable framework which would work on "practically any type". This article is continuation of author previous post: "Overload resolution".

A customizable framework

by Andrzej Krzemieński

From the article:

We want to provide a function (or a set of overloaded functions) that would ‘do the right job’ for ‘practically any type’, or for ‘as many types as possible’. As an example of such ‘job’ consider std::hash: what we want to avoid is the situation, where you want to use some type X as a key in the standard hash-map, but you are refused because std::hash does not ‘work’ for X. In order to minimize the disappointment, the Standard Library makes sure std::hash works with any reasonable built-in or standard-library type. For all the other types, that the Standard Library cannot know in advance, it offers a way to ‘customize’ std::hash so that they can be made to work with hash-maps.

C++11 threads, affinity and hyperthreading -- Eli Bendersky

How to use C++11 threads library for setting various attributes related to thread affinity/hyper-threading.

C++11 threads, affinity and hyperthreading

by Eli Bendersky

From the article:

This post is not a tutorial on C++11 threads, but it uses them as the main threading mechanism to demonstrate its points. It starts with a basic example but then quickly veers off into the specialized area of thread affinities, hardware topologies and performance implications of hyperthreading. It does as much as feasible in portable C++, clearly marking the deviations into platform-specific calls for the really specialized stuff.