Discuss non-framework libraries

[Morwenn]

We still don't have any discussion for libraries too small to be considered frameworks. The kind of utility libraries that are often small, header-only, and dedicated to solving specific tasks. Without too much thought about it, I would nominate cppformat which seems more than decent.

[nabijaczleweli]

👍 for cppformat

[ghost]

butt boost::format?

[nabijaczleweli]

@kBoK See cppformat's Boost.Format README section

[nabijaczleweli]

I also propose CPR - a C++14-isation of libcurl

[ghost]

I agree with CPR - it looks nice. Let's decide whether we put CPR or casablanca - obviously we won't just put two competing libraries into the list

[nabijaczleweli]

They are not competing, looks like.

[ghost]

Actually, it's not obvious. let's just say I'd rather not put two competing libraries, and that at some point we should decide which is awesomer. People look at this list to pick a library, so if you leave two of them, then they still have to continue searching. Just my 0.02$

[ghost]

@nabijaczleweli ok, boost::format is slow, but we still need to agree that cppformat is better :)

[nabijaczleweli]

Casablanca:

Features:

• HTTP client/server, JSON, URI, asynchronous streams, WebSockets client, oAuth
• PPL Tasks - A powerful model for composing asynchronous operations based on C++ 11 features
• Support for Visual Studio 2012, 2013, and 2015 with debugger visualizers

A.k.a. it's (a) a framework, and (b) an all-container.

CPR, however is just a C++14 wrapper around libcurl.

[milleniumbug]

👍 for cppformat.

[milleniumbug]

I nominate Niebler's ranges library.

[Morwenn]

A list of the small yet cool libraries that I starred and that might be worthy of official awesomeness:

• Howard Hinnant's date and time library based on <chrono>.
• sqlpp11 which seems pretty cool. It implements a EDSL to manipulate databases and can even consider standard collections as temporary databases. Pretty powerful.
• Eric Niebler's range-v3 as already mentioned.
• I wish I could nominate boost-tree which isn't part of Boost but the subject of a standard proposal. It offers interesting abstractions and several kinds of generic tree data structures. IIRC it wasn't accepted in the standard library because it was too big without enough experience from users. I think it was proposed to Boost too but it overlapped with several Boost modules; I don't know what happened afterwise.
• expected which is an implementation of the proposed std::expected error reporting utility.

All of them are "modern" C++ libraries and much thought was put in their design.

[nabijaczleweli]

👍 the date library collection and sqlpp11
v0v expected
👎 Boost.Tree - old, unmaintained

[nabijaczleweli]

@milleniumbug's seed11

[ghost]

pugixml ?

[milleniumbug]

Posting here so I won't forget: docopt. Not yet nominating, I have to review it first.

[nabijaczleweli]

I hereby propose:

• jsonpp — excellent, yet simple API
• cereal — all-round dope serialisation

@milleniumbug, how's docopt treating yeh?