Alfred Klomp

Hi, thanks for the suggestion. Currently, this library claims the `base64` namespace for all global functions, macros, structures and other objects exposed in the public headers. I guess this can...

That sounds like a reasonable solution. However at this point it's hard to implement, because of the archaic way that the build system handles symbol visibility. All visible symbols are...

So the AVX detection was fixed (thanks @mayeut), which leaves the macOS libtool problem. @mscdex, could you provide some more information about what exactly is going wrong, such as terminal...

I'm a bit unclear on what exactly you want to do, and with which data type. Do you want to base64-decode an `uint16_t` array which you get from V8? In...

So the proposed solution would be to duplicate the API and the underlying logic to add first-class support for these wide chars (as I'll call them). I'm reluctant to do...

If you're not using the streaming functions, why not use a two-stage pipeline and preprocess the input string using SSSE3 functions before sending it to the decoder? That should be...

The reason to ignore invalid bytes on decode is to skip newlines and carriage returns, right? I've been thinking about adding an option to the library for that actually. As...

Here's the scheme I've come up with so far. It's probably not optimal, and I didn't benchmark or even compile it. Say we have these nine valid bytes of input....

@mayeut while I think your proof-of-concept is remarkable (how are you so damn _fast_? :) I'm hesitant to merge something like it for the reasons mentioned above. In my opinion,...

First off, there may be a license conflict: Boost.Predef is licensed under the [Boost license](http://www.boost.org/users/license.html), whereas this project is BSD 2-clause. Maybe that's not a problem if we ship an...