Tristan Stenner

> I don't have time right now to modify the source code and check what the minimum number of samples in the buffer can be---probably 1 is ok but what...

> > Pull everything in a chunk and take only the interesting samples? > That is only guaranteed to work if the max chunk size is greater than the buffer...

> We are not really breaking the API. We are fixing functions that have always been bugged. Which still crashes with combinations of programs compiled against the wrong library version...

> Right, it's tricky to implement. Not with the Boost Queue, see [spsc_queue::read_available()](https://www.boost.org/doc/libs/1_67_0/doc/html/boost/lockfree/spsc_queue.html#id-1_3_22_6_4_1_1_1_13_14-bb).

That's one possibility, but I'm currently thinking about implementing it instead. Would be easier if Matlab had a binary search built in, but I'll manage without it.

As far as I can see, the SIFT function is functionally identical to the one in `eeglab\functions\statistics\`, so rather than fix this one I propose to remove it from SIFT...

First, install the build environment (see [here](https://github.com/labstreaminglayer/labstreaminglayer/blob/master/doc/BUILD-ENVIRONMENT.md); short: install g++ / clang++, CMake, and optionally git). Then, clone the [liblsl repository](https://github.com/labstreaminglayer/liblsl) or download and unzip [the archive](https://github.com/labstreaminglayer/liblsl/archive/master.zip). In the downloaded...

To speed things up, you could try [a branch with some compilation speed optimizations](https://github.com/tstenner/liblsl/tree/miniboost) and/or install ninja beforehand (add `-G Ninja` to the first cmake invocation)

See also https://github.com/labstreaminglayer/liblsl/pull/2

> From what I have heard, it may take 3 hours On a plain 3B without overclocking and a slow SD card it takes about 13 minutes, but you have...