matlab-strings-howfast

Benchmarks for Matlab string arrays vs. cellstrs and MCOS.

This project is a set of benchmarks to measure how fast the new (R2018a-ish) Matlab string array type is compared to cellstrs and naive MCOS (user-level classes) implementations are. Its purpose is to provide feedback to MathWorks so they can improve the performance of the string type. (Because I think string arrays are Good and that's what everyone should be using, so it'd be nice if they were fast, too.)

This project may also be expanded to include benchmarks for categorical arrays, too, since they're closely related to string arrays.

Licensing note

You should not use this project to benchmark any product that is a competitor to Matlab (like, say, GNU Octave). That could be a violation of the Non-Compete clause of the Matlab license, and you could lose your Matlab license because of it.

If you have a problem with this, contact your MathWorks account rep, and/or your lawyer. Definitely do not contact me about it.

Usage

Download the source tree, unzip it somewhere on your disk, and stick its Mcode/ directory on your Matlab path. Then see the doco for the individual classes for how to use them.

Sorry I don't have a better tutorial here.

Contributing

Contributions of all forms, including just comments and feature requests like "I wish it tested such-and-such" are very welcome. Post them as Issues on the project Issue tracker page on GitHub.

Authors and contact

matlab-strings-howfast is written and maintained by Andrew Janke [email protected].

The project home page is https://github.com/apjanke/matlab-strings-howfast.