If we have read in recent times that the famous tool for requesting in many protocols is being rewritten with the language created by the Mozilla Foundation, it is not.
Daniel Steinberg has clearly detailed it on his blog. The base libcurl library is written in C and will remain so. Even though the language has its own problems and weaknesses are sometimes detected, none has been significant and their numbers are decreasing.
Above all, this software brick found in millions of machines worldwide is known for, stable and appreciated. A full port would be a pharaonic job, even before Google eventually recalled Steinberg.
However, the design of CURL is flexible. It can be compiled with different back-ends, 33 to be precise. Each can be used to support a particular protocol, with its advantages and disadvantages.
The announcement made a few days ago is about the creation of a new back-end for HTTP requests. It will use Hyper, a library used in Rust, a language considered “memory safe”, because of its design and its specialized memory management.
This back-end, which will support both HTTP / 1 and HTTP / 2, will support quiche and Mesalink, already supported as well as written in Rust. Its construction was funded by ISRG, the organization behind Let’s Encrypt N.
Steinberg says the job will take time. Hyper does not yet have an API that allows it to communicate with code written in C, its development is underway. Only after this step can the actual design of the new back-end begin.
A dedicated branch has been created on the CURL GitHub repository. No date has been given yet on the outcome of the project.
