⇧ [VIDÉO] You might also like this partner content (after ad)
Is it possible to code other than by typing on a keyboard? Obviously, yes. Researchers at Cornell University have created an interface capable of transcribing sketches or lines of handwriting into computer code.
Their tool is called Notate, and is based on the use of a stylus. It is not intended to “replace” the way developers usually code, but rather to complement it. It is intended for those who use digital tablets or notepads, and “ allows users to open drawing canvases in lines of code “, explain the scientists in an article published in the magazine Association for Computing Machinery.
Support us by buying a poster that throws:
Support us by buying a poster that throws:
Developers are therefore invited to introduce diagrams, sketches… In short, information that is sometimes much easier to synthesize in this type of format than by long lines of code. A command is then used to “execute” the diagram, in order to see if it has been understood by the program. Another tool even breaks it down.
” Through this work, we seek to question what it means to ‘write’ code say the researchers. Indeed, they also remind us that although the keyboard is today the tool mainly used for coding, this has not always been the case. ” With the advent of programming, the earliest computer programming notations were handwritten, not typed. In the famous 1945 First Report on EDVAC, for example, John von Neumann equated diagrams with text and vice versa. In fact, notations were not serialized and called programming languages until typewriter interfaces became suitable for programming. “, underline the researchers.
A new programming tool to improve
The interface works through a deep learning model. It bridges handwritten and textual programming contexts. The two are completely related: thus, a notation in the handwritten diagram can refer to the textual code and vice versa. For example, Notate is able to recognize handwritten programming symbols, such as “n”, and then relate them to their typed equivalents.
Scientists have already performed some user testing, with some success. The basic principle seems to have worked, but however requires some improvements to be able to be really functional. “ Results from a usability study with novices suggest that users find our basic interaction of implicit cross-context references intuitive, but suggest further improvements to debugging infrastructure, interface design, and rates. of recognition “, they specify in the introduction to their study.
However, they are enthusiastic about the future of such a tool: ” A system like this would be ideal for data science, especially for building charts and tables that then interact with textual code. says Ian Arawjo, lead author of the study, in a press release from Cornell University. “ Our work shows that the current programming infrastructure is holding us back. People are ready for this kind of functionality, but developers of interfaces for entering code need to take note and support images and GUIs inside code “.
