Expectations of Copy-and-pasting formulæ

Paul Libbrecht, PH Weingarten
Matija Lokar, University of Ljubljana
CC-BY

Expectations of Copy-and-pasting formulæ

Paul Libbrecht, PH Weingarten
Matija Lokar, University of Ljubljana
CC-BY

Defining selection, copy, point and paste
Motivation
What’s working thus far
Expectations: types and examples
One flexible approach?
Future

Definitions

Any piece of information or knowledge that can be selected
- should be copyable
- i.e. put in a clipboard
could also be input
- recreated or simply pasted from the clipboard

Technically

Within MacOSX, Windows, GNOME and others the clipboard is
A record: content-types → array of bytes
Applications are free to record whatever type they want
Pasting: apps are expected to select the best type
gotchas:
- there may be parallel data (e.g. internal references)
- agreement on content-types…

So you want to copy?

Antiquity n°1: Just copy text syntax
- syntax match? try TeX with macros
Antiquity n°2: you want standards?
- Use Edit : Copy special : Copy as MathML
Antiquity n°3: a universal language exists (e.g. Wolfram’s)
- The nifty function prime-check-with-a-hardware-chip-that-removes-your-privacy will not be available in OpenOffice soon

Motivation

Several tools, each with their specialty
- need inter-application communication
Facilitating the input is a key cognitive support
- It didn’t accept my answer as correct, despite clear syntax (checked by the syntax checker) and my answer being correct. (with T Smith, paper at MathUI 07)
Somehow, the issue was not observed
- Internal copy and paste of course insures the quality of the copy

Expectations
Svedek: Expectation / Očekávání CC-BY-NC-SA

Expectation: I am not dumb

Users are fully able to understand the encoding differences
- but refuse to be cognitively loaded with it
E.g. differences of syntax or context
Ok if it can’t input half-a-fraction…
- but should still do something useful!

Expectation: Natural Selection

Formulæ are often understood as text
- horizontal selection typically natural
- kind of ok for simple polynomials
Feedback is key!
- MathML selection miserable thus far (a pack of letters)
- e.g. TeXmacs’ tree-selection
Q: is multiple selection display useful?

Expectation: A single copy and paste

Multiple copy operations are generally useless
- E.g. Copy special, Copy as…
- It should depend on where you go to
Often used currently because users want to be control
- Mathematicians who know the syntaxes
- Persons who want a chance to hack the result

Expectation: Choose the One Right Thing

Easy strategies:
- Going to a text editor? plain text (or… TeX? Mathematica?)
- Going to a text-processing? display formula, styled text, pictures
- Going to a CAS? computable formula, s-expression, display formula, styled text? pictures?
- Going to a dynamic geometry figure: i2g construction, computable formula, …
And, in all cases: internal data and proprietary format should take precedence (e.g. Maple has prepared something for Mathematica)

A Surprisingly Complete Solution: the Wiris Input Editor

The current Wiris Input Editor is in JavaScript
- and edits formulæ for display only
Any selection is reflected a MathML text (the only widely allowed format)
Receiving applications can sniff MathML and act accordingly
- could contain computable and display formulæ
- multiple formats (also TeX or differnt computing levels)
Probably a temporary solution until MathML is accepted as safe

Tracking expectations

Expectations above are from our experience:
But system makers know their users well too!
- and they mostly ignore standards
How can they systematically report?
- Probably a wiki powered by a clipboard inspector
Many many many surprises expected: e.g. style breaches, issues of translatations, representation transfers we did not think of
- deep into mathematical competencies

Conclusion

Copy and paste still unpredictable
Could be a game to teach students that they need to think
a lot before c-n-p ing
- Is this a matrix? you’d better use a text editor’s regexp to replace the brackets
- Is this a function definition? Make it computable by assigning it…
Computers seem not there to help but to rob users’ time.
- Lots of natural expectations not assumed
- Reporting could help