Javier Bezos (famous of titlsec
, tensind
, spanish
… packages; and now maintainer of babel
) wrote a nice book on typography and scientific notation in spanish.
He says (pag. 82) that the space between the number and the unit should be a thin space on displayed equations; but a word space on the text. I couldn't find this distinction in the SI documentation; and I have no access to the ISO 80000.
I think that siunitx
puts a thin space always. (I made a small text and \showoutput
shows the smae glue and boxed between the number and unit.
I look to the siunitx
code but I can not say say if \SI
checks if tis in a display equation or in a in-line one.
My questions:
-
Is the advice given by Javier an international convention? or it is only applicable to the Spanish language?
-
Would it be possible to by
siunitx
to automatically detect the kind of equation and apply different spacing?
EDIT 1
From Javier's answer I understand that the typography convention that he explained in the book will work as following
In the text I will write sentences like the following. The value of the parameter $a$ is $5\text{~}\mathrm{m}$, but on the other hand, $3\,\mathrm{km} + 500\,\mathrm{m}$ and probably $c = 3005\,\mathrm{m}$. Finding a displayed equation with units should be possible: \begin{equation} c = 3000\,\mathrm{m} + 5\,\mathrm{m} \end{equation} And isolated magnitude as display equation can be the answer to an exercise. The result is \begin{equation} \boxed{3005\text{ }\mathrm{m}} \end{equation}
Best Answer
At a technical level, detecting whether
\SI
is used inside display math is already implemented for thedetect-display-math
option. Thus adding functionality to make a choice of spacing based on whether\SI
is used inside display math or elsewhere is quite feasible. Presumably this would require splitting thenumber-unit-product
into two parts, with an alias meaning that the existing option continues to work: as usual, suggestions for names would be welcome. My main concern with adding such an option would be that I've never seen this type of variability in printed material: I'd like to see an 'official' source for such an approach before adding it.