29073 - Empty tags in <msup> cause strange behaviour

Status: RESOLVED FIXED

(Core :: MathML, defect, P3)

Description

[Paul Gartside]

<msub> and <msup> each take two parameters (base and sub/super script).
If the base is empty for <msub> then it behaves as I would expect - like latex's
${}_n$.
But if the base is empty for <msup> nothing is displayed - unlike latex's
${}^n$.
(See the url or attached file for a clear example.)

From the point of view of tex to mathml convertors it would be preferable if
empty tags were accepted and displayed in a tex like way.

Comment 1

[Paul Gartside]

Attachment: test file

Comment 2

[Paul Gartside]

Any news on this bug? Are there problems confirming  it? Or have better things

to do!

Comment 3

[rbs]

Saw the problem. Accepting bug.

It is necessary to know the height that should be given to the empty element.
For TeX, the base in ${}^n$ a zero-width box, and the height is non-null
(this way, the sub/super script can be shifted w.r.t. that height)

Do you know what should be the height of ${}$?

Comment 4

[Paul Gartside]

I asked David Carlisle what (la)tex does about ${}^x$. Here is his reply:



> For TeX, the base in ${}^n$ a zero-width box, and the height is non-null

>(this way, the sub/super script can be shifted w.r.t. that height)



>Do you know what should be the height of ${}$?



well the height of the box made by  ${}$ is zero, which I suppose isn't

the required answer.



In fact the superscript gets raised by virtue of





\rule 18a. If the translation of the nucleus is a character box, possibly

followed by a kern, set $u$ and~$v$ equal to zero; otherwise set

$u\leftarrow h-q$ and $v\leftarrow d+r$, where $h$ and~$d$ are the height

and depth of the translated nucleus, and where $q$ and~$r$ are the values

of $\sigma_{18}$ and $\sigma_{19}$ in the font corresponding to styles

$C\mathord\uparrow$ and~$C\mathord\downarrow$.  \ (The quantities $u$

and~$v$ represent minimum amounts by which the superscript and subscript

will be shifted up and down; these preliminary values of $u$ and~$v$ may

be increased later.)



ie font parameters 18 and 19 effectively force a minimum shift.



I am not sure of the current state of the sub project to instill

these rules from appendix G into mozilla, in the particular case of

wanting to know what to do about {}^x I'd just use the x height of the

current font. ie raise it as if it was x^x.

Comment 5

[Paul Gartside]

Fixed I believe. Status to `resolved' ?

Comment 6

[rbs]

Resolving as FIXED.