Re: [EXTERNAL] auto-boxing

Quickly correcting the record:

"Deyan: no tool should auto-box"

I did not say / didn't mean to imply that. I said I see a different
emphasis:  If the natural clipboard UX requires cross-node ill-formed
selections, a tool should/could aim to support that.

The quoted sentence in my previous email was a reframing of your problem as
applied to spans in HTML, rather than my own opinion. Hence the scare
quotes.


On Mon, Jul 12, 2021 at 2:20 AM Paul Libbrecht <paul@hoplahup.net> wrote:

> Hello all,
>
> Deyan: any tool changes the user
>
> indeed so.
>
> Deyan: no tool should auto-box
>
> Well… unless someone insists that this makes sense to protect users from
> ill selection. That was the thought that Murray rendered below, it is quite
> normal.
>
> Neil: the problem is not MathML
>
> well… unless MathML insists on having (lots of) mrows, e.g. because
> intents are better AT-interpreted there. I think I remember Deyan
> indicating that LaTeXML is creating a lot of them. I am under the
> impression that the draft that Deyan circulated about “minimally structured
> MathML” is going to be related to that.
>
> The box that I got came automatically because of my brackets… it was not
> surprising at all and breaking it was the problem.
>
> This is about questioning the fact that most MathML should have these many
> boxes. If yes, then maybe they should be user-visible or maybe not. If they
> are visible, should they impact selection? I remember that we were proud to
> have OpenMath-term-respecting-selection in ActiveMath. And this example
> shows situations where this is undesirable, at least without breaking.
>
> Paul
>
> On 11 Jul 2021, at 20:23, Murray Sargent wrote:
>
> In OfficeMath
> <https://docs.microsoft.com/en-us/archive/blogs/murrays/officemath> if
> you select a math-object start delimiter, end delimiter, or argument
> separator, the whole object is automatically selected. Is this what you
> mean by auto boxing? For example, if you select the start delimiter of a
> delimters object (that used for parenthesized expressions and the like),
> the whole expression is automatically selected. The rationale is that the
> object won’t then be corrupted when copied or deleted. As Neil points out,
> this choice isn’t part of the MathML specification. But I think it’s good
> UI. It’s discussed a bit more in the post Math Selection
> <https://docs.microsoft.com/en-us/archive/blogs/murrays/math-selection>,
> a post you inspired me to write way back in 2007 😊
>
>
>
> Murray
>
>
>
> *From:* Neil Soiffer <soiffer@alum.mit.edu>
> *Sent:* Saturday, July 10, 2021 12:45 PM
> *To:* Paul Libbrecht <paul@hoplahup.net>
> *Cc:* www-math@w3.org
> *Subject:* [EXTERNAL] Re: auto-boxing
>
>
>
> Paul,
>
>
>
> I don't think I understand your issue:
>
>    - If the MathML generated by the new expression is wrong, that is
>    simply a bug.
>    - If you are saying that you don't like the way Word's editor or some
>    other editor works with selection/copy/paste, that's a UI issue and is
>    independent of MathML. Some editors might only allow certain edits, and
>    others might be more free. E.g., Mathematica's editor (full disclosure, I
>    wrote that one) is completely freeform wrt to linear notations. It seems
>    that Word's editor won't allow selection of part of the interior of parens
>    that extends beyond the parens; that's not a MathML limitation and might be
>    considered a feature by some (only allow syntactically meaningful
>    selections?)
>
> I think a simple way to break up the first expr is to select all of it,
> copy/paste it to the right and then delete the contents of parens as
> appropriate. Seems pretty easy and quick. Then add a '2' in front of the
> '5's (could have done this first and saved inserting a char). Of course,
> everyone's editing style differs.
>
>
>
> In the end, I'm not clear why this is a MathML issue but I am likely
> misunderstanding your issue.
>
>
>
>     Neil
>
>
>
>
>
> [image: Image removed by sender.]
> <https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.avg.com%2Femail-signature%3Futm_medium%3Demail%26utm_source%3Dlink%26utm_campaign%3Dsig-email%26utm_content%3Dwebmail&data=04%7C01%7Cmurrays%40exchange.microsoft.com%7Cedc7322bc41340a8781408d943db8535%7C72f988bf86f141af91ab2d7cd011db47%7C0%7C0%7C637615432701286759%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=jXWWHParpZsdAptbpduQTwYFAD96uGBMX%2BM3V07Fk8k%3D&reserved=0>
>
> Virus-free. www.avg.com
> <https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.avg.com%2Femail-signature%3Futm_medium%3Demail%26utm_source%3Dlink%26utm_campaign%3Dsig-email%26utm_content%3Dwebmail&data=04%7C01%7Cmurrays%40exchange.microsoft.com%7Cedc7322bc41340a8781408d943db8535%7C72f988bf86f141af91ab2d7cd011db47%7C0%7C0%7C637615432701296713%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=dMnj2vZUTd9MbbI9uGWKAXAR1IJ3FxsgcdQ%2B98h0L68%3D&reserved=0>
>
>
>
> On Fri, Jul 9, 2021 at 8:28 AM Paul Libbrecht <paul@hoplahup.net> wrote:
>
> Hello dear list,
>
> I had a funny demo yesterday in my lecture.
>
> I had input the following in Word: P(T<-t∨T>t)=0.05:
>
> However, a colleague that watched indicated I should rather split the
> probability measures in two which I started:
>
> Splitting the formula by using cut and paste was not possible anymore,
> because an automatic box (an mrow I assume) had appeared inside the first
> bracket. Only partial cut and paste was possible.
>
> So it was easier to re-input the whole or abandon the change; I chose the
> second ;-). Word has influenced my mathematical discourse!.
>
> A few discussions around intents seem to imply that these boxes are a
> natural requirement which is understandable from the perspective of a
> navigation through the formula or read-aloud or a selection-aware
> presentation_. I would like to agree with that but this implies that some
> boxes will bother the mathematical discourse.
>
> I believe that similar issues are met in other environments (in particular
> TeXmacs has very deep box-nesting.
>
>    - Should users expect “box manipulations” so that the boxes become
>    correct?
>    - Should they be told to care?
>    - Are there situations where boxes would overlap?
>
> paul
>
>