Added support for \not #140

kevinbarabash · 2014-10-01T03:36:21Z

Addresses issue #29. Requires a Huxley test which I will try to add tomorrow. Here's a screenshot:

TODO:

rebase
switch the order so the solidus comes first in the span and is shifted over with CSS
figure out how to make firefox render \not{abc} correctly

kevinbarabash · 2014-10-01T03:43:26Z

src/buildTree.js

+        var not = buildCommon.mathrm(
+            "\\not", group.mode, options.getColor(), ["mrel"]);
+
+        return makeSpan(["mord mrel"], [inner, not]);


This makes an assumption that group.value.body is a "rel". It really could be anything else. LaTeX and MathJax allow this. What they end up doing is placing the combining long solidus overlay glyph infront. Because this glyph already includes metrics that cause it to be rendered over the previous glyph, producing similar behaviour in these edge cases will require extra work.

kevinbarabash · 2015-06-26T15:34:30Z

We'll handle this using macros.

gagern · 2015-07-08T23:56:49Z

I don't see how this would be handled with macros, particularly the interaction with the atom type. Can you illustrate that idea? What would the \not macro look like?

kevinbarabash · 2015-07-09T00:10:30Z

I called the symbol the wrong thing, but I added a symbol for the long solidus. The not method in buildTree.js places one the long solidus after the inner symbol. It seems like a we could have a macro like \def\not#1{\longSolidus #1}. I just made up \longSolidus so maybe this isn't possible.

gagern · 2015-07-09T00:39:44Z

\def\not#1{\longSolidus #1} could almost be abbreviated to \let\not=\longSolidus except for the fact that \not{ab} is essentially \longSolidus{a}b which shouldn't make a difference. In LaTeX the definition of \not is \not=\mathchar"3236 and it is of type \mathrel. Looking at it this way, just your modifications to symbols.js should be sufficient to emulate LaTeX. Anything more fancy might be well intended, but actually contrary to how LaTeX does things, leading to incompatibilities.

kevinbarabash · 2015-07-09T00:43:06Z

In that case I'm going to re-open this.

sophiebits · 2015-07-15T14:20:26Z

Maybe-stupid question: why is this its own command type instead of just being one of the supported symbols?

kevinbarabash · 2015-07-15T14:31:21Z

@spicyj the \not glyph gets rendered after the rel that follows it in the source code so there needs to be some processing in buildTree.js (now buildHTML.js) to handle this. The macro I was suggesting in a previous comment should've been \def\not#1{#1 \longSolidus}. Sorry about the confusion.

sophiebits · 2015-07-15T14:41:19Z

Oh! I see.

gagern · 2015-07-15T21:49:01Z

But if \longSolidus overlaps the preceding character, then it behaves differently from the \not in LaTeX. In particular it will be aligned to the right edge of the group it affects, not the left edge. Perhaps we should add kernings so that the glyph is shifted to affect the following character? Or edit the font to achieve that?

kevinbarabash · 2015-07-15T22:17:48Z

@gagern because the \longSolidus occurs after the inner in the span it should already be right aligned. That glyph has funky metrics which shift it to the left by its width.

gagern · 2015-07-15T22:25:54Z

… it should already be right aligned.

Yes, and that's not consistent with TeX. In LaTeX, \not{abc} will essentially be {\not a}bc, while your {abc}\longSolidus would look like ab{\not c}. Sure, with multi-character groups we could split up the group and make this a\longSolidus bc, but the same problem holds for symbols of different widths, and for more complicated things in that group following the \not.

kevinbarabash · 2015-07-15T22:35:14Z

@gagern sorry, reading too quickly. I can flip the order and shift it forward.

kevinbarabash · 2015-09-26T00:04:17Z

src/buildMathML.js

@@ -454,6 +454,11 @@ groupTypes.phantom = function(group, options, prev) {
    return new mathMLTree.MathNode("mphantom", inner);
 };

+groupTypes.not = function(group, options) {
+    // TODO: it's an operator first and find the correct "not" version of that operator
+    return buildGroup(group.value.body, options);


In the case where there isn't a "not" version of whatever's in the body, I'll add the solidus in manually. This is what MathJax does:

<math xmlns="http://www.w3.org/1998/Math/MathML"> <mrow class="MJX-TeXAtom-REL"> <mpadded width="0"> <mtext>⧸</mtext> </mpadded> </mrow> <mrow class="MJX-TeXAtom-ORD"> <mi>a</mi> <mi>b</mi> <mi>c</mi> </mrow> </math>

kevinbarabash · 2015-09-26T00:39:37Z

The order is correct now. It's too bad firefox doesn't want to render it correctly.

xymostech · 2015-09-26T01:31:56Z

@kevinbarabash do you know why that is? Is it just failing to render the font correctly or something?

kevinbarabash · 2015-09-26T01:59:56Z

@xymostech I'm not sure. It renders everything correctly except for \not{abc}. For some reason the solidus is missing in this case.

sthoch · 2016-02-23T10:28:30Z

The solidus is there in my Firefox 44.0.2 (OS X 10.9.5).

kevinbarabash · 2016-02-23T22:20:03Z

@sthoch this pull request is to add it as a combining mark.

sthoch · 2016-02-29T10:14:30Z

Sorry, let me clarify: If I take your PR and use it to render \not{abc} with my Firefox, the solidus is there and not missing. I thought that was the problem.

kevinbarabash · 2016-02-29T16:15:59Z

@sthoch I didn't realize that you had tried this pull request out. I'll give it another try. It might have been a problem with the specific version of Firefox that I was using. What version are you using?

sthoch · 2016-02-29T16:33:27Z

Firefox 44.0.2 (OS X 10.9.5).

kevinbarabash · 2017-07-03T01:12:01Z

src/buildMathML.js

@@ -639,6 +639,9 @@ const buildExpression = function(expression, options) {
        const group = expression[i];
        groups.push(buildGroup(group, options));
    }
+
+    // TODO(kevinb): combine \\not with mrels and mords


MathJax combines \\not with certain relation operators and outputs the combined HTML entity inside an <mo> and for mords, it puts the HTML entity for not after the identifier inside an <mi>.

kevinbarabash · 2017-07-03T01:19:31Z

I've changed the approach significantly. Instead having a specific not group type and \\not function, buildExpression processes \\not groups and makes necessary changes to the following group.

I left a number of TODOs for subsequent PRs.

kevinbarabash · 2017-07-03T01:27:15Z

kevinbarabash · 2017-07-03T01:29:53Z

The handling of \\not with ords can definitely be improved, but it seems like a bit of an edge case as \\cancel is a better option for ords.

kevinbarabash · 2017-07-03T01:39:00Z

src/buildHTML.js

+    // Process \\not commands within the group.
+    // TODO(kevinb): Handle multiple \\not commands in a row.
+    // TODO(kevinb): Handle \\not{abc} correctly.  The \\not should appear over
+    // the 'a' instead of the 'c'.


I'll take care of these in a separate PR.

kevinbarabash · 2017-07-09T00:17:50Z

@gagern could you have another look at this?

to encourage other people with push access to review

kevinbarabash · 2017-08-12T23:58:12Z

Looks like I'll have to update the screenshots for this.

edemaine

LGTM. One typo in the comments. And need to fix the screenshots (and maybe run the --diff option to make sure changes are OK).

edemaine · 2017-08-20T18:17:01Z

src/buildHTML.js

+            // \u0338 is a combining glyph so we could reorder the children so
+            // that it comes after the other glyph.  This works correctly on
+            // most browsers except for Safari.  Instead we absolutely position
+            // the glyph and set it's right side to match that of the other


it's -> its

Can this same approach be used for #802 and #735?

It could almost be used for those but those accents are centered horizontally over their contents which complicates matters.

edemaine

I guess the screenshots fixed themselves? Anyway, thanks for revitalizing this old PR, @kevinbarabash! A very useful feature.

edemaine · 2017-08-23T16:35:02Z

For some reason I'm unable to actually merge this. Feel free to do so if you can...

kevinbarabash · 2017-08-23T17:10:29Z

For some reason I'm unable to actually merge this. Feel free to do so if you can...

@edemaine the branch needed to be updated. Do you see an "Update branch" in your UI?

edemaine · 2017-08-23T18:12:32Z

I did, but it wasn't clickable, which I hadn't seen before. Presumably a temporary glitch. Anyway, glad you fixed it.

kevinbarabash reviewed Oct 1, 2014
View reviewed changes

kevinbarabash force-pushed the not_command branch from 778093b to 0a28485 Compare October 16, 2014 01:56

gagern mentioned this pull request Jun 26, 2015

Restructure functions, environments, symbols #262

Closed

kevinbarabash closed this Jun 26, 2015

kevinbarabash reopened this Jul 9, 2015

kevinbarabash force-pushed the not_command branch from 0a28485 to 0f6fce9 Compare September 26, 2015 00:02

kevinbarabash reviewed Sep 26, 2015
View reviewed changes

kevinbarabash force-pushed the not_command branch from 0f6fce9 to 6467588 Compare September 26, 2015 00:38

kevinbarabash force-pushed the not_command branch from 6467588 to 36faf98 Compare January 15, 2017 05:48

kevinbarabash force-pushed the not_command branch 2 times, most recently from 9da8bb6 to bafa6e7 Compare July 3, 2017 01:09

kevinbarabash commented Jul 3, 2017

View reviewed changes

kevinbarabash force-pushed the not_command branch from bafa6e7 to 6da473d Compare July 3, 2017 01:13

Added support for \not

ba8a56a

kevinbarabash force-pushed the not_command branch from 6da473d to ba8a56a Compare July 3, 2017 01:24

kevinbarabash added GH Review: review-needed and removed GH Review: needs-revision labels Jul 3, 2017

kevinbarabash commented Jul 3, 2017

View reviewed changes

Merge branch 'master' into not_command

afb12e5

edemaine reviewed Aug 20, 2017

View reviewed changes

Merge branch 'master' into not_command

98181f3

kevinbarabash force-pushed the not_command branch from 6c5040b to 98181f3 Compare August 23, 2017 04:16

kevinbarabash removed their assignment Aug 23, 2017

fix grammar in comment

2dd0090

edemaine approved these changes Aug 23, 2017

View reviewed changes

Merge branch 'master' into not_command

6df5f3a

kevinbarabash added GH review: accepted and removed GH Review: review-needed labels Aug 23, 2017

kevinbarabash merged commit c6647e3 into KaTeX:master Aug 23, 2017

kevinbarabash deleted the not_command branch September 15, 2017 03:34

Added support for \not #140

Added support for \not #140

Conversation

kevinbarabash commented Oct 1, 2014 • edited

kevinbarabash Oct 1, 2014

Choose a reason for hiding this comment

kevinbarabash commented Jun 26, 2015

gagern commented Jul 8, 2015

kevinbarabash commented Jul 9, 2015

gagern commented Jul 9, 2015

kevinbarabash commented Jul 9, 2015

sophiebits commented Jul 15, 2015

kevinbarabash commented Jul 15, 2015

sophiebits commented Jul 15, 2015

gagern commented Jul 15, 2015

kevinbarabash commented Jul 15, 2015

gagern commented Jul 15, 2015

kevinbarabash commented Jul 15, 2015

kevinbarabash Sep 26, 2015

Choose a reason for hiding this comment

kevinbarabash commented Sep 26, 2015

xymostech commented Sep 26, 2015

kevinbarabash commented Sep 26, 2015

sthoch commented Feb 23, 2016

kevinbarabash commented Feb 23, 2016

sthoch commented Feb 29, 2016

kevinbarabash commented Feb 29, 2016

sthoch commented Feb 29, 2016

kevinbarabash Jul 3, 2017

Choose a reason for hiding this comment

kevinbarabash commented Jul 3, 2017

kevinbarabash commented Jul 3, 2017

kevinbarabash commented Jul 3, 2017

kevinbarabash Jul 3, 2017

Choose a reason for hiding this comment

kevinbarabash commented Jul 9, 2017

kevinbarabash commented Aug 12, 2017

edemaine left a comment

Choose a reason for hiding this comment

edemaine Aug 20, 2017

Choose a reason for hiding this comment

kevinbarabash Aug 23, 2017

Choose a reason for hiding this comment

edemaine left a comment

Choose a reason for hiding this comment

edemaine commented Aug 23, 2017

kevinbarabash commented Aug 23, 2017

edemaine commented Aug 23, 2017

kevinbarabash commented Oct 1, 2014 •

edited