WeBWorK and Runestone: preview answer

[Chrissy Safranski]

I think this is mostly for Alex, but maybe Brad too, and a heads up for others.

There is currently no way to preview an answer on WeBWorK problems in AC in Runestone when using the default AIM course for the WeBWorK.  This is a problem because when students get the answer wrong, they don't know whether it's their syntax, so that they have the problem correct on paper but didn't enter what they meant to enter, or their actual work.  Standalone WebWorK has the ability to preview.

The MathView editor I had enabled in the former daemon course showed them the preview (or maybe it showed them it after submission of their previous attempt, I'm not sure).  The MathQuill editor has issues with placement in-context, but the pretty typesetting avoids the need for a preview.  Right now, students don't have either one now that we've changed to using the default AIM WeBWorK course.  I've pointed them to https://webworkbuddy.me/ so that they could type their answer there with pretty typesetting and then copy-paste the supplied formatted answer into the answer box on Runestone, but it's obviously a work-around, and some (most?) students aren't using it yet.

Is it possible to have the down arrow button (the one that appears at the end of each answer box in PreTeXt html webwork problems that students click on to show them the feedback message) also show what they submitted in their previous attempt?  Or is it possible to have a "preview" button alongside the "submit", "randomize", and "reset" buttons?  

Or to set the default AIM course to use the MathView editor? I know, that changes things for everyone, which is why it's my 3rd option, but it would be the quickest fix. :-) 

Chrissy

[Alex Jordan]

When there is a syntax issue, it is reported in the feedback message.
That doesn't pop up by default, because it was considered to be too
intrusive. The button shows the feedback message, as in this
screenshot. I'm not seeing that a Preview button would have an
advantage over the Check button and viewing the feedback. In regular
WeBWorK, the purpose of the Preview button is to reveal syntax issues
without using up an attempt. But here I believe allowed attempts are
unlimited. Is that the case?

> --
> You received this message because you are subscribed to the Google Groups "PreTeXt development" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to pretext-dev...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/dc7a4c0d-8f85-47ef-ba3f-310fa65ca863n%40googlegroups.com.

[Chrissy Safranski]

Yes, attempts are unlimited.  But I think I didn't make clear what I meant.

By syntax issue I meant typing something other than what they really mean.  For example, I have students entering e^5x instead of e^(5x).  Their work on paper is correct and it's transferring that to the online format that is throwing them off.  If they could see that what they had actually submitted was interpreted as (e^5)x they would know what to fix instead of wondering if their work was incorrect.  Reducing that frustration is one of the main reasons I started using Edfinity for first-year students, though standalone WeBWorK has now caught up with MathView and MathQuill. 

But right now in Runestone they have no way of seeing how what they entered was received.

Chrissy

[Alex Jordan]

Oh I see. The solution to that is to wait until we can use MathQuill
properly. I will bring this up with Glenn, the WeBWorK-MathQuill guru.

> To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/442b05cb-822b-4b7e-84d5-f51a0f9b6e9bn%40googlegroups.com.

[Alex Jordan]

More quickly, perhaps I can add a rendered preview of the answer to
that feedback popup. When I work on the other thing, I will look into
this too.

[Alex Jordan]

Met with WeBWorK developers today. WeBWorK's MathQuill will most
likely evolve to:

(1) Move the palette to the bottom of the container (in a row
configuration) when the height of the container reaches some critical
low value. Some scenarios where this is likely to apply are if you are
using your phone in landscape orientation, or with many WW problems
rendered inside an iframe where the height is not tall enough for the
palette-as-column.

Additionally:
(2) Simply not show the palette at all if the container height is
extremely small, as it is with a few exercises used in AC that are
one-liners inside an iframe.

This would still leave an issue with extremely tall iframes, which is
the case with the new 8a preview exercises. The palette would still be
positioned at the 50% mark on the right side of the iframe, and so the
palette may be off the screen depending on where your scroll position
is in the parent window. Directly addressing this is very hard, and
might not be possible without some ugly/annoying side effects.

Anyway, in addition to all of the above, there will most likely be a
way to toggle the palette off for an entire course (in this case the
host course). So MathQuill would still do the live rendering as you
type, but not give you a palette. I would be inclined to toggle off
the palette for the default AIM host course. (But I am open to hearing
arguments against that.)

No word yet on a timeline for any of this.

[Chrissy Safranski]

Doing the live rendering without showing the palette would be my ideal solution. 

But meanwhile, a rendered view of what they submitted (maybe only needed if the answer was not correct?) would be good.  

Thanks for looking into this!

Chrissy

[Alex Jordan]

Glenn did some work on the WeBWorK MathQuill js for the next version
of WW. I tested it, and I've turned it on for the AIM anonymous host
course. So you will see it now in many places, including in the
safranski-acs course on Runestone Academy.

There are some things that are not perfect, but maybe they will get
better. Perhaps the most awkward one is that in the Assignments area
on Runestone, the toolbar is opening quite low within the iframe.

On Windows/Linux, when you have focus in an input field and you use
alt-t, it will toggle the toolbar. This setting will persist from one
input field to another. This at least gives a way to hide the toolbar
if it is covering something important (which is much less likely now).
On OSX, the hotkey will probably end up being command-option-t, but it
is not yet implemented.

> To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/4f478e96-3d5e-4f0d-ba7f-fdff7c582c83n%40googlegroups.com.

[Chrissy Safranski]

It's great!  You're right that the placement is still not ideal, because the menu bar is frequently low and makes you scroll more to hit submit, but so far I haven't seen it cover up anything that you need to see for that box.  If it covers up some of other parts of the question, it moves lower and out of the way when you click on that entry box.  

And alt-t is awesome.  I would love it if that was the default, and you had to hit alt-t to make the menu bar appear rather than the other way around.  But this is amazing.

My (and my students') thanks to you and Glenn!

Chrissy

[Alex Jordan]

Mysteriously, MQ appears for some exercises, but not all. For me here:
https://pretextbook.org/examples/webwork/sample-chapter/html/section-4.html
not on 1, but yes of 4,5,6.

There is no console error so I'm not sure what's up. But I wanted to
report the observation since it may be acting similarly inside a
Runestone Academy assignment.

> To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/71a0a40e-44f3-4cc4-89fc-836e8cf4d8ddn%40googlegroups.com.

[Alex Jordan]

> MQ appears for some exercises, but not all.

This turns out to be something that is fixed for 2.18, but too
complicated to be patched on the AIM 2.17 server. It will just have to
wait until the 2.18 upgrade.

The good news is that it probably does not affect the saffranski-acs
course on RA. It is only an issue for exercises that offer a hint or
solution, and my guess is that these are disabled for this build of
AC.

[Chrissy Safranski]

You're right.  It only affects a couple sections of my extra material, but we are past both of those now. :-)  Everywhere else, the webwork only appear in divisional or project exercises where hints (and solutions) are disabled, and so it shouldn't be a problem.  

One other thing - when MQ is being used, the answer box does not turn green or red after submission.  Students have to click the feedback arrow in order to see how they did.  While it's not a big deal when compared to the benefits of using MQ, students do prefer seeing their result without an extra click.  Is that something that's hard to change, or was it perhaps just an oversight?  See the difference in the two pictures below, the first on a question with MQ after hitting submit and then clicking the down arrow and the second on a question without MQ, immediately after hitting submit.

 

Chrissy

[Alex Jordan]

The background color issue is difficult because when the page loads, that input field is not actually there. Instead, at first the regular non-MathQuill input field is there. Then MQ hides the original input field and inserts some stuff that is perceived as the input field.

What this means is that styling on the original input field is not relevant for what is seen by the user. And I cannot directly style the stuff MQ inserts. So I resorted to using the CSS "+" selection operator to target the span that comes right after the original input field. I do not feel good about this; it may not be very robust. If something else gets inserted right after the original input field, it will break this. But maybe it won't be a problem.

Anyway please check things out now, ideally on a variety of exercises. I only did testing on the PTX sample chapter.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/4629fccc-1a82-4bf2-9441-f62a16d226f5n%40googlegroups.com.

[Chrissy Safranski]

I found the PreTeXt WeBWorK sample chapter online and tested it there (having to look for exercises without hints or solutions) and it looks great!  It also works on the website version you made of this special branch of AC, I tried a few different sections and exercises, but landed here for comparison.  Exercise 1.: Derivative of a power function   I tried Chrome and Firefox.

I'm not seeing any difference in my Runestone courses, however, or on the html version of standard Active Calculus. For comparison:   Exercise 1.: Derivative of a power function   Actually, maybe I can detect a flash of green before it goes away, in both Runestone and AC html?

One nitpicky thing is that after an incorrect answer, the box stays red when you start typing in it again.  It would be nicer if it didn't.  But non-MQ boxes seem to act that way, too.

I just went to ORCCA to see how it acts there, and that's very different behavior from WeBWorK in PreTeXT than I've seen before!  The boxes are outlined in green or red rather than being filled with green or red, and when you start typing in a previously colored box the outline changes to blue, and it's more like standalone WeBWorK with the summary of the prior attempt appearing.  If it doesn't take too much time to explain, why the different behavior there?  I can see it's a different server, but I mean philosophically why?

Chrissy

[Alex Jordan]

Did you clear cache? I was in your course in an assignment last night and things were working. Just not formatting the previous session answer yet.

>  the different behavior there

ORCCA HTML has not been rebuilt in 4 years. You are time traveling to olden times.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/569d1ced-cd5e-4e97-9759-3b8644ef7251n%40googlegroups.com.

[Chrissy Safranski]

*sheepish* No, I didn't.  After clearing cache, it works in Runestone in both Chrome and Firefox. Awesome!

But it still doesn't work for me on AC html in Chrome or Firefox. I can see it flash green but then reverts to no color. Maybe I need to clear more than just cache? Maybe I'll ask a colleague what they see. It doesn't affect me right now at all since we're using Runestone, but I figured I'd make the report.  

And oh.  I guess that's a very short explanation of "why"! :-)

[Alex Jordan]

For me on a Mac, using either Chrome or Firefox, I can visit either the AC book in your RS course or an assignment in your RS course, and entering correct or incorrect answers, the background is appropriately green or red with either a checkmark or alert sign. Can you give me a very specific place to look in case it is something to do with a specific exercise or type of exercise?

Separately, I am still working on the feature that loads a previously submitted answer into the input field. if you see that working, not working, or partially working, stand by for a post about it.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/b086fbd3-f907-4966-9763-f272b6f067d6n%40googlegroups.com.

[Chrissy Safranski]

> Separately, I am still working on the feature that loads a previously submitted answer into the input field. if you see that working, not working, or partially working, stand by for a post about it.

Woohoo!  Students have been wanting that since the beginning, because right now they have to do the entire problem correctly in a single attempt (not necessarily the first attempt) to get full credit.  They resented that because it was more work than Edfinity or standalone WeBWorK if they needed to come back to finish a problem later.  

I am seeing the new coloring work fine in my courses on Runestone.  I know you said you don't feel great about it, but I think it's awesome.  Where I see the new coloring not working is standard AC html, like for example Exercise 1.: Derivative of a power function It has the prior behavior of no color until you click the feedback arrow (although I can sometimes see a flash of green), and I just checked on my colleague's computer and also my phone to be sure it wasn't just my computer.  It doesn't affect my current courses at all and maybe doesn't need to be changed, but I wanted to report it.

You received this message because you are subscribed to a topic in the Google Groups "PreTeXt development" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pretext-dev/5ICwm5yW-WU/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pretext-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CA%2BR-jrd%3D-%3DQiW8-6jaWTZX6vrnUrJgGXr%3DYxDKLG7k3s0kMoSQ%40mail.gmail.com.

[Alex Jordan]

So that AC was built a while ago and is using version 0.13 of PreTeXt javascript. I am only enhancing version 0.2 of PreTeXt javascript. At least for now. If it reaches a point where everything is working well in your AC course (including the loading of previous answers) I may look into patching the old 0.13 in the same way.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CAH%3DJ7ghTnW1hQT9BWSwwyRXKTDggTeoi0PQK98xcgZG%3DboGk3Q%40mail.gmail.com.

[Sean Fitzpatrick]

I turned on MQ for one of the APEX books. It works! (For the one question I tried, at least.)
https://opentext.uleth.ca/apex-calculus/sec_chainrule.html

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CA%2BR-jrdur20Mc6GwN18E0CcmD4SLWX%2BXpO-QqBeA9f1d7Xbazw%40mail.gmail.com.

[Sean Fitzpatrick]

But I discovered one aspect that is not so great: if the answer is long, the MQ toolbar gets in the way. This might be a consequence of us putting exercises in columns.

See this exercise, for example, where the correct answer is
1/2*x*sqrt(x^2+1) +1/2*ln(x+sqrt(x^2+1))+C

Exercise 5

By the time I get to the second term, I'm underneath the toolbar.

The other issue (this is not a MQ problem; this is an "us putting exercises with long answers in narrow columns" problem) is that if you forget the +C, you get marked wrong, and the down arrow button you can click to see the message ("This is not the most general solution") is lost, unless you notice that there is now a horizontal scrollbar at the bottom of the exercise.

[Chrissy Safranski]

In other places, like my Runestone courses and in standard AC html, hitting alt-t (on Windows) turns off the tool bar.  

But it didn't work in your links.  Something about the 2 columns? 

[Sean Fitzpatrick]

Interesting. I didn't know about the alt-t shortcut.
It doesn't work for me, either.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/69a4696f-a63a-470c-883f-84d5bd75741cn%40googlegroups.com.

[Sean Fitzpatrick]

I found a weird workaround: clicking outside the question area turns off the toolbar. But if you click back on the box, the toolbar returns.

However: if you click on the text box, then open a new tab, and then close it, keyboard focus retrurns to the text box without activating the toolbar, but you still get nice rendering.

[Alex Jordan]

Please be advised: the alt-t feature to hide the menu is still being refined. What Sean is using for MathQuill on his WW server is probably at a different stage of development of an *unmerged pull request that is still under review and still getting new commits* compared with what I'm applying to the AIM server. And neither current state will be the final state of affairs. Already it has been decided that "alt-t" will be replaced with a right-click menu option, for example. Soon. Not now.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/d0a00ab6-d4c6-402e-9130-601825ada499n%40googlegroups.com.

[Alex Jordan]

The feature to turn off the MathQuill toolbar is now working well, thanks to Glenn Rice. It might need a clear cache. If you right click (including control-click with a Mac) it gives you a menu to disable the toolbar. The choice you make will persist. So if you want the toolbar to come back for aaa different exercise, you have to right-click again.

[D. Brian Walton]

As I'm thinking about whether this might also work for the formula-based fillin problems I'm working on, I looked at the MathQuill documentation and it looks like it is based on typing in LaTeX into a field and rendering it as you type.

How is WeBWorK converting the resulting LaTeX-like text back to string formulas that WeBWorK normally wants to parse? Or are you actually going the reverse way and feeding MathQuill LaTeX based on what the student is typing into a separate field?

Thanks,

- Brian

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CA%2BR-jrdXLkAUYdyx0BYqaT9wD4hL0pNhrcKfH649zThuXUTcPw%40mail.gmail.com.

[Alex Jordan]

My understanding is that the original MathQuill is dead. Projects that use MathQuill like WeBWorK and Desmos maintain their own fork. So if you haven't found this yet, the WeBWorK MathQuill is here:

https://github.com/openwebwork/mathquill

Glenn has done 99.99% of what is there.

I don't know the precise answer to your question, but I have inferred from conversations that MQ makes "blocks" internally from what the user types, and these blocks have enough structure to either produce LaTeX or the kind of string WW wants for an answer. And on the page there are actually two input fields, but one is hidden. MQ is putting the LaTeX in one, and the WW answer in the other.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CAH7VRoLVyOJkKF%2BTdTrpmcG%2B_dTb4aH4ybN4zenOdUKfnRdsFw%40mail.gmail.com.

[Chrissy Safranski]

Very cool, thank you!  It works well for me without clearing cache.

Chrissy

[Sean Fitzpatrick]

Alex, if my version is out of date, do I just need to do another checkout of that file from Glenn's repository?

As far as the conversation from two weeks ago, I think my version should have been up to date, because I was reporting my results only a few minutes after checking out the needed files from Glenn and doing the npm install.

But it's possible there is some other component I'm missing.

[Alex Jordan]

Yes, all that should be needed is to get those two files mqeditor.js and mediator.scss (and make sure mediator.css is deleted). 

Then run npm install from pg/htdocs. Restart apache. And it should be working.

--

You received this message because you are subscribed to the Google Groups "PreTeXt development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pretext-dev...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/46c7024b-4cd6-4cc0-ae0b-4da5f34e9d06n%40googlegroups.com.

[Sean Fitzpatrick]

Ok. That's what I did the first time.

I'm not seeing a right-click menu on any of the books I'm hosting, including after a refresh.

I could try running npm install a second time.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CA%2BR-jreeL7jzdXQFiZT%3DhgUXymUvY31iHYRzUh8x%3DLdH5q_FrQ%40mail.gmail.com.

[Alex Jordan]

You do need versions of those files that Glenn edited yesterday. Did you get fresh versions?

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CAH%2BNcPZvW2F4NPdsH-dRpr7_QKTw5U25ofLDTxbWTgOP5gTesA%40mail.gmail.com.

[Sean Fitzpatrick]

No, I'll update to yesterday's versions.

I'm also seeing some console errors at https://opentext.uleth.ca/apex-calculus/sec_chainrule.html

when I activate a WW problem.

Theses are are all CORS errors related to MQ fonts, due to no "access-control-allow-origin" header.

But there should be such a header, because it's needed for WW to work in the first place.

I'll double-check this while updating those files.

[Alex Jordan]

You probably have the header that gives access to the webwork2/htdocs folder, but not an additional one for pg/htdocs. Can you add that?

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/d637c2e4-9c59-457b-a3ad-a4b4314ce00en%40googlegroups.com.

[Alex Jordan]

I'll be at drop-in today if we need to troubleshoot.

[Sean Fitzpatrick]

OK, I'll be at drop-in as well.

Which is good, because I am stumped. Console errors when I activate a WW problem include things like:

Access to font at 'https://webwork-dev.uleth.ca/webwork2_files/node_modules/mathquill/dist/fonts/Symbola.woff2' from origin 'https://opentext.uleth.ca' has been blocked by CORS policy: No 'Access-Control-Allow-Origin' header is present on the requested resource. I've added pg/htdocs to be safe, but that doesn't seem to be where the error is coming from.

At least this is not interfering with being able to check answers in a WW problem.

But my webwork.apache2.4-config already was set up to allow that folder.

I have https://github.com/drgrice1/pg.git set as a remote in my webwork/pg folder.

I did git checkout mq-tweak/mqeditor-scrollable -- htdocs/js/apps/MathQuill/mqeditor.scss and git checkout mq-tweak/mqeditor-scrollable -- htdocs/js/apps/MathQuill/mqeditor.js,

then cd htdocs and npm install, followed by a restart of apache and a reboot of the server.

I have the MQ toolbar, but I do not have the ability to turn it off.

When I right-click on a WW blank, I get

Uncaught TypeError: Failed to execute 'getComputedStyle' on 'Window': parameter 1 is not of type 'Element'.
    at hi._getPlacement (dropdown.js:285:19)
    at hi._getPopperConfig (dropdown.js:314:23)
    at hi._createPopper (dropdown.js:255:31)
    at hi.show (dropdown.js:147:12)
    at HTMLSpanElement.<anonymous> (mqeditor.c24c4a0f.min.js:1:4912)
_getPlacement @ dropdown.js:285
_getPopperConfig @ dropdown.js:314
_createPopper @ dropdown.js:255
show @ dropdown.js:147
(anonymous) @ mqeditor.c24c4a0f.min.js:1

[Alex Jordan]

Add a header for the directory pg/htdocs.

Glenn set something up where a file is first sought in webwork2/htdocs, but if not found, it searches in pg/htdocs.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/62d4f2a4-34ef-4633-ac8e-05b5477dd2a0n%40googlegroups.com.

[Sean Fitzpatrick]

I had added this header, but I wasn't careful: in the apache config file, there are a bunch of alias settings done by Perl. the pg/htdocs folder has to be referenced as pg_files.

That takes care of the CORS errors, but not the uncaught type error when I try to right-click on the answer box.

To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CA%2BR-jrfwkyt4D9gtQCjXyR%3D0g7buS2%3Dp1xLwB1zce3unraiqsQ%40mail.gmail.com.

[Alex Jordan]

That sounds like it would work (and I infer that it did!) In my
situation I declared the header using a directory element instead of a
location element. And using the file path is what to do when it's a
directory instead of a location.

> uncaught type error when I try to right-click on the answer box.

That sounds like the error I got yesterday morning. And then Glenn
made some changes that addressed that. Does your version of
mqeditor.js match this one?
https://github.com/openwebwork/pg/blob/3935f46c8e0d19080ca4e14dc80f1bdda9005bec/htdocs/js/apps/MathQuill/mqeditor.js

In particular does it have the most recent version from yesterday? I
think Glenn did some force pushing, so you may need to really check
that git pull worked. If you use git log on Glenn's branch, do you see
3935f46c8e0d19080ca4e14dc80f1bdda9005bec as the most recent commit?

I'm sure you are doing it, but remember to npm install and restart
apache. Since it sounds like you have a slightly older version of
mqeditor.js somewhere.

On Fri, Mar 3, 2023 at 10:59 AM Sean Fitzpatrick

> To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/c97b284c-dd02-1d15-36bc-569f4fae5d82%40gmail.com.

[Sean Fitzpatrick]

I suspect that I didn't properly update Glenn's branch. I'll try again once I'm done ferrying my kids

You received this message because you are subscribed to a topic in the Google Groups "PreTeXt development" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pretext-dev/5ICwm5yW-WU/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pretext-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pretext-dev/CA%2BR-jrdvtPGtkpGbwngGQ_%3Dt1xNcyahkLtxQUOr5sD4TiFoteA%40mail.gmail.com.