upward shift for some users under IE

[Ron Whitney]

Hi.  We have recently been seeing inconsistent behavior in vertical alignment for IE users.  Some of us get

while others get

The inconsistency is across users' browsers, not on different pages or across different math expressions.  A browser showing the bad upward shift in one place, shows it elsewhere.

We're using MathJax 2.6.1, but notice the same behavior when we use 2.7 from the CDN.  All of us are using IE 11 and looking at the same site.

And similar differences are seen when we view the MathJax home page as well as the samples under your features page.

So I gather some IE setting is causing this, but I haven't found anything in the literature about it.  Any suggestions would be appreciated.  Thanks.

-Ron

[Peter Krautzberger]

Hi Ron,

I'm afraid there's not much we can say from a screenshot; a minimal live example (e.g. using codepen) is almost always necessary. 

See the User Group Guidelines for more tips.

Regards,
Peter.

-Ron

--
You received this message because you are subscribed to the Google Groups "MathJax Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mathjax-users+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ron Whitney]

Hi, Peter.  Of course I'm willing to supply whatever you need.  In this case, you have source code already in the form of the samples accessible from the MathJax home page.  I thought the relevant information would lie in a precise understanding of how the layout is vertically ill-positioned for some of our users and that's why I went to a screenshot.  I'm certainly willing to supply a shot of, say, the 2\pi i of the MathJax sample raised up toward the fraction bar, but it will also look much like the D and 2 of the sample I sent.  The display of the quadratic formula is also mis-aligned vertically in fraction and radical for the same users, vaguely so to you now, but a screenshot would show what some of our users are seeing.  The underlying code we're viewing is fine and users are looking at the same pages from the same servers (yours).  Of course, you face the difficulty of reproducing what we're seeing.

Looking at your examples using codepen and MathJax, I wonder how to reproduce IE output, and if I can do that, how to ensure the misalignments we see for some users (but not for others).  And, I suppose if I can do that, I'll have the solution.  I do see that cross-browser testing is possible.  If you consider the methods to do so clear, just reassert that and I'll do my homework (with apologies).  Otherwise, how can I supply what you need to understand the problem?

We've been using MathJax for quite a while, but I cannot say when this might have started happening.  I am guessing (perhaps hopefully, as well) that our problems are due to something well known in IE configuration.  I'm willing to do further searches, but am asking here because I did not find a solution previously.  We only see the problems in IE, not in FF or Chrome.  Users are viewing with IE11 under IE9 compatibility.  Half of our sample of users have the problem, half don't, but I've been unable to determine what difference in IE configuration is causing difference in display.

I don't rule out an obvious solution, but

same browsers, same version, same compatibility;
same servers, same pages;
different vertical placement for 3 out of 6 users.

What remains to check in IE?



I appreciate your help or redirection and apologize for words which appear as noise.

-Ron

To unsubscribe from this group and stop receiving emails from it, send an email to mathjax-user...@googlegroups.com.

[Peter Krautzberger]

Hi Ron,

My apologies. I missed the part where you mention our samples on mathjax.org.

We'll take a look.

Peter.

To unsubscribe from this group and stop receiving emails from it, send an email to mathjax-users+unsubscribe@googlegroups.com.

[Davide Cervone]

What output renderer are you using? Is it the CommonHTML output?

One thing to check is whether your users who are seeing the problem have the MathJax fonts installed on their systems. These were updated in version 2.6, and the old versions of the fonts will produce results like the ones you are seeing when used in CommonHTML output.

There is a test page that you can use to determine if the old fonts are installed. Go to

http://codepen.io/mathjax/full/avZRzM

and see if the message indicates that you have the old fonts. If so, either replace them with the new versions, or just remove them from your system all together.

Davide

[Ron Whitney]

We are indeed using CommonHTML output and installation of the newer fonts corrected the problem of one of our users.  The others are out this week, but I expect they have the same issue.  Thanks much for your help.

-Ron