Not signed in (Sign In)

Not signed in

Want to take part in these discussions? Sign in if you have an account, or apply for one below

  • Sign in using OpenID

Discussion Tag Cloud

Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.

Welcome to nForum
If you want to take part in these discussions either sign in now (if you have an account), apply for one now (if you don't).
    • CommentRowNumber1.
    • CommentAuthorjayprich
    • CommentTimeDec 12th 2017

    the “Profunctors” section of Day convolution contains “⇸” as unicode within MathJax/iTex source which doesn’t render for me on Chrome / Windows 10

    For some reason this character “Rightwards Arrow with Vertical Stroke” (U+21F8), which has no iTex macro, is mangled and renders with three “unknown character” blocks.

    Diagonal Stroke (U+219B) is available as \nrightarrow (viz. profunctor#notation)

      or perhaps using the HTML entity would render consistently ⇸ ⇸ works for me (but strangely not ↛ ↛)

    • CommentRowNumber2.
    • CommentAuthorRodMcGuire
    • CommentTimeDec 12th 2017

    The problem is not just in MathJax but also in the MathML (Firefox) which is the official rendering.

    That is the result of a “fix” that wound up replacing all HTML character entities in the nLab with their unicode translation. See: nforum: internal-server-errors#Comment_658859

    We are still waiting for that fix to be fixed or rolled back.

    • CommentRowNumber3.
    • CommentAuthorUrs
    • CommentTimeDec 12th 2017
    • (edited Dec 12th 2017)

    We are still waiting for that fix to be fixed or rolled back.

    Adeel said he tried to roll it back, but ran into further bugs that prevented it.

    We need help. Richard Williamson is now kindly looking into it. But it may still take time, since fixing “Maruku” seems to be non-trivial. You seem to be knowledgeable. Maybe you could help?

    • CommentRowNumber4.
    • CommentAuthorRichard Williamson
    • CommentTimeDec 12th 2017
    • (edited Dec 12th 2017)

    At the moment, I am working on the non-loading of large pages issue. I will try to take a look at the unicode issue as well once time allows.

    I have an idea or two regarding the large pages issue, but I am blocked at the moment by not being able to build instiki locally (i.e. on my own machine), so that I cannot test (without which I am not going to push any changes to the live nLab). It is non-trivial due to the fact that instiki has quite old dependencies. I will contact Adeel so that we can find a way to test locally. It might be easiest to set up a docker container.

    • CommentRowNumber5.
    • CommentAuthorMike Shulman
    • CommentTimeDec 12th 2017

    With an appropriate interface to the page database, it ought to be possible to write a fairly simple script that would search for non-ascii unicode characters inside math-mode blocks and replace them by their corresponding HTML entities. I believe that that would fix this problem.

    • CommentRowNumber6.
    • CommentAuthorRichard Williamson
    • CommentTimeDec 13th 2017
    • (edited Dec 13th 2017)

    Now that a workaround is in place for non-loading of large pages, I will turn my attention to this, probably at the end of the week or so. It is dependent on my getting the nLab to run locally, but I am making progress on that, with Adeel’s help.

  1. I have begun digging into this. It would be trivial to reverse Adeel’s script, which is basically what Mike is suggesting. However, Adeel mentioned in the other thread that the problem is really with iTex not understanding unicode, and that he would prefer to fix that. I will dig a bit further to see if I can see how to do that.

    • CommentRowNumber8.
    • CommentAuthorUrs
    • CommentTimeDec 17th 2017
    • (edited Dec 17th 2017)

    Thanks, Richard!

    Adeel seemed to have said that he ran into some problem when reverting. But if you think it’s trivial, maybe you could just revert right away, to bring many of our pages back to readable, and then investigate the deeper fix at your leisure?

    • CommentRowNumber9.
    • CommentAuthorRichard Williamson
    • CommentTimeDec 17th 2017
    • (edited Dec 17th 2017)

    It is trivial in the sense that I can just run a script on the database to convert unicode to html. But such an operation on the entire database is not without risk, so I’d at least like to get Adeel’s approval.

    On the other hand, I believe I have now found the place in the itex2MML code which needs to be changed. I can run this by Adeel, and it may be just as quick, and certainly less risky, to try making that change, which is more or less a one liner. I’ll contact Adeel now, and keep people updated here thereafter.

    • CommentRowNumber10.
    • CommentAuthorUrs
    • CommentTimeDec 17th 2017

    Sure, thanks, sounds good.

    • CommentRowNumber11.
    • CommentAuthorUrs
    • CommentTimeDec 19th 2017
    • (edited Dec 19th 2017)

    Adeel and Richard have figured out/done something. It looks like all symbols display correctly now. But everybody please check.

    • CommentRowNumber12.
    • CommentAuthorTobyBartels
    • CommentTimeJan 3rd 2018

    Thank you, Adeel and Richard, for fixing this!