Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
Richards asks that we report bugs in a single thread, and if discussion is needed start a new thread for each.
We have the issue of the link !-modality not rendering. Maybe because of the punctuation at the front. This was discussed here.
Then there’s the issue of consecutive links not always generating a space between them.
Thank you David! I will use this thread as the main place for keeping track of things from now. If it’s not here I may miss it :-).
Investigate use of max-width in mathematics CSS. Consider embedding DejaVu fonts. See here.
Re. #2, my guess is that the problem is that !
is used within Wiki-links by the !include
and !redirects
commands, so a page beginning with !
confuses the parser into thinking it’s dealing with something other than a page name.
I had a problem with itex negative spacing. Thread here.
There seemed to be an SVG problem here.
The “need to make an additional edit for uploaded file link to appear” issue is not new, but for tracking purposes, i’ve added it to the Technical TODO list.
Just to highlight that there is a problem with the rendering of edit announcements by user Ben in the thread on bicategories here
[edit: ah, seeing now that also my link above doesn’t render!]
All the comments are getting interpreted as plain text. Weird. Does that happen to me too?
My comment looks good to me.
The reason for #10 not originally rendering was that ’Text’ was selected as the format type. I have corrected this now. I suspect this was simply a case of Urs in a rush.
The comments in the other thread seem also to have been rendered in ’Text’ format, but I do not immediately see how that has happened. I will look some more.
Aha! It seems that the nForum ’Ben’ user is banned. This must cause comments to be rendered as ’Text’. I suspect the banned ’Ben’ is not the same as the nLab one, so I will try to fix this by deleting the old, banned Ben.
Fixed now by deleting the old, banned Ben and re-assigning the authorship of a few comments.
The page algebra over a monad seems impossible to edit: trying gives “500 internal server error”.
@PaoloPerrone, re. #16: I’m currently looking into this; thanks for reporting!
What happened in this case is that the page failed to be fully created. More precisely, for those familiar with the internal details, a row was created in the table ’pages’ but not in the table ’revisions’. This led also to the 500 error when carrying out a search.
I have now deleted the page, and re-created it: algebra over a monad. I did not have your attempted content available in an easily accessible form, Paulo (I have now tweaked the code so we would have the source available on a future occasion), so I must ask you to edit the page to put in the content. Fortunately, I believe you mentioned that you have a backup.
I will see if I can fix the root cause of this problem when I get a chance.
The page Critique of Pure Reason is impossible to edit giving the error:
An unexpected error occurred when rendering the page
On a different note:
As of this morning, I find that hyperlinks coded as
[arXiv:hep-th/0206192](https://arxiv.org/abs/hep-th/0206192)
do not render properly on the nLab (it works here on the nForum!)
In fact, on the nLab also the escaped form as above does not render properly: the “](” bit gets removed, and that then also explains the bad rendering of the un-escaped form.
I have put an example with the problem in the Sandbox.
Re #20: Fixed now. See here.
Re #19: It seems possible to edit now, I guess it was probably the same root cause as that which led to #20.
Thanks a million! Great.
@Richard: Okay, thank you. I’ll update the content now.
Is there a way to add captions to floating images?
As in the example here, consider a bunch of paragraphs, each flowing around a respective floating image.
Then how to put the source reference for the image (“graphics taken from XYZ”) right next to (below) the floating image, instead of it flowing with the other text?
Is this possible with the existing imagefromfile
-functionality?
If not, I am requesting such a feature hereby! :-)
Yes, it was possible within the existing functionality of imagefromfile
, and even documented at HowTo :-). However, the documentation did not really draw attention to it. Moreover, some care needs to be taken with the margin values to ensure that things look nice. I have now added some more specific guidance to the HowTo around this, and fixed the examples using imagefromfile
at AdS-QCD correspondence. There are some graphics at this page in the second half of it which still do not use imagefromfile
; I think this would be good to change, as imagefromfile
ensures that the caption is correctly handled, also semantically.
Previously the caption did not allow a link within it. I have now made this possible. I also happened to notice whilst browsing the page that the nLab did not have styling for headers of depth h5
; this is now fixed too in the same commit.
There are also some headers of depth h6 at AdS-QCD correspondence. These not only had no styling but were not picked up by the table of contents previously. I have now changed this, i.e. both added styling and ensured that they are picked up by the table of contents. I hope that this does not break anything on other pages; there may have been some reason why I disallowed h6 from the table of contents originally! I think possibly there were some cases of strange formatting involving h6, but hopefully these have been cleared up in the meantime. If anybody notices anything, just raise here :-).
Thanks!! Nice.
Sure, I will change the remaining image inclusions to imagefromfile
, too.
Hi, thank you for h6 again! I would bring up two things:
The compiler reformats when you jump over levels—I mean, when you type this
##
####
###
it becomes
##
###
##
I see the reason for this but it becomes inconvenient, when you want to reflect the structure of certain books like the Critique of Pure Reason.
Maybe the table of contents would become more compact, if the font-size (and spacing) of level h6 would be smaller.
Sorry again for my posting mess: As for troubles with h6. I don’t think that it is a feature that the “Theorem” shows up in the table of contents at HowTo#latex_syntax
Richard,
apparently now all Definition/Theorem/Proof-environments are being picked up by the table of contents.
A clear example is now the entry determinant (which I just edited and saved, causing the problem to appear).
I see now that this is also what Daniel is referring to in #30 above.
Yes, that must have been the reason I did not permit h6 originally. I will fix this as soon as I get chance one way or another. But I do not have the opportunity for the moment; perhaps this evening European time.
I have fixed this now, and re-rendered manually all pages edited since the changes announced in #26-#27 were made (hopefully I didn’t miss any).
Regarding point 1 in #29, thanks for raising Daniel, but for now I don’t think I’ll change this. I explicitly wished to disallow level jumping when I wrote the table of contents parser (for some pages there was a mess previously coming from the fact that it was not enforced), and I think probably it should still be enforced; in a book, one would never hop level for example. What you describe is, it seems to me, a corner case a) which would be solved if all intermediary sections from the book were named, and b) which does not seem disastrous to me even with the level hopping.
Regarding point 2 in #29, thanks again for raising. Yes, the table of contents does sprawl in some cases, it probably would be good to try to make it a little more compact one way or another. Not sure about making h6 smaller, but it is an idea. For now I’ll not prioritise this unless others feel that it should be acted on reasonably soon, but it’ll be listed here, and we can come back to it as time allows.
Hi Richard,
concerning again the imagefromfile
-functionality:
I would like to reduce the vertical distance between the image and its caption a little.
I have tried reducing the bottom margin, but apparently that is counted from below the caption.
Is it possible to adjust the caption positioning itself?
Or maybe it’s good for it to be hard-coded, but then it could maybe be a bit smaller.
The doctype declaration on top of the file https://ncatlab.org/nlab/show/geometry+of+physics+–+representations+and+associated+bundles says:
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1 plus MathML 2.0 plus SVG 1.1//EN" "http://www.w3.org/2002/04/xhtml-math-svg/xhtml-math-svg-flat.dtd" >
<html xmlns="http://www.w3.org/1999/xhtml">
XHTML 1.1 is an anachronism. We should probably replace it with a modern doctype:
<!doctype html>
<html>
XHTML 1.1 was published in May 2001 (!), which reminds me how extremely old Instiki is.
Re #36, thanks for raising! There was a problem with inconsistent section depths on this page, which led to the table of contents not being able to be parsed. I have attempted to correct it; the page displays now, but please check, Urs, if the sections are now at the correct depths (i.e. that things which are sections should indeed be sections and not sub-sections, etc).
Re #37-38: the date in the header is still OK for XHTML, Instiki is not that old :-). But yes, we can consider replacing the header. Recently I have been making progress on my long-term project of rendering Instiki itself obsolete when it comes to the nLab; much of the important code already does not belong to Instiki. Fairly soon, I hope that a completely new nForum will be ready for use with a new, fast and effiicient, Markdown parser written in C, as the first major visible sign of this latest work. This is a step towards moving the nLab/nForum to the cloud; I do not intend to move it as-is, it will be moved once the software is more amenable to it.
Re #35: I have tweaked this now. Let me know if it is OK now.
By the way, may I suggest to consider whether it is a good idea to use floated right images in general. E.g. the first graphic on the page AdS-QCD correspondence is not fully visible on a mobile phone, or at least my mobile phone. I don’t think one typically loses much by simply have an image be placed vertically somewhere below the parapraph which refers to it.
Thanks!
Re #35: On my phone everything looks perfect when I hold the phone horizontally.
(When I hold it vertically, the floating images are placed as ordinary non-floating images, the only problem now being that they get aligned to the right margin of the text (instead of lapping over it as the non’floating images do) and then partially flow out of the screen to the left if they are large.)
Ah, I never use my phone horizontally! If you are fine with how it is now, then all is good :-).
But wait, it seems this issue shouldn’t depend on whether the images float or not:
In vertical phone view (or generally when the window has not enought horizontal space to allow floating, I suppose) the floating images get placed vertically just as the non-floating images do, and it shouldn’t make a difference.
The problem (on my phone at least) arises only because, as soon as horizontal space is lacking, the floating images lap to the left, while the non-floating lap to the right, while the phone shows no whitespace to the left of the text margin, but loads of whitespace to the right of the text margin.
This need not be so, does it? It seems unrelated to the floating functionality.
Hi Richard,
there is a user here having trouble with the spam filter when editing Kakeya Conjecture. Can you see what’s causing the problem?
Re #43: I have replied on that thread now.
Re #42: I have fixed this now. The fix will only apply upon re-rendering a page, but I have done that for AdS-QCD correspondence, so this page should display fine on a mobile phone held vertically now. The problem was in fact related to floating, basically the fix disables floating (and adds a small bit of padding on top) when on a mobile phone, or when the window width is sufficiently small. All that said, I still personally feel that it typically looks better not to use floating, and just to position tables vertically; but of course it is up to you, I am no expert on typography!
The current revisions of constructive model structure on simplicial sets has a strange rendering bug. This bug only shows on the main page linked above, but strangely not on the revision page https://ncatlab.org/nlab/revision/constructive+model+structure+on+simplicial+sets/5 (notice that 5 is the number of the newest revision).
Specifically, in the section “Weak equivalences”, the second list item appears to have a spurious paragraph break in the middle of a list item. But this paragraph break does not appear on the revisions page.
There was some weird character in the source, a carriage return I guess. I fixed it by just making the source line continuous, as it should be. Probably such carriage returns should be tolerated and ignored/corrected, but one should definitely avoid carriage returns in this situation anyway and not rely on the server handling it, so I’ll not fix it for now. It is almost certainly an artefact of the interplay between the old and new renderers, and should be gone once we are moved off the old Instiki completely.
Only the current page uses the new renderer, not any of the revision pages (including the current one). Again, I do not plan to change this until the old Instiki is gone.
With regard to this discussion: There is XyJax which is an extension for MathJax that renders Xy-pic’s.
Formatting, such as italicised text, inside hyperlinks doesn’t seem to work. See *normal*
on this page, for instance.
Could the syntax for links to other pages be added, as suggested by this comment (that also has some other suggestions)?
Also, it would be quite nice to support Markdown for internal references instead of using the [[ page name | text ]]
style, which is entirely different to the style of internal links. I’m not sure if there is precedent for this elsewhere, but something like [text](§page name)
would feel far more consistent with the existing Markdown notation. Is this something anyone else has ever wanted?
Are there any plans to add support for \cite{…} and \bibitem{…}?
I am currently using them (e.g., at numerable open cover) in anticipation that such support will be added.
I thought that translating the \cite/\bibitem syntax into Maruku’s syntax is a matter of simple substitution, or is there more to it?
Is it possible to redirect to a specific section of a page? E.g. “higher-order abstract syntax” should be redirected to this section, but at the moment, it’s just redirected to the page in general.
No, that’s not possible. But to get a hyperlink to a specific section in a page, give it an anchor name by
## A section
{#ASection}
and then refer to it by
see [here](PageName#ASection)
Is it possible to rename nForum posts?
For instance, this post https://nforum.ncatlab.org/discussion/3179/measurable-locales/#Item_7 is title “Measurable locales”, which means that when somebody edits the article measurable locale, the software will create a new thread, ignoring the old one.
Yes, but only by the author of the first message in the thread:
Clicking “edit” on that first message (which it’s original author can do) gives an edit pane for the message itself, but also for the title of the thread.
Just a quick note that I have seen the above posts, thanks for raising. I have been tied up with other things and not been able to take a look yet.
The page cartesian bicategory appears to be uneditable: upon save I get “Invalid LaTeX Block:” with no further information.
This usually occurs when LaTeX is not closed properly, i.e. missing a dollar sign. The old Instiki permitted this (with consequent erroneous rendering of the page), the ’new’/current one does not. I will take a look when I get a chance.
Fixed now. In this case the problem was that someone had tried to include vertical space by an empty LaTeX block, which I removed and replaced by \linebreak
.
Re #54 and #55, I have fixed this in the database now. I do this whenever I see this problem occurring, merging the newly created thread with the old one. Another case where it occurs is where the original thread has a different category than ’Latest changes’.
Just post here if you see other examples of this which I overlook.
Re #48, this is a bug in Maruku, the markdown parser which Instiki uses. I will see if I can find a way around it when I get a chance.
Re #59 Thanks!!
Hi Richard,
does the imagefromfile
-environment admit specifying a URL to make the image be hyperlinked?
(I have tried some evident choices for tags, such as URL
and href
, without success)
Hi Richard,
on a related note, I just encountered a bug with the imagefromfile
-environment:
A minimal example is now in the Sandbox:
The source code of the Sandbox page right now is this:
\begin{center}
\begin{imagefromfile}
"file_name": "HardWallModelPredictions.jpg",
"width": 440,
"caption": "From [Erlich 09, section 1.2](#Erlich09)"
\end{imagefromfile}
\end{center}
while the output of the Sandbox page is this error message:
XML Parsing Error: mismatched tag. Expected: </div>.
Location: https://ncatlab.org/nlab/show/Sandbox
Line Number 314, Column 35:
<div style="text-align: center"></p>
----------------------------------^
The error goes away when removing either (or both, of course) of
the center
-environment,
the image caption
.
Hi Richard,
I have now a “working” example of the bug with modifying the redirects:
I have just started to add a new subsection to diffeological space. In the source code it’s visible as
### Relation to topological spaces
Then I added one redirect to the page
[[!redirects D-topology]]
After that, the added material no longer appears in the output of the page.
Experience shows that if next I made any other edit to the page the output will appear. But I’ll leave it as is for the moment in order to demonstrate the bug.
Thank you! I’ll take a look at these things as soon as I get a chance.
Recently, when visiting the nLab from Google, extra text has been added to the URL, and the corresponding text highlighted on the nLab, which seems to slow everything down, and also looks ugly. An example of what I mean:
https://ncatlab.org/nlab/show/over+category#:~:text=Definition,-The%20slice%20category&text=The%20slice%20category%20is%20a,g%3AX%E2%86%92X%E2%80%B2
I’m not sure who’s responsible (Google/nLab). Is anyone else experiencing this?
Hi, I’ve not noticed this myself. There have not been any changes to the nLab software which should lead to this. I also could not reproduce it in the browsers I use. Let me know if it looks like there is something we can do on the nLab side about it.
It’s been happening sporadically and I could definitely believe it was Google A/B testing something on me that wasn’t working as intended.
Footnotes increase line height, e.g. see Giry monad. Reason is that line height in normal text is not 1em but 1.3em. Thus the <sup>
-tag shifts this extra space to the top creating too much space. Remedy: sup { line-height: 1em; }
.
Thanks for notifying! I made the change you suggested; it looks better at Giry monad now, but I guess not perfect yet. I ran out of time to experiment further; will do so when I get the chance.
I find spacing at the beginning of lists in lists not pleasing, e.g. lax functor. I would suggest li li:first-child { margin-top:.25em; }
.
I have to correct my suggestion:
li ul, li ol { margin-top:.25em; }
li li: { margin-bottom:.25em; }
The is another point about first order lists. There is too much spacing when a list follows a paragraph, e.g.pseudofunctor. My suggestion
p { margin-bottom:.5em; }
ul, ol { margin-top:.5em; }
Speaking of margins, I would also suggest
h6 {
margin-top:2em;
margin-bottom:1em;
}
The only two examples using h6
I am aware of are Science of Logic and Critique of Pure Reason.
Sorry, I correct all my suggestions from #73 again to:
p { margin-bottom:.5em; }
li p { margin-top:.5em; }
and deleting
li p { margin-bottom:.5em; }
The problem with the current margin comes from the fact that the text in lists is put in p
s (in contrast to how tables of content are generated). But p
s have top/bottom margins =0
.
I can’t edit because of “invalid LaTeX block” in space and quantity. I saved the edit I was trying to make as follows. Also it would be really good if there’s a preview button in the editors.
“‘ which underlies much of mathematics is at its heart controlled by the following elementary category theoretic reasoning:
Let be some category whose objects we want to think of as certain simple spaces on which we want to model more general kinds of spaces. For instance , the simplicial category, or CartSp, the category of and smooth maps between them.
An ordinary manifold, for instance, is a space required to be locally isomorphic to an object in . But more generally, a space modeled on need only be probeable by objects of , giving a rule with which, to each test object , we assign the set of probing maps from to , such that this assignment is well-behaved with respect to morphisms in . Such an assignment is nothing but a presheaf on , i.e. a contravariant functor
Therefore general spaces modeled on are nothing but presheaves on :
Of course this is an extremely general notion of spaces modeled on .
For example, any smooth manifold is a presheaf on CartSp by , where we consider CartSpF_M\mathbb{R}^n\mathbb{R}^nM$.
In particular, every object in is a space modeled on , by the Yoneda embedding , whereby every object in is embedded as . That is, any object in is nothing but a consistent way to be probed by all the objects in .
Now take a space modeled on , and consider the set of quantities on with values in . It should be
This defines a covariant functor . More generally, we can consider the S-valued quantities on to be a copresheaf on , namely a covariant functor
One can think of as a generalized quantity which may be co-probed by objects of .
In this vein, one can say, generally, that copresheaves on are generalized quantities modeled on , and we write
Given any such generalized quantity , we can ask which generalized space it behaves like the algebra of functions on. This generalized space should be called and can be defined as a presheaf by the assignment
In total this yields an adjoint pair of contravariant functors between generalized spaces and generalized quantities:
(That this is an adjunction can be understood as a special case of abstract Stone duality induced by a dualizing object.)
Lawvere refers to this adjoint pair as Isbell conjugation.
“‘
Hi all. I noticed there might be a minor rendering issue, for example on this page, where both \mathcal{I}
and \mathcal{J}
are displayed as the latter. The issue persists across multiple devices with Chrome/Chromium, but not on an old version of Firefox (52.6.0). I hope this is helpful.
Thanks for pointing this out. That’s a weird bug. I had not noticed this before.
The word “Tangora” triggers the spam filter. In fact its the string “Tango” in there that triggers it.
I was just trying to update publication data for the textbook by Mosher & Tangora at cohomology operation, but all edits to the page are blocked as long as Tangora’s name appears in the page.
Also all edits to the page Martin Tangora are being blocked.
Re #80: I have tweaked the spam filter now.
Re #78: This has cropped up before, see for example this discussion: it is a MathJax issue which we might be able to workaround, but tweaking itex2MML to do so is not a priority at the moment. Dmitri Pavlov was considering contributing some changes to itex2MML a little time ago; if he does so, we can ask him to try out the workaround.
Thanks, Richard.
When I access Kant I get a message
Redirected from “Kant”.
Note: Immanuel Kant and Immanuel Kant both redirect for “Kant”.
There was a duplicate redirect in the source. I have removed it now.
Hi Richard,
I’d like to use your xymatrix-functionality on the Lab more, but one issue that gets in my way is that the column/row-spacing is too large, to my mind:
For instance, a short exact sequence, typeset with single spacing for the outermost arrows and double spacing for the two inner morphisms, easily stretches over the full page width (I am looking at a concrete example in the Sandbox).
Might it be easy to make the spacing option (like @C=8pt
) work on the nLab?
Or if you don’t have time for that, maybe your script could just reduce the default spacing a little?
Or if I am missing something and I should write different code on my end, please let me know. Thanks!
Hi Urs, I’m reading a bit quickly, but I think adjustment of the spacing is already implemented, and documented in the HowTo. Let me know if it doesn’t meet your needs.
Ah, my bad, I had missed that. Thanks! That’s all I need. Thanks.
One more thought: Given that you script already reads in commands for outside the xymatrix:
How about the possibility to call\usepackage
in the preamble for an xymatrix rendering? :-)
Code-wise this would be possible; the TikZ implentation allows \usetikzlibrary{...}
, and it would be trivial to tweak the code to allow \usepackage
as well. I had preferred to not allow this, however, partly on grounds of security (who knows what some random LaTeX package might do) and partly on grounds of keeping things as generic as possible. My idea was that if a particular package is desirable, I can either add a specific option to be able to use that package, or always have it available. Do you have specific packages in mind?
Sure, sounds good. The diagram that I am currently trying to port to the nLab needs
\usepackage{color}
OK, I think I can safely make that always available :-). I have now added it; let me know if it works (you do not need to call \usepackage{color}
yourself).
I checked myself now that it works at least for some purposes, e.g. one can add @[red]
to an arrow to colour it red. See the example in the current Sandbox. Let me know if it doesn’t work in the case you have in mind.
Thanks, Richard!!
I may have more questions/requests as I start getting serious about writing xymatrices on the Lab. Hope you don’t mind.
Here is another point I run into:
Might we have xymatrices equipped with equation numbers, like we have for Instiki-coded equations?
Might we have xymatrices inside a bullet/numbered item list such that they get indented together with the rest of the text in the item, and such that the item numbering doesn’t get disrupted?
I guess in short this is asking if we can have xymatrices behave inside the ambient text just as Instiki-equations do?
Good that you raise these things, the goal should definitely be that the functionality meets whatever needs you and others have! These are good points, I will try to implement these when I get a chance.
Regarding the second point, I generally find lists in Markdown in this kind of situation to be extremely awkward. I.e. whenever there is some kind of line-break, even if it is possible to get the list to typeset correctly in Markdown, it is typically a considerable effort in my experience. I could try to workaround/extend the Markdown syntax, but I think it might be better to introduce a LaTeX-type list syntax (which would also make it easier for people to paste in LaTeX).
Sounds good.
By the way, these days, I am doing a lot of code where xymatrices are included inside an ambient LaTeX-formula. As in
Maps\left( \raisebox{20pt}{\xymatrix{ A \ar[d]\ar[r] & B \ar[d] \\ C \ar[r] & D }}, X \right)
(just a simplistic example, to illustrate what I mean).
Do you see a chance that such code could work on the Lab?
Another thought:
Could we have our LaTeX-style Definition/Theorem/Proof-environment implement the [...]
-option, as in
\begin{theorem}[Theorema Egregium \cite{Gauss}]
Let... then...
\end{theorem}
?
Re #97: Interesting! This would be a bit of a challenge; probably the way to go would be to convert the XY Matrix diagram directly to some combination of SVG and MathML in this case, rather than using the mechanism we are currently using (actually running LaTeX and then converting to SVG). I am interested in the long-term in doing something to replace or at least significantly ’refactor’ (change the code but not (much of) the functionality of) itex2MML, so this might be a first step in such a project.
Re #98: Yes, that should be possible. Actually \cite
is something I have had in mind for years and have worked a bit on, but never completed yet. I will try to implement this when I have a chance.
I support the implementation of \cite[…]{…} and \bibitem[…]{…}, this would be incredibly useful for the nLab.