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.
My suggestions in #91 haven’t been implemented yet, have they? (I don’t mean to nag, I only ask because Jon seemed to indicate he was already seeing them, but I don’t.)
Changing the math to computer modern font would also be a big improvement! I would have suggested that too if I had known it was what was causing the math to look out of sync with the rest of the text.
I can go along with text justification I think, although I don’t really have strong feelings either way. Any other opinions?
We’ve had more than one request to underline links, too. I think that would be
a { text-decoration: underline }
I was just rereading separable space, and in the end of the sentence with the name Pondiczery in it which ends with ’here’, I couldn’t see any indication that ’here’ links to another part of the nLab unless I hovered over it. The link goes to a specific result and that type of link used to be blue (and maybe underlined as well) without hovering over it, but from my computer at work (Chrome) I couldn’t see a thing.
Can we make that kind of link visible again, please, without having to hover over it?
Todd, is that the same issue of “external links” that was raised already in #91, 97, 98? Links to a specific result (i.e. anchor) on another nLab page use the same syntax as external links, so have the same styling.
Hi Mike,
Yes, I guess you’re right. Sorry for the repetition.
Just to say that Jake seems to have said that he has bowed out of doing further edits, which means that we will have to wait for Richard to implement the requested changes. Unless somebody else would lend a hand!
If someone will point out the files on GitHub, and the specific lines where the changes should be made, that will give interested parties something to aim at.
Sorry, been extremely busy the last few days, I will implement the changes suggested by Mike as soon as I get the chance (may not be today).
I have now made an attempt to implement all of the stylistic suggestions from #91 onwards, except that I have not changed the shade of green used for internal links, since David deliberately wished for this to be different from the old shade, and since it matches the logo I think.
Underlining of the links seems to be the change which helps the most with making the links clear.
Just let me know if further tweaks are needed (preferably with the actual concrete implementation details :-)).
Thanks, Richard!
There is some kind of bug causing certain things to be rendered in the same smaller font size as the context menu, I will fix that as soon as possible.
Thanks, Richard.
I can’t help but think that the line spacing is clearly too small now. It looks now like all pages have a rendering problem with lines tending to crash into each other.
But if the majority of people regularly editing here really wish it to be that way, so be it.
I think it would be partly due to the change in font. The serif font looks denser than the sans fonts, though if there’s actual clashing, then something odd has happened.
When a “y” in one line sits over an “l” in the next, the distance between them is less than that between horizontally adjacent letters (here).
It’s not so important for me either way, we can just as well keep it this way if you all like, just registering that I don’t follow this.
I think it is not what people wanted. Ideally the nLab looks as professionally typeset as a published book or article, rather than some kind of page proofs (as it was before) or something too cramped to read comfortably (as it is now).
Okay, great. Thanks.
By the way, did you figure out which lines on which GitHub pages to edit for these matters? To lend Richard a hand?
No, I didn’t dig around. I was hoping someone who actually knows CSS and/or the nLab source would let the rest of us know.
Richard was being busy and didn’t see your request.
Richard, when you find a spare minute, could you let David know where to turn to for lending a hand with the formatting?! Thanks!
I was even thinking of the other people around here who knew enough CSS to bake their own stylesheets.
The external links now seem to be OK. However, the bulleted lists now have a new problem.
I am not sure if this was intended, but some bulleted points are now smaller than the rest of the text, while others are of the size of the regular text (I see no reason that any bulleted text be smaller than the regular text, on the contrary I find it less readable and less beautiful, unless in footnote-intended passages where the writer explicitly requires footnote style). An example with both regular and smaller bulleted lines in unpredictably differently rendered bulleted lists is smart contract (zoranskoda).
I agree with Urs that now the lines in regular size text tend to be a bit too closely vertically now.
This is the bug Richard mentioned in #111.
Thanks Richard!
I think the problem mentioned in #114 occurs only in bulleted lists. At least, that’s the only place that I see it. But I don’t understand why it happens in some bulleted lists and not others (like the problem in #111).
Why should links have a different color from the rest of the text if they’re already underlined? Sorry if that has a trivial answer. (The text might look more calm if they were the same color.)
@Michael_Bachtold nlab has always had coloured links. I find coloured links easier to see and read. We can also have different coloured links hinting to the individual partaking in clicking where they are going.
I know this is standard in papers but can we have italics just slanted without the curly bits? Some pages, especially ones with lots of quotes like synthetic differential geometry are quite difficult to read when you have a block of curly text. Surely it must be possible to simply slant the text?
I have now fixed the bug mentioned in #111, and made line-height the same in lists and in paragraphs to address #114 and #122.
In addition, I have made line-height 1.3em instead of 1.2em everywhere, to try to address #115 and #120. I am not sure how much difference it has made, though.
Regarding #118, the style changes, including Jake’s original ones, are not committed to master yet, but the relevant file when they have been will be public/stylesheets/nlab.css.
In the meantime, I would strongly encourage anybody unhappy to try out changes for themselves, and provide me with concrete suggestions for what to implement. I am not involving my own preferences in this or even thinking actively about how I would improve it, I am just taking orders.
The active version of nlab.css can be found here. To try out changing line-height, say, just look at the HTML source of some page (most browsers have an option for this, e.g. by right clicking on the page), copy a relevant bit into the Sandbox, and add a style="line-height: 1.3em; font-size: 20px" to the <p> tag (or whatever).
Some browsers also have useful debugging functionality, e.g. in Chromium one can right click something and select ’Inspect this element’, and can see all styling that has been applied to it. Firefox has something similar.
I’m not quite sure what you are referring to in #125, Ali? Ordinary italicised text, or mathematics, or …? The italicised text I see on the page you linked to looks normal to me?
Compare the italicised text on the forum compared with on the nlab. Its working fine on the page but if it was like it is on the nforum I personally would find it much easier to read. I am aware that there are different fonts being used but I can’t be the only one who finds italicised fonts hard to read (especially when they are really curly on the ends).
Ah, then, yes, it is probably the font (a web font version of computer modern). If people wish to change to a different font, let me know (and as in #126, please experiment in the Sandbox to see the effect that the change would have).
Changing the page font to Trebuchet MS makes it much more readable. I also decreased the font size a bit from 19 to 17.
I however do not want the font to really change since people seem to be in favour of a modern font. I don’t know how to make the italicised version of computer modern just look slanted rather than a different font completely.
I think the fontscience way to say it is: I want oblique rather than italic.
I must say that I would be happy to hear others’ opinions about the font. Although it would be great in principle to use Computer Modern, especially since we can use it both for the body text and for mathematics, there is something about the web version that seems slightly off to me, though I’m not sure exactly what it is.
I think the font is good as-is, including italics. At first computer-modern looked odd to me, but then I realized I have no problem reading latexed pdfs on my computer screen, and then it started to look much more normal. Thanks for fixing the bugs, Richard!
I still think there is too much space above and below the page title at the top of the page; if I have some time I’ll try to figure out what could change in the CSS to fix that. Also probably the “Last revised” note at the bottom of the page could be in slightly smaller font. But in general I think it looks quite reasonable and we should give ourselves some time to get used to it and see whether we still have suggestions.
Re #130: According to W3Schools, CSS has font-style: oblique. But trying it out on a blockquote on the nLab, it seems to do the same thing as italics.
Doing some research, it looks like these web fonts are based on computer modern unicode, which has an oblique “Computer Modern Serif Slanted” family. nLab seems to be getting the font from dreampulse, which seems to include Serif Slanted, despite the README.
I think italics for a whole blockquote might be a bit heavy. What if you use oblique for blockquote, and italics for other usually-italics things?
I suspect that mixing italic with oblique on one page would look odd. But is there a reason that blockquotes need to be either italic or oblique? The quote environment in LaTeX is neither by default.
Okay, it looks like the problem with the page title is that in #91 I failed to specify that it was h1#pagename, not ordinary h1, that I wanted to set margin-top and margin-bottom to 10px. I didn’t realize that the same tag h1 was being used for the page name and also for internal headers.
Regarding the “Last Revised” text, it looks like div#revisedBy is already trying to have smaller text with font-size: .8em, but that’s being overridden by the font-size: 19px of the p tag inside it. I don’t know enough CSS to know the right way to fix this.
I have made the changes requested in #135 now. Using 10px for margin-bottom was too cramped, I have used 20px instead to try to obtain what I think was the intended effect. It might still be too cramped though, i.e. maybe there should be a little more space between the page title and the table of contents.
I have also made an attempt to improve the styling of the menu at the very bottom of the page.
One thing that strikes me is that having the page title be entirely lowercase now seems a bit strange to me. It worked with the old design, but now it seems to me that it would look better if the first letter were capitalised. Do others have any opinion on this? I can do the capitalisation through a code change if desired, we do not need to edit every page.
I find WIkipedia’s insistence of capitalising the first letter of the title of every page annoying. You get oddities like https://en.wikipedia.org/wiki/Pi (when it’s about pi, lowercase, not Pi, uppercase). And e on the nLab for instance would become ’E’.
Maybe there could be something added to the syntax to force lowercase titling in such exceptional cases, if that is the only concern?
I’m not seeing any \to or \rightarrow symbols rendering – it happens on my mac (safari) and my iphone (also safari, so maybe it’s a safari thing). Is this just me? Pic for reference: http://tinypic.com/r/1jaz9j/9
Could you try to see if they display using a different font, following the instructions in #126 say? The relevant part of the CSS file is probably math/mtext.
I have missing arrows on macOS and iOS too. On my Mac (10.14.4), setting my stylesheet in Safari to the below caused the the arrows to appear:
math, mtext {
font-family: 'STIXGeneral' !important;
}
I don’t know how to override the stylesheet on iOS, but I saved the page I was looking at as XHTML and served it on my local network (and so it was not using the nlab css) and on iPad the math looked fine including arrows. The iOS web inspector listed a bunch of fonts it was trying for the MathML node with the arrow in it, and one of them was STIXGeneral, so perhaps that is a safe fallback for both versions of Safari?
Just to note that firefox still seems to be working fine (for me anyway).
I have reverted to Stix for mathematics now. It seems that the support for mathematical symbols in the computer modern web font is limited: see here for a list. In particular, it lacks all kinds of arrows. It was not working on Firefox either, it is just that Firefox was using Stix fonts anyhow; I’m not sure exactly why/how, perhaps it caches them or something like that, or else you may have the fonts locally.
Now, itex does not actually make use of a very wide range of unicode symbols, so we could possibly ask at github to see if they can add the symbols we need, but I would guess it will not easily be possible, I think the web font is in turn relying on other earlier work. Also the repository has not been updated for four years.
I also happened to see this, which will affect us unless a change is made.
I would be tempted to try to find a font on all operating systems that works well for both mathematics and text. When I made some experiments last year, it seemed possible. On Linux, I think DejaVu together with DejaVu Math can be used. On Windows, there seem to be fonts available as well, though I do not remember them off the top of my head. I am not sure about Mac.
Alternatively, it seems like Latin Modern has a Math variant which should have widespread support, and apparently Latin Modern is very close to Computer Modern. We could try that for both text and mathematics if people prefer.
And according to this it is not actually difficult to create one’s own font these days (see right at the bottom). I have not tried it, so don’t know about that in practise.
That page gives a very useful list of appropriate fonts. DejaVu as I mentioned will I think be on almost all Linux systems; I think Cambrian is one that is on Windows systems by default. That leaves Mac. We can always serve Stix as a web font as backup if necessary.
Thanks for looking at this. I think I agree that it would be better to find one font that works for both math and text; it looks somewhat jarring to me right now to have the text in computer modern but the math in stix.
Working now, thanks Richard. I agree that now there is somewhat of a clash between the fonts, since CM is pretty spindly. But it does convey all the information!
I have noticed now that some apostrophes are being rendered backwards; for instance in the list beginning with “Parmenides”.
Hi Jon, thanks for raising, could you please link to the relevant page?
Richard, my bad! Somehow I thought I’d linked to it, and now I see that my message must have been very mysterious! :)
https://ncatlab.org/nlab/show/Science+of+Logic
Have’t been following the discussion lately, just want to say:
The way the typesetting comes out now I find excellent!
Thanks to everyone who helped make this happen!!
Re #149 and #151: this seems unfortunately to be a particularity of the computer modern web font.
I am getting the feel that we should look to change font (e.g. #147). This is a bit unfortunate now that people are just getting used to computer modern. For comparison, I have now made DejaVu Serif / DejaVu Math TeX Gyre the first choice with Cambria and Cambria Math as backup, but have not removed cached rendered pages except for comma category. If you’re on Linux or Windows and have those fonts (as you probably will), please take a look at comma category and let me know if it looks OK. To me on Linux it looks fine and is easier to read than the computer modern web font. I no longer have easy access to a Windows machine and cannot test there myself.
comma category looks good to me on my Mac with Chrome. Will test Firefox as well.
Has anything changed on the Mac, David? I wasn’t aware that a Mac would have either of the two fonts I added!
Apparently not :-)
No problem! I’m not sure if there is a good font we can try on a mac or on an android phone.
It looks pretty good to me. Somewhat reminiscent of the NYTimes.
comma category looks okay to me on Linux. (Other pages now seem to also be using this font? At least, I think they all look different.) We might want to reduce the default font size a smidgen with this font relative to CM, I’m not sure.
Thanks all for checking! Yes, any page which was not cached from before, or has been edited since, will have the new font.
Re #160: thanks very much for letting me know! How does it look? Can anyone else on a Mac confirm use of Cambria in Firefox?
My Firefox lists Cambria as being used. (NB I do not have MS Office : see #155) It also lists a Dejavu font.
I’m now seeing DejaVu Serif, and I have to say that I feel it is a lot more readable than before. Thank you for the continued updates and improvements, I really appreciate it!
Great! Thanks for the feedback everybody, very helpful. I have now removed the cache, so all pages should use either DejaVu Serif or Cambria now when these fonts are available. I have also removed use of Computer Modern completely, using just a local serif font and Stix instead when neither DejaVu Serif or Cambria are available (e.g. on a mobile). If anybody finds any other mathematics fonts that we should add to the list, just let me know.
Thanks very much Richard, I think the new fonts look very nice.
What do other people think about reducing the default font size a smidgen? With the new fonts at their current size I think the nLab looks best at about 90% zoom in Firefox.
I think you may be right, I will try that out when I get the chance unless somebody objects.
I also think the new font is more readable, especially on mobile, and on a beamer. I would also try to make it smaller, except on mobile, where it’s already pretty small.
Thanks to everyone for the work!
Good idea to reduce font size a little bit, I second that suggestion.
By the way, a font which is still in the spirit of similar to DejaVu Serif but which looks IMO quite a bit better is Noto Serif (https://fonts.google.com/specimen/Noto+Serif). Noto is more delicate but still has a similar feel.
Hah, the irony. I originally tried using Noto Sans, but then we went all Serif-y. I certainly don’t object to the font size suggestions (smaller on desktop, not on mobile).
Re #165 and following: I have now decreased font size from 19px to 18px. Let me know if it is better or if further changes are needed.
Re #168 and #169: The only issue with Noto Serif is that I assume there is no ’maths font’ version of it, whereas we have been aiming for fonts which do have a maths version, like Cambria and DejaVu. If you don’t think this is an issue, I can add it. It would only affect Mac I guess.
As of this change, the blockquote text is very small for me. Specifically, 14px, compared to regular text’s 18px. Was this intended? I don’t much like it. What about making it 18px or 16px?
It was intended to be noticeably smaller, yes. It is fine for my eyes on both computer and phone, but I can make it larger if people wish.
Belated news about the font changes (Re #155, #160, #162):
Cambria does seem to come from MS Office, on a Mac. I’ve come across two Macs not using it. They use the default serif. Also, my system has Cambria but not Cambria Math. It’s using STIX for math. Wikipedia makes it sound like this odd distribution of Cambria on Mac is typical.
Re the quote environment:
I just notice that
one can use the numbered list functionality also inside the quote environment (nice!)
but it comes out un-indented and un-shrunken (here is an example).
Not a big deal. But maybe it’s not hard to adjust this?
