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.
There has been a discussion about good practice with theorem numbering, particularly in light of a certain bug, under regular elements in a Heyting algebra. I’m moving it here.
I think that’s a good suggestion, Urs (and I’ve begun doing that, but not consistently).
I’d like to propose another suggestion: when including redirects, make a point of including plural forms. That way a link can be made directly from the plural, without extra fuss.
If we do use numbered theorem environements, I think we should always also include an anchor name:
+-- {: .num_remark #MyRemark}
IIRC there is an instiki bug that rears its head when some, but not all, of the numbered theorems on a given page have anchor names assigned.
*shrug* Don’t ask me.
Bug or no bug, there’s not much point to numbering the theorems/remarks/etc if you can’t refer to them later by number. And the only safe way to do that is with the anchors.
Bug or no bug, there’s not much point to numbering the theorems/remarks/etc if you can’t refer to them later by number. And the only safe way to do that is with the anchors.
That’s what we currently do anyway, being forced by the bug: if I want to refer to something by number, I need to anchor everything in between in order for the numbers to come out right.
But what I am hesitant (no: unwilling) to do is to preemptively think up anchor names for everything in sight even before I refer to a single one. It would be good, I agree, but there is some limit being reached here.
But what I am hesitant (no: unwilling) to do is to preemptively think up anchor names for everything in sight even before I refer to a single one.
Me too. But in that case, I just leave them unnumbered until I decide that I wish to refer to them.
I agree, the bug should be fixed; and when it is, then you can just make everything numbered to start with, without thinking of names. However, so long as you cannot think of a name, then numbering it gives you no additional benefit (and leaving it unnumbered works around the bug, while we have it).
The so-called bug is a “feature”. To fix it would involve quite a change in how theorems are numbered. The benefit that would be lost is the ability to create new numbered environments without changing the core code. When this was first raised (incidentally, this’ll be a test of the redirections for the new location - if that link works, they’re fine) I mentioned it to Jacques and he explained that the current method had this flexibility. He didn’t rule out changing it, but I think it would need raising again to get him to look at it.
So to have things numbered, we need anchors. But if an anchor isn’t going to be used, its name doesn’t have to be particularly meaningful. They do have to be unique (otherwise the numbering system seems to get messed up - just experimented on the Sandbox), but dummythma and so forth should be usable without stretching thoughts too far.
Might it be possible to have dummy anchors added automatically somehow, the same way that dummy anchors are created for all headers when there is a TOC?
That would be a reasonable compromise. I’ve suggested it to Jacques. We’ll see what he says.
Interesting to see the discussions on that forum. I had no idea that this existed.
I may have this wrong - so don’t quote me - but Jacques originally set that up as an experiment in having a forum (and for fairly obvious reasons he wanted to have on one the same language as instiki). He mentioned it to me, and I started using it as a way to talk to him about instiki/nlab, and it’s sort of morphed into a general instiki-and-Jacques’-other-projects forum, but I don’t know if it’s ever been announced as such.
Jacques response:
Ah.
That is indeed a bug. Thanks.
So looks hopeful.
This has now been fixed. See the Sandbox for a working example.
Great! Thanks!
Incidentally, this shouldn’t break any existing theorem numbers. Please say if someone spots some that are now out of sync with their references.
Awesome! I’ve deleted the now-unnecessary FAQ “My reference cites the wrong theorem”.
(And I learned something, too; I didn’t know you could link with href=”#anchor” to the id=”…” of a <div> as opposed to the name=”…” of an <a>.)
@Mike: You can also link to the id attribute of an a element too. In fact, you can link to the id attribute of any element, and this is now preferred. (The name attribute of a elements are still supported for backwards compatibility, but it’s deprecated.)
Shows how long it’s been since I’ve written HTML. (-:
Silly question: is the only way to link to an #anchor in a different nlab page to write a [text](URL) link?
Cheers,
–Jesse
Numbered equations can be done with \eqref, but that's the only way that I know of for numbered theorems.
re #22:
yes, I think so. But at least a relative URL is possible and should be preferred. So
[linkname](pagename#anchor)
Oh, that does seem to work… the odd thing is, leaving out the pagename means implicitly the #anchor on this page, which is a little bit surprising… oh, that probably is the html default behaviour as well. Very good.
Thanks!
1 to 25 of 25