Multiple references to a footnote cause duplicated footnotes

[GoogleCodeExporter]

So I was writing up a little something, and I created a footnote [^foo]
with the respective entry at the bottom of the page ([^foo]: full ref) and
then I happened to later cite the same paper and so I chucked in another
[^foo] in there.
Later I happened to notice that this worked, but my reference appears
*twice*! That is, I had expected something like Wikipedia's references
where multiple references link to the same anchor and then back - eg.
https://secure.wikimedia.org/wikipedia/en/wiki/Fujiwara_no_teika#cite_note-famin
e-66

Original issue reported on code.google.com by gwe...@gmail.com on 21 Feb 2009 at 3:20

Attachments:

• xwd-123516683030627.png

[GoogleCodeExporter]

Original comment by fiddloso...@gmail.com on 5 Dec 2009 at 8:00

• Changed state: WontFix

[GoogleCodeExporter]

Why is this won't fix? It wastes space and misleads readers for no benefit at 
all,
that I can see, and coalescing footnotes is perfectly possible in even HTML.

Original comment by gwe...@gmail.com on 5 Dec 2009 at 1:53

[GoogleCodeExporter]

I don't see a good way to fix it without major changes to pandoc's architecture.
The problem is that, after the reader has finished, the AST can't distinguish 
between
two cases:  (a) the case where one footnote reference was used twice, as above, 
and
(b) the case where two references were used, but the footnote text happens to be
identical.  Presumably the notes shouldn't be coalesced in the latter case 
(even in
HTML).

Original comment by fiddloso...@gmail.com on 5 Dec 2009 at 4:59

[GoogleCodeExporter]

> (b) the case where two references were used, but the footnote text happens to 
be
identical.  Presumably the notes shouldn't be coalesced in the latter case 
(even in
HTML)

Why not? If they are literally the same exact note with the same text, then 
this is
just good compilation - I would expect my Haskell code which says 'let x = 1 in 
let y
= 1...' to need space only for 1 1, not 2 1s, for example.

Original comment by gwe...@gmail.com on 6 Dec 2009 at 12:13

[GoogleCodeExporter]

Here's a stronger consideration. In pandoc's html output, the note itself 
contains a
backlink to the reference. This won't be possible if multiple note references 
link to
the same note.

Original comment by fiddloso...@gmail.com on 6 Dec 2009 at 1:09

[GoogleCodeExporter]

As I said before, Wikipedia does this just fine but having multiple backlinks 
in the
note. This is even a feature, not a workaround - with heavily referenced 
articles &
sources, it can be valuable to see where else a reference is being made.

(Is this an objection that 'it's a good idea, this no-duplicate-footnotes idea 
of
yours, but this is a second reason that actually doing this good idea would be
difficult', or this some other kind of objection?)

Original comment by gwe...@gmail.com on 6 Dec 2009 at 1:41

[GoogleCodeExporter]

I'm not yet convinced it's a good idea.

When I click on a note reference to read the note, I don't want to have several
back-links to choose from.  I just want to get back to where I was, and how do 
I know
which back link does that?

Original comment by fiddloso...@gmail.com on 6 Dec 2009 at 3:19

[GoogleCodeExporter]

> When I click on a note reference to read the note, I don't want to have 
several
back-links to choose from.  I just want to get back to where I was, and how do 
I know
which back link does that?

Well, there's the Back button, unsurprisingly. However, Wikipedia, I am told (I 
have
too many customizations etc. to be sure), will highlight via CSS the exact 
backlink
you came from; you can probably try it out yourself at a heavily referenced 
article
like
https://secure.wikimedia.org/wikipedia/en/wiki/Fujiwara_no_teika#cite_ref-keene-
683_82-0
Between the Back button & the highlighting, I can't say I've ever heard anyone 
complain.

Original comment by gwe...@gmail.com on 6 Dec 2009 at 6:50

[GoogleCodeExporter]

When I try it, there is no indication of which backlink I came from.  Maybe 
it's a
customization you have.

Original comment by fiddloso...@gmail.com on 6 Dec 2009 at 7:37

[GoogleCodeExporter]

Before I commented, I asked some people in the #wikipedia IRC channel to try it 
out;
all 3 reported seeing the indicative highlighting and 1 told me that this was 
done
via CSS.

Original comment by gwe...@gmail.com on 6 Dec 2009 at 7:49

[GoogleCodeExporter]

What can I say?  I don't see it -- maybe it's my color vision? -- and I can't 
see how
the CSS would know which link was the one you most recently clicked on.  
Perhaps if
all this were explained to me, I could be persuaded to implement this feature.  
But
I'm still skeptical that the increase in complexity would be justified by the 
slight
increase in compression of the document that would be achieved.  There's also
additional complexity for the reader, who must now select from several possible
back-links instead of stupidly clicking on the one link provided.

Original comment by fiddloso...@gmail.com on 6 Dec 2009 at 8:28

[GoogleCodeExporter]

> and I can't see how the CSS would know which link was the one you most 
recently 
clicked on

I guess that it stores the information by having multiple links go to the same 
line, 
but not the same anchor. Here's one double-footnote use:

Teika's appeals to the unrelenting Michichika failed,<sup id="cite_ref-and-god-
hardened-his-heart_24-0" class="reference"><a 
href="#cite_note-and-god-hardened-his-
heart-24"><span>[</span>25<span>]</span></a></sup> and so Shunzei stepped in 
with an 
eloquent letter (the well-known <i>Waji sojo</i>; "Appeal in Japanese" - 
writing in 
Japanese as opposed to the official Chinese was considered a mark of 
sincerity<sup 
id="cite_ref-and-god-hardened-his-heart_24-1" class="reference"><a 
href="#cite_note-
and-god-hardened-his-heart-24"><span>[</span>25<span>]</span></a></sup>) 
addressed 
to Go-Toba, arguing that such an exclusion was without precedent, and motivated 
by 
base jealousy on their opponent's part

Which links to:

<li id="cite_note-and-god-hardened-his-heart-24">^ <a href="#cite_ref-and-god-
hardened-his-heart_24-0"><sup><i>a</i></sup></a> <a href="#cite_ref-and-god-
hardened-his-heart_24-1"><sup><i>b</i></sup></a> pg 16 of Brower 1972</li>

I don't know much about the HTML involved, but the # in the hrefs makes it look 
as 
if one can have sub-footnotes, if you will.

---

And it's not just the compression. It's DRY; it's honesty. 2 things should be 
separate only if they are different. If a reference is being used more than 
once, 
users should be able to easily see where else it is being used, without 
performing 
an ad hoc whole-document search; is it the crucial linchpin of the work or just 
one 
of many referenced works? The answer can have major implications for how the 
reader 
should interpret the work.

Original comment by gwe...@gmail.com on 8 Dec 2009 at 6:44

[GoogleCodeExporter]

I see how the HTML works, but as far as I know there's no CSS property that you 
could
appeal to to light up the anchor you linked to most recently.

I don't see how DRY is relevant. DRY applies to source; this is a generated 
file. In
the markdown source you're free to reuse a single footnote as many times as you 
like.

I also don't see how it is "dishonest" to use two duplicate notes.  (If you 
really
believe that "2 things should be separate only if they are different," then 
shouldn't
you also introduce short abbreviations for any *sentences* that you happen to 
repeat
twice in your document?)  Anyway, the scheme you're proposing wouldn't be 
enough to
tell you all the places where the reference is being used, since there might be
another note that cites it with some accompanying text.  This note wouldn't be 
an
exact duplicate, and would still be a distinct note, so you'd still have to 
search.

Original comment by fiddloso...@gmail.com on 9 Dec 2009 at 4:49

[GoogleCodeExporter]

> I see how the HTML works, but as far as I know there's no CSS property that 
you
could appeal to to light up the anchor you linked to most recently.

I dunno either; there's too much CSS involved in page loads for me to look 
through,
(not that I would recognize the relevant code even if I saw it).

> I also don't see how it is "dishonest" to use two duplicate notes.  (If you 
really
believe that "2 things should be separate only if they are different," then 
shouldn't
you also introduce short abbreviations for any *sentences* that you happen to 
repeat
twice in your document?)

The dishonesty (perhaps 'misleading' would be better) is in presenting as 
separate
things that are not separate. ('Let me introduce you to Janet. Let me also 
introduce
you to my wife, Mrs. Jessup. What do you mean, where's my wife? She's right 
here, I
just introduced you.')

One should refer back. If I'm working through a proof and I have some elaborate
theorem or axiom, do I write it out in full each and every time I use it, or do 
I use
some short identifier like 'IV'? And sentences may vary for literary purposes or
analysis, such as in poems - what purpose is served by some endnotes being
fruitlessly & redundantly multiplied?

> Anyway, the scheme you're proposing wouldn't be enough to
tell you all the places where the reference is being used, since there might be
another note that cites it with some accompanying text.  This note wouldn't be 
an
exact duplicate, and would still be a distinct note, so you'd still have to 
search.

And what's wrong with this? If I write a reference using source B, and mention 
in it
that I didn't use source A (mentioning source A by name), why is it a blackmark
against this scheme that that note will not be merged with others? You do what 
you
can in clear unambiguous identical situations and default to different notes for
different references.

Original comment by gwe...@gmail.com on 13 Dec 2009 at 1:55