collapse question

I have another question, about how collapsing should be done.
specification says the following:

“citation-number” collapses numeric citations from [1, 2, 3] to [1-3]

“year” collapses authors from (Doe 2000, Doe 2001) to (Doe 2000,

2001)

“year-suffix” collapses as “year”, but also collapses (Doe 2000a,

Doe 2000b) to (Doe 2000a, b) (ignored if

disambiguate-add-year-suffix is missing)

I’m wondering how should this collapse: (Doe 2000a, Doe 2000b, Doe 2001)

If a valid year-suffix implies year, then does this become: (Doe 2000a, b, 2001)

… or should it be handled in some other way?

It’s a rare case, I know, but if I deal with it on authority now, it
will be out of the way.

@Bruce, thanks for the edit to the specification of groups, I think
that puts it in context (although you never know, I do have a
well-tapped but largely unexplored capacity for misunderstanding).

Frank

I have another question, about how collapsing should be done.
specification says the following:

“citation-number” collapses numeric citations from [1, 2, 3] to [1-3]

“year” collapses authors from (Doe 2000, Doe 2001) to (Doe 2000,

2001)

“year-suffix” collapses as “year”, but also collapses (Doe 2000a,

Doe 2000b) to (Doe 2000a, b) (ignored if

disambiguate-add-year-suffix is missing)

I’m wondering how should this collapse: (Doe 2000a, Doe 2000b, Doe 2001)

If a valid year-suffix implies year, then does this become: (Doe 2000a, b, 2001)

That looks a little strange, but I think that would have to be
correct; I see no other obvious way.

Does anyone have a different opinion?

BTW, this is exactly the sort of thing we ought to formalize in a test suite.

It’s a rare case, I know, but if I deal with it on authority now, it
will be out of the way.

@Bruce, thanks for the edit to the specification of groups …

Sure thing.

Bruce

Interesting… I didn’t think about this possibility. I would think
that, in order to collapse to (Doe 2000a, b, 2001), you would need both
the year and year-suffix options set, though. Right?

In this case my implementation should only handle the first option,
whichever it might be.

Andrea

I have another question, about how collapsing should be done.
specification says the following:

“citation-number” collapses numeric citations from [1, 2, 3] to [1-3]

“year” collapses authors from (Doe 2000, Doe 2001) to (Doe 2000,

2001)

“year-suffix” collapses as “year”, but also collapses (Doe 2000a,

Doe 2000b) to (Doe 2000a, b) (ignored if

disambiguate-add-year-suffix is missing)

I’m wondering how should this collapse: (Doe 2000a, Doe 2000b, Doe 2001)

If a valid year-suffix implies year, then does this become: (Doe 2000a, b, 2001)

That looks a little strange, but I think that would have to be
correct; I see no other obvious way.

Does anyone have a different opinion?

BTW, this is exactly the sort of thing we ought to formalize in a test suite.

That’s a great idea. I’m at the point now where the engine is stable,
and it’s just waiting for the definitions of the attributes. It’s
quite a mountain to climb, and a full set of tests would hugely
simplify the remaining work. As a newcomer to CSL, there are a lot of
details that I don’t fully understand. I don’t mean to be a burden
(obviously), but the questions and calls for explanation will keep
coming.

I’ll work up a simple format for high level input/output tests that
can be parsed easily by system XYZ, and written up with a minimum of
typing. I can put it into citeproc-js, but maybe a better location
would be the csl svn?

Frank

I’m wondering how should this collapse: (Doe 2000a, Doe 2000b, Doe 2001)

If a valid year-suffix implies year, then does this become: (Doe 2000a,
b, 2001)

That looks a little strange, but I think that would have to be
correct; I see no other obvious way.

Does anyone have a different opinion?

BTW, this is exactly the sort of thing we ought to formalize in a test
suite.

A quick search with Google Scholar yielded these results (with queries like
"1999a b 2000" and “1999a 1999b 2000”):
(Alcock et al. 1999a,b, 2000)
(Vander Zanden et al. 1999a,b, 2000)
(Steudie, 2000a b; 2001)
(Geller et al., 1998c,d, 2000a,b)
(for reviews, see Pertwee, 2000b; 2001a,b, 2002; Piomelli
et al., 2000).

So a slightly more robust test case might use different delimiters between
the year-suffix-disambiguators (e.g. a comma between the a and the b) and
between citations from different years (e.g. a semi-colon, or " ,"
[space-comma]). I don’t know if this can already be described in CSL though.

Rintze

I’m wondering how should this collapse: (Doe 2000a, Doe 2000b, Doe 2001)

If a valid year-suffix implies year, then does this become: (Doe 2000a,
b, 2001)

That looks a little strange, but I think that would have to be
correct; I see no other obvious way.

Does anyone have a different opinion?

BTW, this is exactly the sort of thing we ought to formalize in a test
suite.

A quick search with Google Scholar yielded these results (with queries like
"1999a b 2000" and “1999a 1999b 2000”):
(Alcock et al. 1999a,b, 2000)
(Vander Zanden et al. 1999a,b, 2000)
(Steudie, 2000a b; 2001)
(Geller et al., 1998c,d, 2000a,b)
(for reviews, see Pertwee, 2000b; 2001a,b, 2002; Piomelli
et al., 2000).

So a slightly more robust test case might use different delimiters between
the year-suffix-disambiguators (e.g. a comma between the a and the b) and
between citations from different years (e.g. a semi-colon, or " ,"
[space-comma]). I don’t know if this can already be described in CSL though.

Rintze,

It would be great if you could produce a test case or cases for this,
when the archive is set up.

FB

I believe I already added this variable (‘year-suffix’).

Bruce

I noticed that update to the schema, but I don’t understand how to use the
variable ‘year-suffix’ in CSL. At what point in the style (for instance APA)
do you declare it?

Rintze

It’d go after the year (in the macro), of course :wink:

But AFAIK, it’s not yet implemented.

Bruce

Uhm, okay. Is this (based on APA) then the supposed usage:

How do you then assign a value to the year-suffix variable (e.g. “,”)?

Rintze

In my implementation that variable is automatically instantiated with
the disambiguation suffix. That is to say, the above example, coupled
with a

would held something like:

(Rossato, 2008aa; Rossato, 2008bb)

Is that right?

Andrea

Scrap that last sentence from my previous email: I was confusing support for
formatting of year-suffices (a problem that should indeed be solved with the
addition of the variable “year-suffix”) with the ability to set custom
delimiters for subsequent year-suffices (for which I wrote a ticket some
time ago:
http://sourceforge.net/tracker2/index.php?func=detail&aid=2212677&group_id=117435&atid=678021).
Do you have any ideas the latter could be implemented? E.g. in the example

(Steudie, 2000a b; 2001)

the delimiter between different years ("; “) is different from that between
subsequent year-suffices (” "). To stress this point: I have just been
toying with Zotero’s csledit.xul, and if you replace “” with “” in the APA style and remove all the commas in the
style, you still get:

(Doe 1999a, b, 2000)

Rintze

Scrap that last sentence from my previous email: I was confusing support for
formatting of year-suffices (a problem that should indeed be solved with the
addition of the variable “year-suffix”) with the ability to set custom
delimiters for subsequent year-suffices (for which I wrote a ticket some
time ago:
http://sourceforge.net/tracker2/index.php?func=detail&aid=2212677&group_id=117435&atid=678021).
Do you have any ideas the latter could be implemented? E.g. in the example

(Steudie, 2000a b; 2001)

the delimiter between different years ("; “) is different from that between
subsequent year-suffices (” "). To stress this point: I have just been
toying with Zotero’s csledit.xul, and if you replace “” with “” in the APA style and remove all the commas in the
style, you still get:

(Doe 1999a, b, 2000)

Fun one! And an occasion for a mea culpa from me (to Bruce
particularly) about the unseemly fuss I made over capitalization of
terms. This requires very similar logic.

At first I also thought that this should be handled by a header
option, but it’s simpler to tell the year-suffix itself to change the
back-splice string when it’s the first item in the item cite that gets
rendered. So something like …

… would do the trick. If the engine knows how to identify that a
cite is not the first in a cluster to be rendered, and that an element
is the first in the cite to produce rendered output, it can pick this
up and do the right thing. Plus it would work with other cases
requiring splice changes that might come down the pipe.

Frank

But that’s foreign to how CSL has been designed. Why not just build on
the existing delimiter support?

Bruce

Scrap that last sentence from my previous email: I was confusing support for
formatting of year-suffices (a problem that should indeed be solved with the
addition of the variable “year-suffix”) with the ability to set custom
delimiters for subsequent year-suffices (for which I wrote a ticket some
time ago:
http://sourceforge.net/tracker2/index.php?func=detail&aid=2212677&group_id=117435&atid=678021).
Do you have any ideas the latter could be implemented? E.g. in the example

(Steudie, 2000a b; 2001)

the delimiter between different years ("; “) is different from that between
subsequent year-suffices (” "). To stress this point: I have just been
toying with Zotero’s csledit.xul, and if you replace “” with “” in the APA style and remove all the commas in the
style, you still get:

(Doe 1999a, b, 2000)

Fun one! And an occasion for a mea culpa from me (to Bruce
particularly) about the unseemly fuss I made over capitalization of
terms. This requires very similar logic.

At first I also thought that this should be handled by a header
option, but it’s simpler to tell the year-suffix itself to change the
back-splice string when it’s the first item in the item cite that gets
rendered. So something like …

… would do the trick. If the engine knows how to identify that a
cite is not the first in a cluster to be rendered, and that an element
is the first in the cite to produce rendered output, it can pick this
up and do the right thing. Plus it would work with other cases
requiring splice changes that might come down the pipe.

But that’s foreign to how CSL has been designed. Why not just build on
the existing delimiter support?

Could do that too. I guess it’s unlikely that other doctoring of
splices will be needed, and that would be more intuitive when writing
CSL code.

Frank

I guess that should do the trick. The delimiter between subsequent years
(e.g. the "; " in (Steudie, 2000a b; 2001)) can probably default to the same
delimiter used to delimit grouped and non-collapsed citations (in this case
(Doe, 2000; Joe 2001)), but that might be more of an implementation issue,
although it could be documented in the schema. The list of examples I gave
includes one exception from this rule:

(for reviews, see Pertwee, 2000b; 2001a,b, 2002; Piomelli et al., 2000)

where I would expect “2001a,b; 2002;” instead of “2001a,b, 2002;”. Unless
some of the major style guides mention something to this effect, I would
write this one of as a fluke (it might very well be an error, and it doesn’t
make a lot of sense).

Rintze