[osis-core] Re: Call me if you do not agree with my statements regarding refsys(work) rather than work(refsys)!

[Patrick Durusau]

Guys,

Todd forwarded the following question that I think we need to answer 
before I write yet another regex!

Todd Tillinghast wrote:

>Patrick,
>
>Call me if you do not agree with my statements regarding refsys(work)
>rather than work(refsys)!
>
>This is a very important point!  
>
>If work(refsys) is not just an error or misconception, then I would like
>to understand the need that is driving to make refsys optional and work
>mandatory and work to provide appropriate defaults, shortcuts, or
>aliasing solutions that allow us to retain refsys(work) as the
>appropriate structure.
>
work(refsys) was not in error. (Although I  think we need to use "[]" 
just to be consistent in the regex.)

Reasoning:

1. Default case: For the majority of cases, there is no practical 
difference between the work and the reference system. Bible.KJV could 
just as easily be a reference to the reference system of the KJV (used 
by all major English translations, save the TEV, CEV and then by 
reference). Note that even the Catholic translations follow it save for 
extension, not inconsistent with the common parts.

2. It seems to me less confusing to users to say: "Use the work 
reference for osisRef just as you would a normal citation." Distinguish 
this from describing to users that they should use the refSystem (which 
looks to them like the "work") and if they are citing particular "work" 
they have to indicate that as well. That sort of distinction, which I 
agree fully is useful, looks like something we could do as part of 
processing and not as part of markup.

3. Note that having the optional refSys allows Harry to address the 
legitimate case  where he has a standard work that has multiple 
reference systems to address that work.

I understand the notion that always having a refSys and an optional work 
appears to be more transparent but that depends upon the idea of having 
a universal refSys than can be applied to any work. Hmmm, I suppose 
based on #1 you could argue that is the case for most Bible references 
but I still think it would be confusing to most users, at least with the 
names we have chosen for these attributes.

(I really hate replying to your messages because they are likely to make 
me change my mind or at least reconsider my positions. Guess I need to 
start filtering your mail. ;-)

Suppose, just suppose, not commiting on this, we had the following:
(I am using different words so as to not confuse  with other terms)


citation(particular edition) (citation implying a reference system or 
being a shorthand for a reference system?)

In other words:

Bible.KJV:Gen.1.1   -> (Bible.KJV being a defaulted refSys in current terms)

Bible.KJV[NIV]:Gen.1.1 -> A citation to the NIV edition known as Gen.1.1 
using the Bible.KJV refSys (in current terms)

Harry:

Does this get you:

augustine.confessions[pusey.1880]:X.iii.20 -> which defaults to a 
standard reference system, yet indicates a particular edition?

Noting that augustine.confessions and pusey.1880 would have to be 
defined elsewhere in the text.

So,  the fuller syntax would be:

refsys[work]:canonical[split]@grain-canonical[split]@grain

Or in the terms I propose below:

WORK[EDITION]:citation[split]@grain-citation[split]@grain

Note that I "sold out" and gone over the Todd's position!


SEMANTIC CHANGE:

In order to "sell" this to users I think we need to change the 
nomenclature that we use for these (and possibly other) syntax constructs.

***user instructions***

Optionally record the WORK, then optionally an [EDITION], followed by 
COLON, followed by the CITATION.

Most texts (check your documentation) have a default WORK and EDITION, 
in those cases you can just enter the CITATION. To pick a particular 
EDITION of the Bible text (that uses KJV numbering), for instance, 
enter: [NASB]:Gen.1.1 to make a reference to the New American Standard 
Bible translation of the text of Genesis 1:1.

(note to OSIS group, note in the last sentence I am defaulting part of 
the current refSys to Bible.KJV. Would require making the portion prior 
to EDITION optional, as well as EDITION being optional in the regex.)

***end user instructions***

Realize that is not how we have spoken of it in the past but I am 
getting concerned that if we (at least me) can get confused with all our 
syntax and its changes, we will have an elegant system that is about as 
readable as ESIS format (a remark meant for Steve's enjoyment). 
(Actually I never found it that hard to read but that's SGML folks for you.)

I will try to post a little later some thoughts on the various postings 
for syntax in the header. (Although I also need to get a TEI paper out 
this morning, so bear with me.)

Patrick


>
>
>Todd
>

-- 
Patrick Durusau
Director of Research and Development
Society of Biblical Literature
pdurusau@emory.edu