[osis-core] Syntax Check

[Steve DeRose]

At 12:39 AM -0600 08/26/02, Todd Tillinghast wrote:
>Patrick,
>
>This is a departure from what I understood regarding the structure for a
>reference. 
>
>We MUST ALWAYS have a REFERENCE SYSTEM and WORK is always OPTIONAL.

that's the part I still don't get. If we don't have both a refsystem 
and a work,  at least by default, a reference is meaningless, right? 
Like what is "foo.1.1"?

But given that, isn't it simply a question of which one(s) we allow 
people to leave to default?

I think when average people write "KJV" or "Bible.KJV" or whatever, 
they will be thinking of the work, not the references system -- most 
people don't even know that there is more than one reference system. 
Thus, I think that if someone writes just "Bible.KJV:Matt.1.1" it 
behooves us to treat that as a work.

At one point we had the notion that works would be declared in the 
header, and that a work declaration could also specify the reference 
system(s) it supports. If we allowed works to simply specify their 
default reference system within their <work> declaration element, 
then specifying the work without a reference system anywhere else, is 
well-defined because you simply use that default.

I don't have a problem with allowing a reference like 
"(Hebrew):Psal.1.1", which gives only a reference system -- the 
meaning would then be, *any* version that supports that reference 
system (which itself is declared in the header).

I'm thinking that we may want to reserve prefixes in a later spec, so 
that OSIS-aware systems can count on them without everybody having to 
re-declare them all the time; but that we can add easily enough.

Does that make sense?

>
>There has been discussion of short cuts and aliases but in the end when
>all of the short cuts and aliases are expanded it is ok for a reference
>or self-id to have ONLY a reference system and it is NEVER ok for a
>reference or self-id to have ONLY a work.

Did I address this ok above?

>
>This is the first I am seeing "(split)" as a part of reference.  Can you
>explain what its purpose is?

That's the segmentID or splitID or whatever, for identifying the 
various pieces of a segmented verse.

>
>The way I understand the regex is the following.
>
>refsys(work):canonical@grain-canonical@grain
>
>Where:
>1) work is optional
>2) @grain is optional in both places
>3) there must be a value follwing the "-" symbol
>4) refsys: can be defaulted based on the declaration of defaultRefSys
>attribute in <osisText>
>
>
>the other alternative is
>alias:canonical@grain-canonical@grain
>where alias is refsys or refsys(work) based on an alias element in the
>header.
>
>If we want to allow a "work" name to substitute for "refsys(work)" based
>on the reference system declared for the work in the header, that would
>be doable if we required a reference system attribute for every work
>element.  I don't think this is a good idea because it overloads the
>meaning of the suffix.
>
>
>Todd
>
>>
>>  Guys,
>>
>>  Before I re-write the regex, is there general agreement on Steve's
>>  proposal:
>>
>>  work(refsys):canonical(split)@grain-canonical(split)@grain
>>
>>  (This may have been what Steve outlined while we were in Canada and I
>>  mashed into something else for posting, Apologies if that was what
>>  happened.)
>>
>>  So, whatever the declared syntax of "work" it is followed by optional
>>  "(refsys)" plus ":" plus "canonical" plus an optional "(split)" (to
>>  indicate a subdivision that is not part of the canonical reference),
>>  followed by an optional grain "@grain (which is either str or cp
>>  followed by a colon, then square brackets), then the range indicator,
>>  plus all the foregoing again.
>>
>>  Comments, modifications?
>>
>>  Thanks!
>>
>>  Patrick
>>
>>  --
>>  Patrick Durusau
>>  Director of Research and Development
>>  Society of Biblical Literature
>>  pdurusau@emory.edu
>>


-- 

Steve DeRose -- http://www.stg.brown.edu/~sjd
Chair, Bible Technologies Group -- http://www.bibletechnologies.net
Email: sderose@speakeasy.net
Backup email: sjd@stg.brown.edu