Support pointer to a subfield of a given field about marcspec

[nichtich]

Looks like the MARC21 documentation (see for instance here also uses the proposed syntax where indicator can also be first:

$a - Title
...
245 10$aNineteenth-century bird prints.

This all would be valid MARCspec, the first two are typical in documentation:

• $a
• 245_10$a
• 245a
• 245$a
• 245a_10
• 245$a_10

[cKlee]

; refer to a character position or range
characterSpec = [ fieldTag ] "/" characterPositionOrRange

Why is the fieldTag optional?

Does the MARCspec $a refer to all subfields "a" in the record?

Are these true:

100$_ ; valid (subfield "_" of field "100")
100$_1 ; valid (subfields "_" and "1" of field "100")
100$__ ; invalid (missing indicators)
100$__1 ; valid  (subfield "_" of field "100" within context of indicator 1 == 1)

?

Because of the allowed characters for subfield tags

 ! " # $ % & ' ( ) * + , - . / : ; < = > ? { } _ ^ ` ~ [ ] \

I expect some problems:

100_ ; valid subfield "_" or invalid because missing indicators?
100$a_ ; valid subfields "a", "_" or invalid because missing indicators?

Maybe the solution is to restrict the position of subfieldSpec at the end of the marcspec and require a "$"?

245_10$a ; valid
245__0 ; valid
245_1 ; valid
245$a ; valid
245$$ ; valid (subfield "$")
245$_ ; valid (subfield "_")

245_$ ; invalid (missing indicators)
245$ ; invalid (missing subfield tags)
245a ; invalid (missing "$")

[cKlee]

There is another problem with indizes see #4

How is something like 245$a[0] interpreted? Is it the first subfield "a" or is it the subfields "a", "[", "0" and "]"?

Another possibility is to prefix every subfield tag with a "$".

245$a$[$0$]$$ ; valid (subfields  "a", "[", "0", "]" and "$")

245$ab ; invalid (missing "$" for second subfield tag)

245$a[0] ; valid (first subfield "a")

245$__ ; invalid (missing indicators)

245$__1 ; valid (subfield "_" within context of indicator 1 == 1)

245_1$_ ; valid (same as above)

245_1$_a ; invalid (missing "$" before subfield tag "a") 

[cKlee]

It really troubles me using only a subfield tag or a character position or range as a MARCspec. Shouldn't a MARCspec always be a reference to data in a MARC record? Otherwise it must be defined what a MARCspec references within the referenced data?

[nichtich]

This thread mixes too many different topics.

[cKlee]

Subfield ranges and indizes are done with tag v0.4draft. I don't think that it is a good idea to allow a MARCspec relative to a given subfield. When exchanging MARCspecs then you have to state if the MARCspec is relative to a record, a field or a subfield. This won't work.

[nichtich]

ok, pointers to parts of a field may be expressed with strings that follow the syntax rules fieldSpec, characterSpec or subfieldSpec, but this should not be called MARCspec. I note on reusing part of MARCspec syntax should be included nevertheless.

[cKlee]

This issue is done with v0.5draft