Discussion List Archives

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ddlm-group] Adding namespaces to dREL

I'm not at all attached to '|' and I believe that any of John's suggestions would also work grammatically. The double full colon '::' appeals to me most.

Yes, associating namespace identifiers with dictionaries is outside the scope of dREL. Any context in which dREL is used is responsible for specifying how objects in that context are bound to dREL category objects. While I have labelled this proposal 'namespaces' essentially it is just offering optional two-component names which contexts can use and bind as they see fit.

On Tue, 1 Dec 2020 at 02:11, Bollinger, John C <John.Bollinger@stjude.org> wrote:
Dear James and group,

Adding namespacing to dREL is a fine idea.  I'm on board.

I'm not so sure about the specific proposed syntax, however.  The '|' symbol is tightly associated with alternation (logical and / or bitwise) in my mind, and indeed, it is used (doubled) for that purpose in dREL.  There is a variety of other possibilities.  Some that I like better are

 - use '::' instead of '|', following the namespace separator convention of C++, Ruby, and a few other languages.
 - use '//' instead of '|', like a filesystem path, but with the '/' doubled to avoid ambiguity with the dREL division operator. This is a bit reminiscent of XPath, though XPath does not double the `/'.  Python does have a '//' operator, however, which might be cause for objection.
 - use '%' instead of '|'.  The association of '%' with the arithmetic modulus operation is not a problem for me because it seems vaguely thematically consistent.
 - reverse the identifier order and use '@' instead of '|', reminiscent of an e-mail address.

Although I like them less, in part because they also have operator associations for me, I would prefer even '!' or '~' over '|'.

Also, a question: do I understand correctly that associating namespace identifiers with dictionaries is left outside the scope of the dREL language itself?



John C. Bollinger, Ph.D., RHCSA
Computing and X-ray Scientist
Department of Structural Biology
St. Jude Children's Research Hospital

From: ddlm-group <ddlm-group-bounces@iucr.org> on behalf of Herbert J. Bernstein <yayahjb@gmail.com>
Sent: Friday, November 27, 2020 8:17 AM
To: James Hester <james.r.hester@gmail.com>; Group finalising DDLm and associated dictionaries <ddlm-group@iucr.org>
Subject: Re: [ddlm-group] Adding namespaces to dREL
Caution: External Sender. Do not open unless you know the content is safe.


On Thu, Nov 26, 2020 at 10:17 PM James Hester <jamesrhester@gmail.com> wrote:
Dear DDLm group,

Please have a look at the following proposal for adding namespaces to dREL and provide any feedback you have. Note that this has no implications for how data names are written in data files or for DDLm.
I have implemented this syntax by machine-parsing the adjusted grammar to create executable code and am happy that it is workable and does not introduce any parsing ambiguity.

and the plain text is below.

all the best,

# Proposal to add namespaces to dREL

## Summary

It is proposed that the new construction `<namespace> "|" <identifer>`
be added to dREL to allow disambiguation of `<identifier>` when
necessary. Note this is not relevant to the way in which data names are
written in data files.

## Introduction

dREL methods operate within a relational context. Tables
("categories") from this context are automatically bound to variables
within the dREL method. Occasionally, contexts may identify tables using a
two-part name: for example, if the table names are drawn from
different domains which happen to use the same table identifier, a
label associated with the domain itself could be used for

dREL currently does not have syntax to allow for such two-part

## New syntax

[the formal grammar for dREL](https://github.com/COMCIFS/dREL/blob/master/annotated-grammar.rst)
contains the following line defining an identifier:

ident = ID ;

where `ID` is defined by a suitable regular expression. The proposal is to change this to
the following:

    ident = ID ;
    nspace = ID BAR ;
    nident = [ nspace ] ident ;

and replacing `ident` by `nident` in those productions that may refer to categories.
In the following productions any appearances of `nident` were previously `ident`:

call = nident  LEFTPAREN [expression_list] RIGHTPAREN ;
dotlist_assign = nident "("  dotlist  ")" ;
att_primary = nident | attributeref | subscription | call ;
loop_stmt =  LOOP ident AS nident [ COLON  ident  [restricted_comp_operator  ident]] suite ;
with_stmt = WITH  ident  AS  nident  suite ;

## Example

Code to transform dictionary information from ddl2 to ddlm.

loop dd as ddl2|dictionary {
    ddlm|dictionary(.title = dd.title,
                    .version = dd.version

## Comments

1. Use of a namespace for a table identifier is optional (`nident = [ nspace ] ident`)
2. Vertical bar `|` has been chosen as the separator as this leads to an unambiguous grammar.
The only appearance of vertical bar in the grammar is as `||` meaning "bitwise or". `:` (colon)
leads to ambiguity due to its widespread use in other places in the grammar.
3. Namespaces are made available for function calls as DDLm dictionaries can define functions,
which may therefore have namespaces associated with them.
4. DDLm dictionaries already include a `_dictionary.namespace` attribute.
5. This proposal has no implications for the structure or appearance of data names
in data files.
6. Use of this facility is only likely (i) when interoperating with outside
domains, over which we have no naming control, or (ii) categories are named identically,
even if the data name as a whole is distinct.
7. The productions above have been tested using the [Lerche grammar parser](https://github.com/jamesrhester/Lerche.jl) to create Julia code that successfully transforms between DDL2
and DDLm dictionaries using dREL expressions, disambiguating the "dictionary" and
"category_key" categories that appear in both.
T +61 (02) 9717 9907
F +61 (02) 9717 3145
M +61 (04) 0249 4148
ddlm-group mailing list

Email Disclaimer: www.stjude.org/emaildisclaimer
Consultation Disclaimer: www.stjude.org/consultationdisclaimer
ddlm-group mailing list

T +61 (02) 9717 9907
F +61 (02) 9717 3145
M +61 (04) 0249 4148
ddlm-group mailing list

Reply to: [list | sender only]
International Union of Crystallography

Scientific Union Member of the International Science Council (admitted 1947). Member of CODATA, the ISC Committee on Data. Partner with UNESCO, the United Nations Educational, Scientific and Cultural Organization in the International Year of Crystallography 2014.

International Science Council Scientific Freedom Policy

The IUCr observes the basic policy of non-discrimination and affirms the right and freedom of scientists to associate in international scientific activity without regard to such factors as ethnic origin, religion, citizenship, language, political stance, gender, sex or age, in accordance with the Statutes of the International Council for Science.