OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

oslc-ccm message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [oslc-ccm] Order of contributions in a configuration

The rationale for using strings is that it allows a configuration provider to insert a new contribution into an existing list without having to renumber all contributions. Since this ordering string is not expected to be a natural language name of any form, locale-specific collation is not desired.  I will update the draft spec to indicate this, and to recommend that tools restrict the string to ASCII characters to remove any possible ambiguity with encodings, etc.


Inactive hide details for David Honey ---07/09/2014 10:13:00 AM---The current configuration management spec at  https://tools.oDavid Honey ---07/09/2014 10:13:00 AM---The current configuration management spec at  https://tools.oasis-open.org/version-control/browse/ws

From: David Honey <david.honey@uk.ibm.com>
To: oslc-ccm@lists.oasis-open.org
Date: 07/09/2014 10:13 AM
Subject: [oslc-ccm] Order of contributions in a configuration
Sent by: <oslc-ccm@lists.oasis-open.org>

The current configuration management spec at https://tools.oasis-open.org/version-control/browse/wsvn/oslc-ccm/trunk/specs/config-mgt.html shows this for defining the order of contributions in a configuration: 
    Prefixed Name 
    An indication of the order of the contribution, relative to other contributions. Contributions are sorted lexicographically on this property to resolve component skew.

One of the issues with using a string is to define what is meant by "
sorted lexicographically". Best practice in Java is to use a Collator for a specific locale. In which case should this be the locale of the client, a consuming server, the contributing server? This makes it very difficult for the provider of the contributor to know how the order data it provides will be determined and whether it represents the order used by that provider. 

I have a suggestion to make this better defined. What about requiring that the value uses an integer related
xsd type such as xsd:integer, xsd:int, xsd:long and so on. That allows a precise locale-independent order to be defined based on integer values. If we want to represent undefined ordering, then perhaps the specification should make this zero-or-one instead of exactly-one. On the other hand, if the intent is that the contributions for every oslc_config:Configuration must be fully ordered, perhaps the specification should define that each contribution must have a unique value of oslc_config:contributionOrder. In Java, most collators seem to be case-independent, such that the string "ABC" has the same order value as "abc", "aBc" and so on. That's another advantage for using a numeric integer datatype instead of a plain literal. 

Best regards,
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]