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


Help: OASIS Mailing Lists Help | MarkMail Help

office message

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

Subject: Re: [office] Not an JIRA issue, yet....

Patrick Durusau <patrick@durusau.net> wrote on 08/06/2010 09:56:27 AM:

> I wanted to ask about something before possibly raising it as a JIRA 
> To what extent, if any, are " host-dependent properties: " 19.644 
> table:formula, different from being implementation dependent?

See section Part 2, section 3.4 "Host-Defined behaviors".  If we use 
"host-dependent" then this is an error.  You probably also want to give a 
cross reference in 19.644 pointing to Part 2, Section 3.4.

Host-defined is different than implementation-defined.

If something is called implementation-defined, then the implementation 
must define the behavior.  But for some things in Part 2, the behavior is 
not determined by the formula evaluator, but by the hosting application. 
If we called it implementation-defined, then this would presume that the 
formula evaluator is aware of its host.  This breaks the desired 
independence of the component model.  We want the formula evaluators 
(ideally) to be plugable, meaning that they do not have an explicit 
dependency on their host.

This is a standard engineering technique.  If you have a mutual dependency 
between components A and B and you want to factor out that dependency, 
then you introduce a new interface, call it C, and use that to express the 
dependencies.  So instead of having Part 2 depend on Part 1, we use 
host-defined properties to express the dependencies and move their 
definition into Part 1.

> I ask because we use terms like "HOST-CASE-SENSITIVE" and then give 
> the attribute with a default value. 
> This *does not appear in ODF 1.0/1.1* so we are not committed to 
> this terminology, yet.


> Note that HOST-LOCALE doesn't have a default value but is said to be
> "implementation-defined." 

Correct.  In this case, since we are in Part 1, the implementation is the 
ODF Spreadsheet Consumer.

> Is there a distinction between an implementation and a host? 

Host is a species of implementation.  In theory you could use 
"implementation" everywhere, but then you would need to be very careful to 
say exactly which implementation you were giving a requirement on: the 
formula evaluator (Part 2) or the ODF Spreadsheet Consumer (Part 1).

Not that saying that something was "implementation-defined" in Part 2 
would imply that the formula evaluator would need to define it.  But this 
is not what we want.  We really want to say "This property is defined by 
the spreadsheet application".  So we could say, "ODF Spreadsheet 
Consumer-defined (Part 1) behavior".  But this is also not desired, since 
Part 2 could be used in other contexts other than an ODF spredsheet 
document.  But in those other contexts the existence of host-defined 
properties is still relevant.  We want to express a dependency on a "host" 
where the host could be Part 1, or it could be something else.


> Hope everyone is looking forward to a great weekend!
> Patrick
> -- 
> Patrick Durusau
> patrick@durusau.net
> Chair, V1 - US TAG to JTC 1/SC 34
> Convener, JTC 1/SC 34/WG 3 (Topic Maps)
> Editor, OpenDocument Format TC (OASIS), Project Editor ISO/IEC 26300
> Co-Editor, ISO/IEC 13250-1, 13250-5 (Topic Maps)
> Another Word For It (blog): http://tm.durusau.net
> Homepage: http://www.durusau.net
> Twitter: patrickDurusau

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