| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
This is one of the places where a strong semantic is enforced to good effect. Misinterpretation of units data would obviously result in erroneous stimulus, measurements, etc, and at worst case could be catastrophic. Relying on "best judgment" is never a good idea when thinking of general applications over the long haul - well-intended judgment could also be horribly mistaken. I *strongly* urge you not to weaken this very important semantic validation.
Case in point:
The Mars Climate Observer crashed into the surface of Mars, a victim not of a technical problems, but of a semantic misunderstanding concerning the units of measure used to calculate the thrust.
'nuff said.
Tim
_________________________________
Timothy J. Wilmering
'314.234.6781
-<mailto:timothy.j.wilmering@boeing.com>
From: Ralph, John (Mission Systems) [mailto:John.Ralph@NGC.COM]
Sent: Wednesday, December 14, 2005 8:16 AM
To: stds-scc20-atml@listserv.ieee.org
Subject: Common schema units of measure
As some of you may be aware, there is a 'units' attribute applied to several elements in the Common schema. A regular expression has been defined to limit the contents to standard SI and some English units of measure.
Late breaking input has shown that this is too stringent, and doesn't permit "extended" units such as "V rms" or "V peak-to-peak", etc. Accordingly, two suggestions have been put forth:
1. Drop the regular expression validation on 'units'
2. Add a free-text "modifier" attribute that applies to 'units'
In option 1, we would then be dependent on the "good engineering judgement" of schema users to ensure that unusable or non-standard units of measure are not introduced.
Option 2 would still impose a regexp validation on the 'units' attribute, and schema users could put any sort of "modifier" they needed in the 'unitQualifier' attribute. This has the advantage of at least SOME level of standardization on units of measure.
This needs to be resolved ASAP, as there are participants currently implementing and they need a "locked down" schema to work against.
Thoughts?
John Ralph