[DAS] Re: Our identifier doc and proposal

Brian King kingb@doubletwist.com
Tue, 27 Nov 2001 19:44:46 -0800


This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01C177BF.05BAED80
Content-Type: text/plain;
	charset="iso-8859-1"


 >    urn:lsid:informatics.mpi.com:plate/glycerol/freeze:12345
> 
> I think the top level namespace, e.g. "plate" should be hard-and-fast
> data types.  Is this envisioned by the I3C?
> 
> Lincoln

Lincoln,
 
I'm one of the authors of the I3C ID proposal.  We've just started
discussing the ontology-in-ID issue, so take my answer as only my own point
of view. I believe we should not encode information such as object type in
the I3C IDs.  An ID should just be a unique string, and nothing more.  Since
ID creation is decentralized in the I3C interoperability architecture we
need to encode an authority and namespace to ensure uniqueness, but that's
the only purpose of that information.  The problem with having information
encoded into IDs is that developers immediately start parsing this
information to speed queries or map data to physical locations, or worse,
create their IDs based on such assumptions.  Such schemes are unreliable and
unstable.  They're unreliable because different paths in the ID imply
different processing, and since IDs are passed everywhere, the logic for
this processing gets scattered all across the system.  They are unstable
because the information that seemed relevant in the last system becomes
obsolete.  I have been a software contractor on several projects, and often
the first task on entering a project is untangling the IDs.  We should avoid
turning the ID into a query language in itself.
 
Regards,
Brian King
DoubleTwist, Inc

------_=_NextPart_001_01C177BF.05BAED80
Content-Type: text/html;
	charset="iso-8859-1"

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">


<META content="MSHTML 5.00.2314.1000" name=GENERATOR></HEAD>
<BODY>
<DIV><FONT face=Arial size=2><BR>&nbsp;&gt;&nbsp;&nbsp;&nbsp; 
urn:lsid:informatics.mpi.com:plate/glycerol/freeze:12345<BR>&gt;&nbsp;<BR> &gt; 
I think the top level namespace, e.g. "plate" should be hard-and-fast<BR>&gt; 
data types.&nbsp; Is this envisioned by the I3C?<BR>&gt;&nbsp;<BR> &gt; 
Lincoln<BR></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=080191403-28112001>Lincoln,</SPAN></FONT></DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=080191403-28112001></SPAN></FONT>&nbsp;</DIV>
<DIV><SPAN class=080191403-28112001><FONT face=Arial size=2>I'm one of the 
authors of the I3C ID proposal.&nbsp; We've just started discussing the 
ontology-in-ID issue,&nbsp;so take&nbsp;my answer as only my own point of view. 
I believe we should not encode information such as object type in the I3C 
IDs.&nbsp; An ID should just be a unique string, and nothing more.&nbsp; 
Since&nbsp;ID creation is decentralized in the I3C interoperability 
architecture&nbsp;we need to encode an authority and namespace to ensure 
uniqueness, but that's the only purpose of that information.&nbsp; The problem 
with having information encoded into IDs is that developers immediately 
start&nbsp;parsing this information to speed queries or map data to physical 
locations, or worse, create their IDs based on such assumptions.&nbsp; Such 
schemes are unreliable and unstable.&nbsp; They're unreliable because different 
paths in the ID imply different processing, and&nbsp;since IDs are passed 
everywhere, the logic for this processing gets scattered all across the 
system.&nbsp; They are&nbsp;unstable because&nbsp;the information that seemed 
relevant&nbsp;in the last system becomes obsolete.&nbsp; I have been a software 
contractor on several projects, and often the first task on entering a project 
is untangling the IDs.&nbsp; We should avoid turning the ID into a query 
language in itself.</FONT></SPAN></DIV>
<DIV><SPAN class=080191403-28112001><FONT face=Arial 
size=2></FONT></SPAN>&nbsp;</DIV>
<DIV><FONT face=Arial size=2><SPAN 
class=080191403-28112001>Regards,</SPAN></FONT></DIV>
<DIV><SPAN class=080191403-28112001><FONT face=Arial size=2>
<DIV><FONT face=Arial size=2>Brian King</FONT></DIV>
<DIV><FONT face=Arial size=2>DoubleTwist, 
Inc</FONT></FONT></SPAN></DIV></DIV></BODY></HTML>

------_=_NextPart_001_01C177BF.05BAED80--