[MOBY-l] Re: MOBY and the "REST/SOAP" debate

[mwilkinson at gene.pbi.nrc.ca]

Lincoln Stein wrote:

> If we have the data formats in hand, it doesn't matter whether we fetch them
> by GETing or URL or POSTing a SOAP message.

This is precisely the sentence that was in my head when I woke up this morning!
Since we aren't really using RPC (at least, not at this point),  there is nothing
that we are doing so far in MOBY that couldn't be done through plain old GET.

I think it is more useful at this point to focus on:

(1) the behaviour of the system we are trying to build - e.g. are tangential
relationships derived from cross-references going to produce interpretable results?
Are cross-references themselves going to be uniformly interpretable and reliable
since they are optional and server dependent?  Is it "dangerous" to so loosely
define the relationship between the input and output of a service?

(2) the level of data complexity required to achieve this behaviour - e.g. how many
ontologies do we need, how detailed do they need to be, how much data should we pass
in our objects, should they be hierarchical or loose collections...

(3)  the data formats - presumably XML, but how should we structure that data to
achieve the behaviour and complexity we need?

I have a lot of ideas about the behaviour of the system, and have discussed them at
length both on and off the list... but certainly my conversations with Carole
(myGrid) a few weeks ago opened my eyes to the potential dangers lurking in my
approach.  I am still hashing out these things, especially with the myGrid people,
though sadly much of this discussion is happening off the list.

As I see it, we're in a bit of a holding pattern at the moment while we work out
what role ontologies will play in determining MOBY behaviour, and whether or not
these need to be built from scratch, or whether existing ontologies, such as those
generated by the myGrid project, will work in the somewhat looser (currently) MOBY
environment.  Once we have these nailed down, we will be in a much better position
to decide what underlying technologies we need to achieve this...  I think the
current discussion, though academically interesting, is likely a bit premature in
its intensity and that the best implementation choices will become much clearer when
we know for certain what exactly we need to achieve.

I'm also convinced that there are some things that we will not be able to determine
through debate and academic missives...  There are probably multiple equally valid
ways to achieve the goal, though they may or may not be equally difficult to
implement - I'm sure there will  be times when we'll just have to set things up and
try them out!  :-)

M



--
--------------------------------
"Speed is subsittute fo accurancy."
________________________________

Dr. Mark Wilkinson, RA Bioinformatics
National Research Council, Plant Biotechnology Institute
110 Gymnasium Place, Saskatoon, SK, Canada

phone : (306) 975 5279
pager : (306) 934 2322
mobile: markw_mobile at illuminae.com