Andy I wasn't suggesting we get rid of query_uri - quite the opposite
in fact. just that the single source uri would have to be specified
with a uri as conceptually all other commands may not contain the root
uri. This also seems to me means we will have to update das1 code to
cope with multiple query uris.
On 15 Sep 2009, at 17:56, Andy Jenkinson wrote:
On 15 Sep 2009, at 16:35, Jonathan Warren wrote:
I agree with Andy on both these (we talked about versioning before).
The version numbers really have no meaning at the moment (no web
pages anywhere actually explain what a different version means) and
don't seem to be used at all in data sources ( I'm guessing people
end up just copying the version numbers from examples given.
I've always had an issue with the commands like this http://www.ebi.ac.uk/das-srv/genomicdas/das/eqtl_rat_cis_fat
not being a valid das command as it's the most natural request for
a person new to das to make. So giving it a specific purpose and
response is a good idea.
My only concern is how to handle these if we start using the power
of multiple query_uri s per das source (inherited from DAS2, which
we have started to talk about, rather than the das1 style where all
urls have a root) as currently there is no "root" url specified in
the DAS2 spec in the sources document...?? So this would have to be
specified as another capability? or you could infer it from the
features command, but obviously not the sources cmd!!!
My take on this is that the root URI identifies the source. In a
conceptual sense the definition of a source is merely a combination
of commands acting on a common set of data. It is not really
important where that information comes from (a registry, a server, a
flat file...) because a server by itself does not really mean
anything. So the URI http://www.ebi.ac.uk/das-srv/genomicdas/das/eqtl_rat_cis_fat
is not actually meaningful, even less so given it is not even a
resolvable URL.
The query URI system inherited from DAS/2 has the potential to allow
the commands to be served from different locations on the web. It is
not something we have needed up to now (all query URIs start with
the same path), and does add confusion but I can see it being used
for stylesheets. For example a "sequence ontology stylesheet" served
from a single location.
But the biggest reason to have it is because of the registry. The
registry assigns its own root URIs for a DAS source (e.g. DS_1234),
which means it is necessary to provide another URI used to actually
query it. Since we already have a way of doing it in the sources
document, I don't really want to change it now. It seems we might as
well just embrace the extra flexibility and merely describe it better.
On 15 Sep 2009, at 15:47, Andy Jenkinson wrote:
On 15 Sep 2009, at 15:19, Thomas Down wrote:
Capabilities are stated in the sources document:
<CAPABILITY type="das1:maxbins" />
Ah, interesting. I'd seen that, of course, but hadn't explicitly
linked this with the idea of capabilities as listed in the X-DAS-
Capabilities header (although of course it makes a lot more sense
to have one set of capability metadata, rather than two!). There
are a couple of issues here:
1. The SOURCES examples all say "das command" in the
type attribute of the CAPABILITY element, whereas many of the
capabilities don't actually map to commands. I notice that the
latest DAS1.6 draft does give an example to clarify this.
2. X-DAS-Capabilities entries are versioned whereas
SOURCES capabilities aren't, which makes them look rather
different. (and I note that the 1.6 spec is bumping up the
version numbers on some of the existing capabilities...)
How about versioning capabilities in SOURCES, e.g.:
<CAPABILITY type="features" version="1.1" query_uri="http://noranti.derkholm.net/das/mydata/features
" />
<CAPABILITY type="maxbins" version="1.0" />
Assume any missing version attributes are "1.0" and everything
should be backwards compatible.
Indeed I did increment the version, just because it seemed the
right thing to do. However as far as I am aware these per-
capability versions are totally superfluous when taken in context
with the X-DAS-Version header, i.e. we do NOT want to make it
possible to implement DAS 1.6 and features 1.0, for example. This
could create a whole world of pain!
IMO the per-capability version is unnecessary and confusing.
ProServer does use it internally, but that can be easily changed.
Getting rid of it would make the spec less confusing, but will of
course break things that depend on the current format (if there
are any).
What do others think?
The only snag is that right now you have to parse all sources.
Technically both the registry and proserver allow you do do:
http://www.ebi.ac.uk/das-srv/genomicdas/das/sources/
eqtl_rat_cis_fat
But IIRC I didn't include this in the spec to keep things simple.
If this isn't specified yet, how about allowing:
http://www.ebi.ac.uk/das-srv/genomicdas/das/eqtl_rat_cis_fat/sources
?
Then it's possible to stick with the model of passing a single
URI around to refer to a "DAS datasource", and stick a command on
the end of it to get the data you're after.
Well, the reason we didn't use this format is simply that it
doesn't "read" well, if only because "sources" is plural. What
would perhaps make sense, and which would allow for quickly
'pinging' a source for other similar uses, is to use this URL
format:
http://www.ebi.ac.uk/das-srv/genomicdas/das/eqtl_rat_cis_fat
Again, this is what seems most 'sensible' to me but I am happy to
go with the consensus.
_______________________________________________
DAS mailing list
[email protected]
http://lists.open-bio.org/mailman/listinfo/das
Jonathan Warren
Senior Developer and DAS coordinator
[email protected]
Ext: 2314
Telephone: 01223 492314
--
The Wellcome Trust Sanger Institute is operated by Genome
ResearchLimited, a charity registered in England with number
1021457 and acompany registered in England with number 2742969,
whose registeredoffice is 215 Euston Road, London, NW1 2BE.
Jonathan Warren
Senior Developer and DAS coordinator
[email protected]
Ext: 2314
Telephone: 01223 492314
--
The Wellcome Trust Sanger Institute is operated by Genome Research
Limited, a charity registered in England with number 1021457 and a
company registered in England with number 2742969, whose registered
office is 215 Euston Road, London, NW1 2BE.
_______________________________________________
DAS mailing list
[email protected]
http://lists.open-bio.org/mailman/listinfo/das
