CADAC/WebService

Google Code

Most of this work has moved to SimCat, a Google Code project.

Web Service

Design notes on the CADAC web service.

Protocols and Standards to Build On

Definitely

Maybe

Resources

{member}
A CADAC member.
{run}
Job on a computer, batch or interactive, simulation or analysis.
{tag}
Classic short tag, used to organize runs.
{computer}
Machine used to do the work.
{program}
Software doing the work.
{dataset}
Output of the run.

URIs and Operations

Canonical URIs - GET Returns XML

URI relative to http://cadac.sdsc.edu GET PUT POST DELETE
/xml/members list of {member}s
/xml/members/{member} {member} information add {member} modify {member} remove {member}
/xml/members/{member}/runs list of {member}'s runs
/xml/members/{member}/runs/{run} {run} information add {run} modify {run} remove {run}
/xml/members/{member}/runs/{run}/tags list of {run}'s {tags
/xml/members/{member}/runs/{run}/tags/{tag} add {tag} modify {tag} remove {tag}

XHTML Rendered - GET Returns XHTML

URI relative to http://cadac.sdsc.edu GET
/members list of {member}s
/members/{member} {member} information
/members/{member}/runs list of {member}'s runs
/members/{member}/runs/{run} {run} information
/members/{member}/runs/{run}/tags list of {run}'s {tags

Convenience & Querying URIs

URI relative to http://cadac.sdsc.edu GET
/members/{member}/tags list of {member}'s {tag}s
/members/{member}/runs/tag/{tag} list of {member's {run}s with {tag}
/members/{member}/runs/rss rss or atom feed of {member}'s {run}s
/members/{member}/runs/{run}/rss rss or atom feed of {run}
/members/{member}/runs/{tag}/rss rss or atom feed of {run}s with {tag}

Representation

Need a means to distinguish between XHTML and XML representations. Perhaps the default with GET will be XHTML, to support browsers, and XML for the PUT and POST operations. XML could be specified for GET using ?fmt=xml, or something similar. And the Atom or RSS feeds will be in their appropriate format.

Permissions

For, now, assume all GETs are world readable, and all other operations require authentication by {member}.

Implementation 0.1

Use Apache mod_rewrite and ProxyPass to map between the web server and the XML database. Use XSL to generate XHTML from native XML formats. Having the XML formats represented under something like /xml/members, allows /members/{member} to be rewritten to /xml/members/{member}?_xsl=/stylesheets/member.xsl. That way, eXist would build the XHTML. Likewise, RSS/Atom feeds will be generated using /{foo}/rss mapping to /xml/{foo}?_xsl=/stylesheets/rss.xsl.

Still need to deal with collection listings and {run}/tags, which is not a collection.

Implementation 0.2

I don't think I can build all of this using just rewrite rule. In particular, because we shouldn't make the client store all of the logic to deal with the database. The serving up the XHTML using a style sheet is simple and direct, but not handling POSTs and PUTs that update the database.

Components

  • pyfo ("Easily generate XML from Python data structures." True by all accounts.)
  • selector (Maps URI's to functions.)
  • resolver (selector dependency.)

selector example using the standard module wsgiref 

def say_hello(environ, start_response):
   start_response("200 OK", [('Content-type', 'text/plain')])
   return ["Hello %s!" %environ['selector.vars']['name']]

from selector import Selector
s = Selector()
s.add('/myapp/hello/{name}', GET=say_hello)
from wsgiref.simple_server import make_server
httpd = make_server('', 8000,s)
httpd.serve_forever()

pyfo example: 

import pyfo
doc = ('Run',('Computer','Cobalt',{'ivo-id':'http://www.ncsa.uiuc.edu/UserInfo/Resources/Hardware/SGIAltix/'}))
n = pyfo.pyfo(doc, pretty=True, prolog=True,encoding='utf8')
print n
<?xml version="1.0" encoding="utf8"?>
<Run>
  <Computer ivo-id="http://www.ncsa.uiuc.edu/UserInfo/Resources/Hardware/SGIAltix/">Cobalt</Computer>
</Run>