-<xlog server="chat.freenode.net" channel="#existdb" date="2008-02-03">

<message nick="DiZzZz" time="12:02:37.898Z">morning....</message>

<message nick="wolf77" time="12:02:55.512Z">Hi there</message>

<message nick="wolf77" time="13:29:37.867Z">Hi Pierrick</message>

<message nick="perig" time="13:30:24.601Z">hehe nice features coming up ? </message>

<message nick="wolf77" time="13:31:40.974Z">just the beginning of a feature ;-)</message>

<message nick="wolf77" time="13:32:57.659Z">Next step would be to optimize path expressions on very small node sets.</message>

<message nick="perig" time="13:32:59.07Z">yeah, looks nice though : it's always better to collaborate on something that is existing than on something that isn't :-)</message>

<message nick="perig" time="13:34:55.268Z">BTW, I've started to introduce generics into the code. Should I commit or wait a little bit in order to prevent problems on 1.2.1 ?</message>

<message nick="wolf77" time="13:35:44.806Z">I created that exist:no-index pragma months ago. Using it makes some real queries about 10x faster. However, to apply it automatically, the optimizer needs to know the number of descendant nodes that can be expected for a given element.</message>

<message nick="wolf77" time="13:36:22.176Z">Mmmh, I did not use generics yet, because it would make it more difficult to port changes back to 1.2.1.</message>

<message nick="wolf77" time="13:36:41.272Z">But well, at some point we have to face this ;-)</message>

<message nick="perig" time="13:37:06.647Z">OK, I'll wait then. I will try to work with the new features Ã the office.</message>

<message nick="wolf77" time="13:39:15.141Z">Ooops, forgot to commit some files...</message>

<message nick="wolf77" time="13:41:25.969Z">Another interesting idea would be to store type annotations in the element distribution graph... Well, I see lots of possibilities here.</message>

<message nick="perig" time="13:43:37.317Z">+1 for type annotations, but is it the right place ?</message>

<message nick="wolf77" time="13:44:54.692Z">it's not the right place for enforcing type checks, but it would help to choose better optimization strategies.</message>

<message nick="perig" time="13:45:15.871Z">for instance ?</message>

<message nick="wolf77" time="13:47:30.995Z">e.g. automatic creation of indexes</message>

<message nick="perig" time="13:48:22.885Z">back later : will go and dance a few breton traditional dances to please Maiwenn and, hopefully, myself</message>

<message nick="wolf77" time="13:48:38.036Z">sounds good. have fun ;-)</message>

<message nick="bkc" time="14:28:58.593Z">I'm looking for an opinion on a change to doGet in RESTServer.java</message>

<message nick="wolf77" time="14:29:23.097Z">what change?</message>

<message nick="bkc" time="14:29:37.806Z">I would like to be able to get xqueries to handle left-over path components in the url</message>

-<message nick="bkc" time="14:29:51.265Z">

so .. GET or POST to

<a href="http://somewhere/path/xquery/extra/stuff">http://somewhere/path/xquery/extra/stuff</a>

</message>

<message nick="bkc" time="14:30:05.154Z">would execute xquery and pass '/extra/stuff' as the remaining path to xquery as an arg</message>

<message nick="bkc" time="14:30:37.017Z">to do that, if getXMLResource returns null, the code could trim off trailing path components until it finds an xquery file</message>

<message nick="bkc" time="14:30:46.139Z">if it fails to find one, raises the NotFound as usual</message>

<message nick="wolf77" time="14:30:56.724Z">yes, that should be ok</message>

<message nick="wolf77" time="14:31:55.003Z">however, you don't need to pass an extra arg. the XQuery can find out how it has been called.</message>

<message nick="wolf77" time="14:32:14.259Z">by looking at path-info</message>

<message nick="bkc" time="14:32:15.561Z">yes, but then the xquery would have to know where it was located in the uri and parse the extra args itself</message>

<message nick="bkc" time="14:33:08.042Z">I am thinking of Python's WSGI, i heavily rely on PATH and SCRIPT_NAME in the environment </message>

<message nick="bkc" time="14:33:36.811Z">I guess I'd choose a suitable name and add that as a value in the request object?</message>

<message nick="wolf77" time="14:34:44.843Z">Mmmh, you could add it as an additional request parameter</message>

<message nick="bkc" time="14:35:37.835Z">but then, it would have to be something not likely to be passed as an http query argument.</message>

<message nick="bkc" time="14:35:47.789Z">I don't see any other way around it.. </message>

<message nick="wolf77" time="14:36:46.749Z">you could also declare it as an external variable, but that's not as clean as passing a request parameter.</message>

<message nick="bkc" time="14:37:39.27Z">my java-fu is low, but I'd expect HttpServletRequest to already have the concept of SCRIPT_NAME and PATH</message>

<message nick="wolf77" time="14:38:56.976Z">I guess those correspond to getServletContext() and getPathInfo(), but I don't know enough about WSGI.</message>

<message nick="bkc" time="14:38:59.9Z">yeah.. getPathInfo() is supposed to return the "extra path information"</message>

<message nick="bkc" time="14:40:09.745Z">I don't see any way to set that value</message>

<message nick="wolf77" time="14:41:01.355Z">eXist wraps its own class around the HttpServletRequest. You could overwrite the method there.</message>

<message nick="wolf77" time="14:41:35.111Z">see RequestWrapper, HttpRequestWrapper</message>

<message nick="bkc" time="14:43:34.693Z">so, I could create a private variable in the HttpRequestWrapper class and a setter for it, initialize it with serveletRequest.getPathInfo()</message>

<message nick="bkc" time="14:44:54.451Z">ok, thanks</message>

<message nick="bkc" time="15:34:55.818Z">what compiler compliance level should I use w/i eclipse? 1.3, 1.4, 5.0 or 6? (default is 6 and I have tons of warnings). Build.properties has 1.5 </message>

<message nick="bkc" time="15:35:51.423Z">1.5 isn't an option for me in eclipse settings.. hmm</message>

<message nick="bkc" time="15:37:45.749Z">w/ 5.0, I get 38 errors and 4253 warnings</message>

<message nick="bkc" time="15:37:57.494Z">I'll just stick to ant to build</message>

<message nick="perig" time="15:38:27.225Z">Mmmh.... Ecli^pse should definitely able to build. What are the errors ?</message>

<message nick="bkc" time="15:38:52.649Z">most are in xmldbRealm.java, line 81.. debug cannot be resolved</message>

<message nick="bkc" time="15:39:05.927Z">GenericPrincipal cannot be resolved to type.. </message>

<message nick="bkc" time="15:39:14.639Z">I had some problems w/ the libs, maybe that's the cause</message>

<message nick="bkc" time="15:39:29.539Z">you can tell I'm not a java person!</message>

<message nick="bkc" time="15:40:17.304Z">the build path in eclipse referenced a bunch of libs that didn't exist, I tried to update the version numbers on as many as I could find.</message>

<message nick="perig" time="15:41:03.617Z">That probably explains then</message>

<message nick="bkc" time="15:41:18.538Z">yeah, I had an svn conflict w/ .classpath file. </message>

<message nick="perig" time="15:47:21.661Z">org.apache.catalina.realm.RealmBase is the base class you need</message>

<message nick="bkc" time="15:47:36.09Z">I reverted .classpath and .project, now I'm down to 16 errors, 5100 warnings. most errors in xslfomodule.java fopFactory cannot be resolved</message>

<message nick="perig" time="15:49:47.439Z">yeah : it's downloaded from and external process (but is not required)</message>

<message nick="perig" time="15:50:10.15Z">warnings are due to the fact that th code is still very 1.4</message>

<message nick="bkc" time="15:50:40.01Z">works fine with ant.. so I'm not going to worry</message>

<message nick="perig" time="15:51:30.609Z">I usually worry when I encounter such problems:-)</message>

<message nick="perig" time="15:51:53.68Z">I really want eXist to work out of the scratch on a standard Eclipse install</message>

<message nick="bkc" time="15:53:25.164Z">I have added 2 methods to HttpRequestWrapper. Now when changing doGet in RESTServer, the 'request' param is HttpServletRequest, shouldn't it be HttpRequestWrapper?</message>

<message nick="bkc" time="16:30:43.514Z">I am getting failures in org.exist.http.RESTServiceTest, anyone else? </message>

<message nick="bkc" time="16:30:57.894Z">6 failures, even in code I haven't broken yet. :-(</message>

<message nick="bkc" time="16:35:52.804Z">nevermind.. rtfm</message>

<message nick="bkc" time="16:50:58.765Z">oh, looks like I *had* broken the code. junit and mylar..great stuff</message>

<message nick="perig" time="16:56:49.471Z">fromEclipse, I just rirght-click on the desired test</message>

<message nick="bkc" time="17:38:41.573Z">when using junit w/i eclipse, where can I find the exist.log file generated. it's using server.xml and conf.xml, but I can't see where the logs do</message>

<message nick="perig" time="17:41:11.927Z">WEB-INF, as usual</message>

<message nick="bkc" time="17:41:36.801Z">I looked there, but the logs are old.</message>

<message nick="bkc" time="17:42:18.958Z">I think it uses that when running ant based test, but not from junit?</message>

-<message nick="bkc" time="18:54:34.527Z">

Function request:getPathInfo() is not defined in module namespace:

<a href="http://exist-db.org/xquery/request">http://exist-db.org/xquery/request</a>

</message>

<message nick="bkc" time="18:54:52.04Z">where is the declaration for xquery methods defined?</message>

<message nick="bkc" time="18:57:22.324Z">nevermind</message>

<message nick="bkc" time="20:17:51.734Z">in REST doGet checks both the mime type and binary status of xquery objects. Shouldn't the mime type test be sufficient?</message>

<message nick="bkc" time="20:36:46.044Z">doGet is declared to get HttpServletRequest, but it's really a wrapper, I can't cast it to HttpRequestWrapper in RESTServer, so how do I call the new methods on it?</message>

<message nick="bkc" time="20:39:02.56Z">shoot, nevermind. declareVariables.. hmm</message>

<message nick="bkc" time="20:57:54.507Z">green bar! woo!</message>

<message nick="bkc" time="21:50:19.658Z">do I email my patch or post it in a tracker somewhere?</message>

<message nick="bkc" time="21:51:25.238Z">nevermind.</message>

<message nick="bkc" time="21:58:51.223Z">submitted 1885912</message>

<message nick="wolf77" time="22:18:43.753Z">Thanks :-)</message>

<message nick="wolf77" time="22:19:20.513Z">so it looks like you solved all problems?</message>

<message nick="bkc" time="22:38:26.629Z">well, other than learing to read before asking questions</message>

<message nick="bkc" time="22:38:34.47Z">learning..</message>

</xlog>