-<xlog server="chat.freenode.net" channel="#existdb" date="2008-03-25">

<message nick="Leela" time="10:27:20.729Z">Good Afternoon</message>

<message nick="Leela" time="10:28:53.935Z">Hi perig_</message>

<message nick="Leela" time="10:29:49.961Z">any one there?</message>

<message nick="Leela" time="11:13:59.73Z">sorry the channel didn't work for a long time</message>

<message nick="Leela" time="11:14:28.96Z">perig_ I have few doubts regarding the XSLT Processor</message>

<message nick="Leela" time="11:22:40.467Z">Hi Perig_</message>

<message nick="perig_" time="11:22:42.15Z">well : I'm sorry, I have to go. I'll be back in a little bit more than 1 hour</message>

<message nick="Leela" time="11:23:04.16Z">ok will be after an hour or so</message>

<message nick="perig_" time="11:23:09.216Z">However, feel free to explain your problem : I'll read the logs when I'm back ;-)</message>

<message nick="Leela" time="13:37:09.401Z">hi perig_</message>

<message nick="perig_" time="14:03:21.917Z">hi leela</message>

<message nick="Leela" time="14:31:20.507Z">sorry was away for some work</message>

<message nick="Leela" time="14:31:31.862Z">i require some clarification with XSLT Processor</message>

<message nick="Leela" time="14:34:57.371Z">is the proposal to implement the XSLT processor all fresh, or planning to use any open source XSLT Processor and use that for our purpose</message>

<message nick="Leela" time="14:37:09.851Z">hi perig_</message>

<message nick="Leela" time="14:37:23.56Z">sorry there seems to be some problem with my browser today</message>

<message nick="hircus" time="15:22:13.633Z">are the Javadocs for eXist available for download anywhere?</message>

<message nick="wolf77" time="15:34:06.912Z">if you have the source installed, called "build.sh javadocs"</message>

<message nick="hircus" time="15:34:30.854Z">wolf77: thanks</message>

<message nick="hircus" time="15:40:51.321Z">wolf77: hm, it only created documentation for the fluent extension?</message>

<message nick="wolf77" time="15:41:40.117Z">mmh, no, there should be documentation for all classes. Check webapp/api now</message>

<message nick="hircus" time="15:42:22.335Z">ah! silly me, I was looking for a 'javadoc' folder</message>

<message nick="hircus" time="15:43:28.274Z">Netbeans is saying it does not have index-files folder, or index-all.html file -- I can just presumably symlink index.html to index-all.html ?</message>

<message nick="wolf77" time="15:44:16.219Z">not sure what this means. I never used netbeans with the javadocs</message>

<message nick="hircus" time="15:45:06.588Z">seems to work when I symlink -- basically Netbeans does not consider a folder / JAR to be a valid javadocs tree if it does not have index-all.html</message>

<message nick="hircus" time="15:45:33.334Z">(is that part of the Javadoc spec? in which case it'd be nice to have eXist's index.html renamed)</message>

<message nick="hircus" time="15:47:24.085Z">looking at another javadoc archive I have (for JGraphT) and.. interesting. index.html is like eXist's (small in size). index-all is huge</message>

<message nick="hircus" time="15:48:02.557Z">so not sure. I'm just going to 'touch' the file to make sure it exists for now</message>

<message nick="wolf77" time="15:48:37.544Z">check build/scripts/build-impl.xml</message>

<message nick="wolf77" time="15:48:51.373Z">in line 391, set noindex="false"</message>

<message nick="wolf77" time="15:49:00.706Z">I guess this should generate an index-all</message>

<message nick="wolf77" time="15:49:30.677Z">probably because index-all.html is so large?</message>

<message nick="hircus" time="15:50:49.313Z">fair enough. thanks for your help!</message>

<message nick="hircus" time="15:51:12.565Z">3.9M .. decent size</message>

<message nick="_delirium" time="17:29:44.785Z">wolf77: when i check out trunk and then run build.sh - i am missing jar's for Validation, GeoSpatial Index and XSLFO module - is this normal? its ages since i have done a fresh checkout!</message>

<message nick="wolf77" time="17:30:13.794Z">yes, that's correct</message>

<message nick="wolf77" time="17:30:24.489Z">we discussed this a while ago</message>

<message nick="_delirium" time="17:34:19.868Z">perig: there is an app for xslt 2.0</message>

<message nick="wolf77" time="17:37:20.664Z">not really an application though</message>

<message nick="_delirium" time="17:37:24.829Z">wolf77: I changed build.properties in extensions/indexes but the GeoSpatial jars are not downloaded, also I cannot fund an option for the jars needed for Validation (which is core!?)</message>

<message nick="perig" time="17:38:43.757Z">asking for details...</message>

<message nick="wolf77" time="17:39:43.788Z">_delirium: I don't know. The build system works for me. No problems with the spatial jars once you enabled them.</message>

<message nick="_delirium" time="17:40:00.581Z">its the indexes setting in build.properties and conf.xml?</message>

<message nick="wolf77" time="17:40:17.033Z">extensions/indexes/build.properties</message>

<message nick="wolf77" time="17:40:37.822Z">back to debugging ...</message>

<message nick="_delirium" time="17:42:06.652Z">hmm still cant get it to work here! weird!</message>

<message nick="_delirium" time="17:42:12.29Z">will try checking out again</message>

<message nick="_delirium" time="17:43:59.836Z">perig: do you know if eXist supports "declare copy-namespaces preserve, inherit"</message>

<message nick="perig" time="17:44:30.878Z">_delirium : it should</message>

<message nick="perig" time="17:45:06.557Z">the "only" problem is no-inherit as far as I remember ; it would need substantial changes to our serializer...</message>

<message nick="_delirium" time="17:45:12.625Z">perig: okay a Steve Harris from Oxford uni is reporting that he cant set the media-type from an XSLT which is called using transform:transform from XQuery</message>

<message nick="_delirium" time="17:46:25.023Z">he also reported that he tried to use declare copy-namespaces but it made no difference and so I was wondering is we supported it?</message>

<message nick="perig" time="17:47:15.291Z">Adam, please run the XQTS or read the results on the W3C site : the answer is there :-)</message>

<message nick="perig" time="17:50:30.006Z">but at first glance, I'd say that we should have a way to pass our serialization options to the XSLT processor</message>

<message nick="_delirium" time="17:51:34.543Z">perig: i think its more in the other direction - i need to capture the xsl:output and set</message>

<message nick="_delirium" time="17:51:46.84Z">and set the HttpResponse from the RESTServer/XQuery</message>

<message nick="_delirium" time="17:52:22.939Z">just looking into that possibility now - once i have checked out from svn ;-)</message>

<message nick="wolf77" time="17:52:36.257Z">no: if transform:transform is used inside an XQuery, the serialization settings of the XQuery should be applied</message>

<message nick="wolf77" time="17:52:55.473Z">transform:stream-transform will use the XSLT's options.</message>

<message nick="_delirium" time="17:55:45.885Z">wolf77: sorry - I am meaning transform:stream-transform, Steve claims it doesnt work - I was going to take a look and if it doesnt try and fix it</message>

<message nick="wolf77" time="17:56:06.583Z">ah, ok, stream-transform is a different matter</message>

<message nick="_delirium" time="17:58:06.526Z">will take a look - dunno what i did wrong a minute ago, have the GeoSpatial jars now - but still i have no idea where to get the Validation Jars??? needs something called Jing? from com.thaiopensource ?</message>

<message nick="perig" time="17:58:26.283Z">my consideration was more general : we should have a way to instruct the processor (say an option in the xslt namespace). </message>

<message nick="wolf77" time="17:59:35.377Z">as for parameters</message>

<message nick="wolf77" time="18:00:17.584Z">you have to ask DiZzZz for the jing thing</message>

<message nick="perig" time="18:01:20.599Z">Dizzz forgot to add this one</message>

<message nick="_delirium" time="18:02:20.521Z">thanks :-)</message>

<message nick="perig" time="18:03:00.305Z">Feel free to add it : my .classpath is definitely too hacked ;-)</message>

<message nick="wolf77" time="18:03:05.682Z">Hurrrraaaaaay</message>

<message nick="wolf77" time="18:03:33.41Z">I found the source of the corruption: a node id computation error</message>

<message nick="perig" time="18:03:39.588Z">corupption fixed ?</message>

<message nick="perig" time="18:06:01.831Z">Didn't you say that it worked differently in single/multiple thread ? How a node ID computation could interfer then ? Well... let's see the patch</message>

<message nick="wolf77" time="18:06:33.394Z">quite simple: the problem appears if you are inserting nodes into a gap in the tree, say 1.5.6 is the first child of 1.5 (1.5.1 to 1.5.5 were removed previously). eXist will fill up the available node ids until it reaches 1.5.6 ---- and continues filling, thus duplicating 1.5.6, 1.5.7 and so on.</message>

<message nick="wolf77" time="18:07:22.972Z">this means you end up with duplicate ghost ids for 1.5.6 and following, which completely corrupts the db.</message>

<message nick="perig" time="18:07:46.233Z">sounds very similar to the one we were experiencing with the XQTS</message>

<message nick="perig" time="18:08:16.945Z">(with attributes as particular children)</message>

<message nick="wolf77" time="18:09:10.558Z">well, it took more than a day to discover the cause of the error. It first looked like a dom.dbx page split issue. Only when I printed out all the pages did I see the duplicate ids.</message>

<message nick="wolf77" time="18:10:21.174Z">now I just need to find and remove all the debug code I added... sigh!</message>

<message nick="_delirium" time="19:19:12.014Z">wolf77: hmm Steve Harris was right - stream-transform doesnt set the media-type correctly</message>

<message nick="wolf77" time="19:52:21.336Z">just found another node id corruption :-(</message>

<message nick="wolf77" time="19:53:11.414Z">I´m reaching the end of my power, sigh</message>

<message nick="_delirium" time="19:54:59.731Z">wolf77: just committed a fix to stream-transform()</message>

<message nick="DiZzZz" time="21:16:38.925Z">build is broken</message>

<message nick="DiZzZz" time="21:18:18.159Z">well, I guess this is a pure "I build only in my IDE" issue</message>

<message nick="DiZzZz" time="21:18:35.843Z">(reading his Jing 'issue')</message>

<message nick="perig" time="21:26:00.357Z">should I commit a patch that would reallow compilation ?</message>

<message nick="DiZzZz" time="21:28:55.365Z">I sms-ed him, I guess he is nervous right now :-)</message>

<message nick="DiZzZz" time="21:29:29.742Z">patch is simple... extend HttpResponseWrapper</message>

<message nick="DiZzZz" time="21:29:41.858Z">I guess he forgot to extend 2 classes</message>

<message nick="DiZzZz" time="21:29:51.491Z">and one interface</message>

<message nick="perig" time="21:30:01.855Z">sure. I wonder what to do with the Cocoon one though</message>

<message nick="DiZzZz" time="21:31:08.396Z">hmmmmm</message>

<message nick="DiZzZz" time="21:31:26.08Z">well adam says it compiled for him...</message>

<message nick="perig" time="21:32:54.309Z">one thing is sure, the test suite couldn't pass...</message>

<message nick="DiZzZz" time="21:33:03.362Z">can not believe it....</message>

<message nick="DiZzZz" time="21:33:09.911Z">test suite.... :-)</message>

<message nick="perig" time="21:33:38.77Z">Yeah : I still wonder why I run it before each commit.</message>

<message nick="DiZzZz" time="21:33:47.201Z">perig: feeling better btw?</message>

<message nick="DiZzZz" time="21:34:07.489Z">he confirms... forgot to commit :-)</message>

<message nick="DiZzZz" time="21:34:26.274Z">perig: you have been ill a few times recently...</message>

<message nick="DiZzZz" time="21:34:28.898Z">right!</message>

<message nick="DiZzZz" time="21:34:35.257Z">aaaaaah mr adam :-)</message>

<message nick="perig" time="21:34:58.198Z">Dizzzz : indeed, I'd like this series to stop :-)</message>

<message nick="perig" time="21:35:48.977Z">worst thing is that each time one is ill here, the 3 other ones will be ill the next week</message>

<message nick="DiZzZz" time="21:37:02.727Z">hmmmm buy a big box of multi-vitamins?</message>

<message nick="DiZzZz" time="21:37:20.291Z">one pill a day keeps the doctor away :-)</message>

<message nick="perig" time="21:37:36.002Z">glups</message>

<message nick="DiZzZz" time="21:38:54.599Z">_delirium: what went wrong?</message>

<message nick="_delirium" time="21:44:11.609Z">DiZzZz: hey thanks for the text</message>

<message nick="DiZzZz" time="21:45:12.262Z">at ASML I have to buy cake if I break the build :-) :-) :-) :-) :-) :-) :-) </message>

<message nick="_delirium" time="21:45:50.003Z">DiZzZz: wow wouldnt mind, but well i think it may rot before it reaches you in the post</message>

<message nick="DiZzZz" time="21:46:00.327Z">hehehehheehehehhe</message>

<message nick="wolf77" time="21:46:39.72Z">DiZzZz: I'm not sure you can call a stored XQuery via SOAP.</message>

<message nick="wolf77" time="21:47:55.824Z">at least I can't remember a way to do it ;-)</message>

<message nick="_delirium" time="21:48:19.997Z">SOAPServer ;-)</message>

<message nick="DiZzZz" time="21:48:47.794Z">adam.... why in cocoon response.setHeader("Content-Type", contentType); and in the other setContentType? for a reason or by accident?</message>

<message nick="_delirium" time="21:50:46.606Z">DiZzZz: Cocoons reponse doesnt have a convenient setContentType method</message>

<message nick="perig" time="21:50:55.428Z">cocoon protects the content type, which is determined by the serialization policy</message>

<message nick="DiZzZz" time="21:51:25.989Z">_delirium: ah ok. perig oh ok?</message>

<message nick="_delirium" time="21:51:53.827Z">I assume then it will be overwritten</message>

<message nick="perig" time="21:52:21.435Z">that's what I meant when I said "I wonder what to do with the Cocoon one though"</message>

<message nick="DiZzZz" time="21:53:19.473Z">wolf77: I am reading the book on RESTfull webservices (O'reilly) ; the author(s) nice write about doing complex things with soap while it could be done sooooo much easier</message>

<message nick="DiZzZz" time="21:53:46.34Z">uber-abstracting the real subjext</message>

<message nick="wolf77" time="21:53:51.026Z">yeah, sure. I really can't see why I should use SOAP at all.</message>

<message nick="DiZzZz" time="21:55:40.014Z">well for satisfying managers that already paid too much on .net $$</message>

<message nick="DiZzZz" time="21:56:02.204Z">well nice book, i'll spend some hours on it</message>

<message nick="DiZzZz" time="21:56:18.837Z">gents, good night........</message>

<message nick="DiZzZz" time="21:56:22.652Z">sleep tight</message>

<message nick="DiZzZz" time="21:56:27.108Z">wrong or right</message>

<message nick="wolf77" time="21:56:27.128Z">I think I already read parts of it on safari</message>

<message nick="DiZzZz" time="21:56:39.495Z">wolf77: yeah the book is there</message>

<message nick="wolf77" time="21:56:47.776Z">today I'll sleep tight --- after fixing the second corruption today</message>

<message nick="DiZzZz" time="21:56:59.002Z">chapter 4 is interesting........ (for me as n00b)</message>

<message nick="perig" time="21:57:09.626Z">night dizzz</message>

<message nick="DiZzZz" time="21:57:15.464Z">zzzzzzzzzzzzzzzzzzzzzzzzzz</message>

<message nick="perig" time="21:58:22.805Z">so, the 2nd corruption is fixed ?</message>

<message nick="wolf77" time="21:59:08.517Z">yeah, it falls under the same category as the first one</message>

<message nick="wolf77" time="21:59:26.332Z">just using a different node insertion method</message>

<message nick="perig" time="22:00:44.728Z">I really look forward to examine the patch : spent so much time lat year to track a persistent DOM issue...</message>

<message nick="perig" time="22:01:25.253Z">But, well, that's an excellent news</message>

<message nick="wolf77" time="22:01:41.986Z">Yeah, it takes a damn lot of time</message>

<message nick="wolf77" time="22:02:29.951Z">the only chance to find the bug was to record the whole HTTP traffic between application </message>

<message nick="wolf77" time="22:02:30.241Z">and database, then replay the whole story.</message>

<message nick="wolf77" time="22:23:21.356Z">ah, looks like we even have a hidden node id problem in one of the xupdate test cases</message>

<message nick="wolf77" time="22:26:40.557Z">I extended NativeBroker.checkNodeTree with a few more checks, e.g. to test the sibling </message>

<message nick="wolf77" time="22:26:40.557Z">and parent-child relation between a pair of consecutive nodes</message>

<message nick="wolf77" time="22:37:27.015Z">ok, I'm committing my changes so far. Test suite should be ok unless you enable </message>

<message nick="wolf77" time="22:37:27.045Z">consistency checks.</message>

<message nick="michy80" time="22:40:34.88Z">i'm interested to improve index-support for google summer of code</message>

<message nick="michy80" time="22:40:58.142Z">can i speak with someone?</message>

<message nick="michy80" time="22:42:04.662Z">index support to order by it can improve with B+tree</message>

<message nick="michy80" time="22:43:33.072Z">in a data manipulation like create, update and delete, it have to update the b+tree index</message>

<message nick="wolf77" time="22:44:35.387Z">yes. we have a modularized framework for new indexes, which makes it easier to implement </message>

<message nick="wolf77" time="22:44:35.417Z">those operations.</message>

<message nick="michy80" time="22:45:28.248Z">what have to do a google SoC student?</message>