BaseX-Talk

basex-talk@mailman.uni-konstanz.de

9 participants
5241 discussions

XQuery FLOWR and Context
by Dave Glick 18 Jun '10

18 Jun '10

Hello again, You guys do such a great job of answering and responding to questions and concerns on this board, thanks again for that Christian. This next one might be a case of misunderstanding the standard on my part, but I'm seeing some odd things with XQuery when using a different initial context than the document root. Take the following XML: <test> <a> <aa/> <ab> <aba/> <abb/> <abc/> </ab> <ac/> </a> <ba/> <bb> <bba/> <bbb/> <bbc/> </bb> <bc/> </test> Here is the result of several different XQuery commands with different initial contexts (I've loaded the document into the BaseX GUI and am setting initial context by double-clicking the element in the grid view)... Query: . Context: document, Result: The entire document (as expected) Context: , Result: The element and all its content (as expected) Query: let $v := . return $v Context: document, Result: The entire document (as expected) Context , Result: The entire document (was expecting the element and all its content!) Query: .//ab Context: document, Result: Result: The <ab> element and its content (as expected) Context , Result: Nothing (as expected) Query: let $v := //ab return $v Context: document, Result: The <ab> element and its content (as expected) Context: , Result: The <ab> element and its content (as expected) Query: let $v := .//ab return $v Context: document, Result: The <ab> element and its content (as expected) Context: , Result: The <ab> element and its content (was expecting nothing!) As you can see, it seems to behave counter to expectations when the initial context is not the document and I use the initial context in the FLOWR expression. I attempted to test in other tools, but it appears that setting the initial context for an XQuery expression is actually a difficult feature to find. It's not a totally academic question, though - the system we're building uses initial contexts pretty extensively. I know that context is not maintained in XQuery, but can the initial context not be used in variable assignment or other components of a FLOWR expression? Thanks, Dave

2 2

tablemem
by NewIntellectual 18 Jun '10

18 Jun '10

I've tried to figure out how to use the described 'tablemem' command ("Loading database table into main memory and speedup evaluation times"), unsuccessfully so far. Can this be invoked both in the BaseX GUI and in a client call to the server? Surely it would only make sense to make the call once after server startup.

2 1

default basex data directory
by szabolcs kurdi 18 Jun '10

18 Jun '10

Hi! I know that one can move the BaseXData directory from his or her home, but wouldn't this directory be better placed into something like "Application Data" (Windows), or ".BaseXData" on linux and friends by default? Is there a reason that the data directory is so "visible"? Maybe I have missed something here and that's how it should be - thanks! Szabolcs Kurdi

2 1

Leave whitespaces in XML when loading XML file with new CreateDB
by Jiří Kadlec 17 Jun '10

17 Jun '10

Hi Guys, How can I preserve whitespaces in XML document, when i'm using new CreateDB(..) object for creating new database from Java code. When I export data with new Export(..) I get document with choped whitespaces. I found out, that there is a possibility to preserve whitespaces when I create DB from BaseX gui. There is a possibility to uncheck option chop whitespaces which gives me required result. Is there any way how to do this from Java. Thank you Jiri

4 9

heterogeneous collections, filter notation, and/or add command
by C. M. Sperberg-McQueen 15 Jun '10

15 Jun '10

I'm trying to build a BaseX database containing the current contents of the XSD 1.1 test suite, and have run across a problem I suspect is easily soluble, but for which I am not finding the solution. The directory tree containing the database includes a large number of test cases, mostly files with names ending in .xml, and a smaller number of schema documents for the test cases, which mostly have names ending in .xsd. (One set of test contributions has three or four .xml files which are not in fact XML but which are testing the behavior of the processor in the presence of ill-formed input. I've renamed these so their names end in .xmlnwf, because they were causing attempts to index all the XML documents to fail.) There are also XML files with metadata for the test cases, some of which have names ending in .xml and some with names ending in .testSet. And there is some random stuff that I'm not particularly interested in indexing: some README files, a Microsoft Word document included by one organization contributing test cases, and (since I'm working under Mac OS X) some .DS_Store files containg OS-level binary metadata about the directory. There may be other files in the directories as well; I haven't tried to make a full census. If I could, I'd tell the database creation dialog to index everything it can parse as XML and ignore the rest. But when the parser encounters ill-formed input, it issues an error message and aborts the creation of the database. (And then the message disappears, so the user doesn't know which file caused the problem. If you could change that, it would be great, but really that's a separate issue.) Alternatively, I'd like to tell the creation dialog to use a filter that accepts either .xml files or .xsd files or .testSet files. But I can't figure out what notation to use for that. Since the example given ("*.xml") looks like a bash wildcard, I've tried "*.xml, *.xsd, *.testSet", and "*.xml,*.xsd,*.testSet" and "*.xml *.xsd *.testSet" and "*.{xml,xsd,testSet}", none of which match any files. Is there a way to specify a filter that matches several file extensions? Guided by http://www.inf.uni-konstanz.de/dbis/basex/commands I then tried add /home/xmlschema/xsts11/*/*.xsd in the command area in the GUI, but it tells me the command ADD is not known. (Perhaps this is because I'm still using 5.7; perhaps ADD came later?) I'm hoping wildcards are allowed in ADD commands, because the task of finding and adding fourteen thousand .xsd files would be a bit tedious. (I suppose I could use the command-line interface to do it.) It would also be nice to have the ADD functionality better integrated into the GUI. (How spoilt I have become! I can't believe I'm not using the command line, but the syntax checking in the query window is too helpful for me.) Any advice on (1) the notation of the filter parameter, (2) the use of wildcards in the ADD command, or (3) any other way to solve this task and build this database? I'm using the GUI interface to BaseX (odd; my name for the app makes clear I think it is version 5.7, but the About dialog box gives the number 12.2). I'll send this note, and then download and install 6.1 to see if it helps. -- **************************************************************** * C. M. Sperberg-McQueen, Black Mesa Technologies LLC * http://www.blackmesatech.com * http://cmsmcq.com/mib * http://balisage.net ****************************************************************

3 2

Oops
by NewIntellectual 15 Jun '10

15 Jun '10

A couple of corrections to my previous post. First, I realized after posting that my statement "including retrieval and processing of title information for the documents" isn't quite true. I forgot that I now lazy-load that particular metadata for each book, on demand (which is also fast.) However, it is definitely true that BaseX is very rapidly finding all of those hits, with retrieval of book-level metadata for each hit. Also, apologies to Christian Grün for misspelling his last name as Guen in my last post.

1 0

Experience with BaseX (reply)
by NewIntellectual 14 Jun '10

14 Jun '10

Erik - My own experience with BaseX 6.0+ has been extremely positive. I needed a fast, multi-user (reading) database that implemented not just XQuery, but XQuery with standard Full Text extensions. And not for a toy database, but for a database of historical documents, initially with about 200,000 pages of text with significant internal structure. I initially stress-tested BaseX for multiple users/queries and found a problem, which Christian Guen took seriously and rapidly fixed. (When I say stress-tested, I mean I really hammered it.) The same has been true for a few other glitches and questions. Overall, BaseX has been very solid and very fast. It is crazy-fast for the complete indexing that it does, and it is really fast for queries. Fast enough to find tens of thousands of hits in a structured XQuery+full text search of about 200,000 book pages in under 2 seconds, including retrieval and processing of title information for the documents. So I really like BaseX for its practical utility, and as a programmer I also appreciate that the overall approach to the system is clean and pretty rigorous. They set out to implement actual W3C standards for XQuery and XQuery with Full Text extensions, and that's what you get, plus some useful extensions. The included simple GUI is actually very useful, not just a toy. I place BaseX in distinct contrast to the older and apparently more established eXist system. My experience testing with it, over time as the system has evolved, has not been positive. It is cumbersome to set up - especially for indexing - and it does not follow W3C standards for XQuery Full Text but instead uses its own syntax, and if you look at the mailing list for it, it seems to have a large number of ongoing bugs. It really feels like a big, unstable hack. From a performance perspective, a much smaller subset of my full database began to seriously slow down queries. As far the future goes, my biggest hope is that BaseX becomes better at being a true writable database system. While it appears to handle standard XQuery Update without problems (which itself is impressive!), this is restricted to one writer at a time, as I understand it. I would not try to use BaseX currently as a general purpose writable database, but I also don't need it. In my view, there is a place for XML databases alongside use of standard relational systems. In that regard, I highly recommend the H2 database system if you need a clean, powerful, fast Java relational DBMS - it is greatly superior to, e.g. Derby. So it probably goes without saying, but I am a big fan of BaseX and I hope to see it grow in popularity and features, as it deserves. Phil Oliver Oliver Computing LLC

1 0

XQuery Update syntax
by NewIntellectual 14 Jun '10

14 Jun '10

As an FYI for those starting with XQuery Update, I've found this to be a short but very practical and useful guide: http://www.xmlmind.com/_tutorials/XQueryUpdate/index.html I've successfully used BaseX to apply large numbers of updates to a database containing roughly 200,000 pages-equivalent of text.

2 1

exists(<xpath>) returns empty, database corruption
by Swaminathan S 14 Jun '10

14 Jun '10

Hi, I am using BaseX version 6.03 using client server mode. I face 2 kinds of problems: 1) sometimes, BaseX returns nothing for exists(<xpath>) function (only true or false is expected) - in these cases, I execute some XQueries(XQuery-updates and XQueries) using XQuery API.This starts happening consistently for a few xpaths but not for all xpaths. Just to give you the scenario, my xupdates are always done such that I export the database after every database change. Also, I operate on only on one database(one xml file). 2) also sometimes, the database gets corrupted (some elements get moved beyond the end of the root element). I want to know if you have ever encountered this problem, the cause and if they have been fixed in newer versions. Thanks, Swaminathan Sahiram

4 4

Search term highlighting?
by NewIntellectual 09 Jun '10

09 Jun '10

Is there a ready way to perform search term highlighting in a fulltext query result? I found a reference to a 'ft:highlight' function at http://www.xmlmind.com/qizx/_distrib/docs/manual/fulltext_extensions.htmlbut this does not apparently exist in BaseX or as a standard function.

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

BaseX-Talk