Re: conf.xml

> Is a schema/DTD for the config file available? I'd like to know what
> settings are optional. To kick things off:
>
> * What happens after the max number of connections as specified in
> <pool/> is reached? No more connections? Does this also apply to
> embedded deployments? What happens if I omit the <pool/> declaration?

I attached my schema file for conf.xml, but I wrote it some time ago, so it 
might not be up to date. The <pool> directive specifies the number of storage 
backend instances available to process requests. Backend instances are shared 
between all threads accessing the database. A backend instance is retrieved 
for each single action and released afterwards. It is thus ok to have less 
backend instances than client threads. You need at least two backend 
instances during the initialization phase of the database.

> * What happens if I omit  <at> index-deth from <indexer/>? What happens if I
> omit the <indexer/>?

The  <at> index-depth setting is described in the docs. There are default settings 
for all configuration entries, so if you remove <indexer/>, default settings 
will be applied.

> * What happens if I omit the <log4j:configuration/>

It doesn't matter where you define the log4j configuration. Basically, the 
last configuration wins. So if you have an external log4j configuration (e.g. 
via JBoss), you can just delete the <log4j:configuration/> section from 
conf.xml and define the settings for eXist somewhere else.

Wolfgang

Re: conf.xml

On Mon, Sep 01, 2003 at 12:59:49PM +0200, Wolfgang Meier wrote:

> I attached my schema file for conf.xml, but I wrote it some time ago, so it 
> might not be up to date. The <pool> directive specifies the number of storage 
> backend instances available to process requests. Backend instances are shared 
> between all threads accessing the database. A backend instance is retrieved 
> for each single action and released afterwards. It is thus ok to have less 
> backend instances than client threads. You need at least two backend 
> instances during the initialization phase of the database.

OK.

> > * What happens if I omit the <log4j:configuration/>
> 
> It doesn't matter where you define the log4j configuration. Basically, the 
> last configuration wins. So if you have an external log4j configuration (e.g. 
> via JBoss), you can just delete the <log4j:configuration/> section from 
> conf.xml and define the settings for eXist somewhere else.

OK, but I'm going to run this embedded in an tomcat-hosted servlet,
non-j2ee. So I reckon I'd still need this.

Would it be possible to have <db-connection files="$databaseroot"/>
<db-connection files="$exist.home"/> ? It would make moving
around/automatically instantiating my (embedded) database easier.

Thanks,
--

-- 
Emiliano

Re: Logging and exist

Outgoing mail is certified Virus Free.
Checked by AVG Anti-Virus (http://www.grisoft.com).
Version: 7.0.167 / Virus Database: 259.11.3 - Release Date: 19/08/2003

Nested queries

Re: Nested queries

Hi,

> When performing a nested query using the LocalXPathQueryService and
> eXist 0.9.2, I get a NPE:
>
> java.lang.NullPointerException
>         at
> org.exist.xmldb.LocalXPathQueryService.query(LocalXPathQueryService.java
> :92)

The call to LocalXMLResource.getNode() in line 83 of 
org.exist.xmldb.LocalXPathQueryService probably returns null, but the 
following code assumes it is a valid object. This should be checked.

However, if getNode() returns null, the XMLResource is probably not a valid 
query result node. Could you check if you are passing the right object (one 
obtained from a ResourceSet)? Otherwise, could you send me the corresponding 
fragment of your code?

Regards,

Wolfgang

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Xpoint

Hi,

Ok. I fixed XInclude. Unfortunately, you will have to wait until tomorrow for 
an updated CVS version. There have been lots of changes over the past days 
and I have to make sure that the CVS code compiles.

Wolfgang

On Thursday 28 August 2003 04:44 pm, Giulio Valentino wrote:
> Hi Meier,
> after your last modification in to the cvs,  xpoint of Xinclude is not
> working correctly
>
> int this example
> <?xml version="1.0" encoding="ISO-8859-1" ?>
> <nodo>
> <xi:include href="/db/legislazione/anno2003/decreto.xml#xpointer
> (//intestazione | //parent::articolato/articolo[ <at> id='art1'])"
> xmlns:xi="http://www.w3.org/2001/XInclude" />
> </nodo>
>
> it comes back  null
>
> Could you check it
>
> G. Valentino
>
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> Exist-open mailing list
> Exist-open <at> lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/exist-open

--

-- 
Wolfgang Meier
Darmstadt University of Technology
Dpt. of Sociology
Residenzschloss         D-64287 Darmstadt
Tel.: +49 (0) 06151 16 4458

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Logging and exist

Hi,

I'm using Jboss 3.2.1 and I'm having some problems with eXist (0.9.2)
and Jboss. The configuration for logging messages that comes with eXist
seems to badly interact with jBoss. 

For example, when eXist reads its conf.xml file, the following message
appears:

	ERROR: invalid console appender config detected, console stream
is looping

Then, no more output is displayed on the console, and worse, everything
goes into the exist.log file even Jboss' own output. All the messages
related to Jboss are written in WEB-INF/logs/exist.log.

I tried to change/remove the log4j configuration in conf.xml, but it
always interact with Jboss. If I remove the log4j part in the conf.xml
file, Jboss turns its logging level to DEBUG generated a lot of
messages.

I have carrefully deactivated Cocoon to be sure this is not coming from
Cocoon. With or without Cocoon, I'm still having the same problem. The
sole difference is that, with Cocoon, additional log files are generated
in WEB-INF/logs.

Any idea?

Regards.

Olivier.

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problems accessing CVS updates

Latest news said that sourceforge is currently migrating to a new CVS 
infrastructure. According to sf, the new CVS should be available within the 
next days.

Wolfgang

On Saturday 30 August 2003 03:02 am, Leo wrote:
> Hi folks
>
> I am still experiencing problems accessing the
> CVS
>
> Leo

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

CVS update: new XPath parser

The new XPath parser is now integrated into the CVS version. This is a 
complete rewrite of the XPath language parser, which replaces the old 
implementation. The new implementation is much closer to the XPath spec (in 
fact, it is more or less an exact copy of the XPath 1.0 subset of the EBNF 
grammar given in the 2.0 spec).

The new implementation uses two phases to analyze an XPath expression: first, 
a syntax tree is created from the input XPath string. Second, the tree parser 
walks through the generated tree to generate an executable representation. 
This results in much cleaner code and it is easier to resolve ambiguous 
expressions. Previous problems with "document" or "text" occurring as 
function as well as element names should be solved.

The new parser currently implements all of XPath 1.0 except the following:

* axis specifiers preceding::, following:: and namespace::
* translate() function
* variable references ($var)
* node type specifiers comment() and processing-instruction()

The missing features will be implemented soon.

Some things that are now possible:

* search with namespace-prefix and wildcard: //dc:*, //*:subject
* reference another document: 
collection("/db/shakespeare")//SPEECH[SPEAKER=document()/test/speaker]
* numerical operators
* //div[lang("en")]

Things that changed:

The expression document(*) is no longer valid. Instead, use document() without 
parameters to select all documents in the database.

Regards,

Wolfgang

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: CVS update: new XPath parser

I also uploaded a new snapshot release:

http://prdownloads.sourceforge.net/exist/eXist-snapshot-030903.zip?download

Wolfgang

-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf