BaseX-Talk September 2019

basex-talk@mailman.uni-konstanz.de

17 participants
23 discussions

Full-text index on mixed content
by Schopper, Daniel 23 Sep '19

23 Sep '19

Dear all, chatting after a session of the ongoing TEI conference ( https://graz-2019.tei-c.org) I was asked about plans to support fulltext indexes on mixed content nodes in BaseX – I did not know of any, so I wanted to pass the question on to this list: Is there a plan to implement this feature in the near (or not-so-near) future? If not, did somebody of the core devs estimate the effort to get this done? (needless to say that it would an awsome feature to have in BaseX ;-) Thanks in advance & best Daniel (just being curious) -- Mag. Daniel Schopper Austrian Centre for Digital Humanities Working Group Data, Resources and Standards Austrian Academy of Sciences Sonnenfelsgasse 19 1010 Vienna T: +43 1 51581-2216 E: daniel.schopper(a)oeaw.ac.at

2 1

Suggestion for concurrency of RestXQ
by Marco Lettere 23 Sep '19

23 Sep '19

Dear all, on one of our applications based on RestXQ services running on 9.0.2 we experience a strange behaviour related to concurrency. It is a stateless service running on a server with http-local set to true. So no database access. We removed all accesses to sessions and the code doesn't even write to files or other external resources. It's really only computation. When I run multiple requests, I can clearly see that they are accepted simultaneously (up to 8 which matches PARALLEL). They even complete out of order according to the session id reported in the log. [1] The strange thing is: performance seems to be as if calls were executed serially. If T is the time to complete one call then the pattern is T + T + T + T + T. I have to add that the algorithm is rather complex. Lots of functions (possibly inlined, private and the main one even recursive) which filter and transform a large XML request into a possibly large XML output. If I simplify the called RestXQ to something like [2] then I get a log like [3] which shows a pattern like the one I'm expecting T+d+d+d+d where d is just a slight overhead (orders of magnitude smaller). My question is what do I have to check in order to understand where the hard synchronization of the calls occurs? Might this be related to non-deterministic functions? What else? I've tried to isolate parts of the code but I'm unable to get predictable results. Thank you very much for all the hints you will be able to provide! Regards, Marco. [1] 17:10:19.204 127.0.0.1:58038 admin REQUEST [POST] http://localhost:8081/rate-selector/select-matching-rates 17:10:19.204 127.0.0.1:58040 admin REQUEST [POST] http://localhost:8081/rate-selector/select-matching-rates 17:10:19.205 127.0.0.1:58042 admin REQUEST [POST] http://localhost:8081/rate-selector/select-matching-rates 17:10:19.205 127.0.0.1:58044 admin REQUEST [POST] http://localhost:8081/rate-selector/select-matching-rates 17:10:19.208 127.0.0.1:58046 admin REQUEST [POST] http://localhost:8081/rate-selector/select-matching-rates 17:10:21.325 127.0.0.1:58040 admin 200 2121.01 ms 17:10:23.174 127.0.0.1:58046 admin 200 3965.74 ms 17:10:25.035 127.0.0.1:58038 admin 200 5830.63 ms 17:10:26.889 127.0.0.1:58044 admin 200 7683.76 ms 17:10:28.739 127.0.0.1:58042 admin 200 9534.46 ms [2] (:~ : Test concurrency :) declare %rest:path("/test") %rest:POST("{$body}") function rs:test($body as node()) { <a>{$body}{fold-left(1 to 10000000, 0, function($acc, $next){ $acc + math:sqrt($next)})}</a> }; [3] 17:48:02.782 127.0.0.1:48262 admin REQUEST [POST] http://localhost:8081/rate-selector/test 17:48:02.783 127.0.0.1:48252 admin REQUEST [POST] http://localhost:8081/rate-selector/test 17:48:02.783 127.0.0.1:48258 admin REQUEST [POST] http://localhost:8081/rate-selector/test 17:48:02.784 127.0.0.1:48260 admin REQUEST [POST] http://localhost:8081/rate-selector/test 17:48:02.784 127.0.0.1:48254 admin REQUEST [POST] http://localhost:8081/rate-selector/test 17:48:04.428 127.0.0.1:48262 admin 200 1646.03 ms 17:48:04.509 127.0.0.1:48260 admin 200 1725.45 ms 17:48:04.710 127.0.0.1:48254 admin 200 1926.27 ms 17:48:04.772 127.0.0.1:48258 admin 200 1988.41 ms 17:48:04.904 127.0.0.1:48252 admin 200 2120.33 ms

2 1

Problem with db:create and addraw
by Lizzi, Vincent 17 Sep '19

17 Sep '19

Greetings, I am seeing different behavior between XQuery db:create() and the command CREATE DB when trying to create a database from a zip file, which is causing a problem. I want to import a zip file that contains XML and binary files (images, ets.) into a new BaseX database using db:create() in a web app. The command "CREATE DB" works as expected. For example, after running this script all contents of the zip file, including XML and binary files, are stored in a new database named "test". SET ADDRAW true SET ADDARCHIVES true SET ARCHIVENAME false CREATE DB test C:\path\to\file.zip The XQuery function db:create() is not working correctly. The following query should be equivalent to the above command script, however binary files get placed in a new folder that is created in the folder where BaseX was launch, and the binary files cannot be accessed through BaseX. The binary files should be placed in a "raw" folder within the folder for the database within BaseX' s data directory the same as what CREATE DB produces. db:create('test', ' C:\path\to\file.zip ', (), map{ 'addraw': 'true', 'addarchives': 'true', 'archivename': 'false' }) I've tried this in BaseX versions 9.0.2 and 9.2.4, both produce the same results. Is there a way to resolve this problem? Thanks, Vincent Vincent M. Lizzi - Digital Production Manager Taylor & Francis Group 530 Walnut St., Suite 850, Philadelphia, PA 19106 E-Mail: vincent.lizzi(a)taylorandfrancis.com<mailto:vincent.lizzi@taylorandfrancis.com> Phone: 215-606-4221 Web: http://www.tandfonline.com/ Taylor & Francis is a trading name of Informa UK Limited, registered in England under no. 1072954 "Everything should be made as simple as possible, but not simpler." Information Classification: General

2 5

Re: [basex-talk] binding-types?
by Bridger Dyson-Smith 16 Sep '19

16 Sep '19

Hi Ben - Does something like: ``` declare variable $name external; for $i as xs:integer in 1 to 5 return element { $name } { $i } ``` work any differently? Best, Bridger On Thu, Sep 12, 2019 at 10:49 AM ben.engbers(a)be-logical.nl < ben.engbers(a)be-logical.nl> wrote: > xs:string is neither accepted and says that no namespace is defined for > 002xs:string > > Ben > > > -------- Oorspronkelijk bericht -------- > Onderwerp: [basex-talk] binding-types? > Van: Ben Engbers > Aan: Basex Mail-lijst > Cc: > > > Hi, > > While creating a R-package, based on my R client-implementation, I found > that the binding function was malfunctioning. > > After rewriting that function, the following is accepted: > query_txt <- "declare variable $name external; for $i in 1 to 5 return > element { $name } { $i }" > query_obj_1 <- Query(Sess, query_txt) > success <- query_obj_1$queryObject$Bind("name", "number") > print(query_obj_1$queryObject$ExecuteQuery()) > > results in: > "1" "2" "3" > "4" "5" > > When I change the line > 'success <- query_obj_1$queryObject$Bind("name", "number")' in > success <- query_obj_1$queryObject$Bind("name", "number", "xs:integer") > the following error is produced: > "[XPST0081] No namespace declared for '\002xs:integer'." > > What types are accepted? > > Ben > >

2 1

XQuery Update and Namespace Manipulation
by Andreas Mixich 14 Sep '19

14 Sep '19

Hi, copy $c := <lorem>ipsum dolor sit amet</lorem> modify insert node namespace {"lipsum"} {"lorem-ipsum"} into $c return $c Expected result: <lipsum:lorem xmlns:lipsum="lorem-ipsum">ipsum dolor sit amet</lipsum:lorem> but I get <lorem>ipsum dolor sit amet</lorem> copy $c := <lorem>ipsum dolor sit amet</lorem> modify replace node $c/namespace-node() with namespace {"lipsum"} {"lorem-ipsum"} return $c Here I get an error: [XQST0134] Namespace axis is not supported. How can I change namespaces for nodes with the simple update expressions? For background: I try to implement a function: declare function local:add-namespaces-to-nodes( $nodes as element()*, $namespaces as array(*)*) as element()* { for $namespace in $namespaces for $node in $nodes return copy $c := $node modify insert node (namespace {$namespace(1)} {$namespace(2)}) into $c return $c }; Thanks! -- Minden jót, all the best, Alles Gute, Andreas Mixich

3 7

Re: [basex-talk] binding-types?
by ben.engbers＠be-logical.nl 12 Sep '19

12 Sep '19

1 0

binding-types?
by Ben Engbers 12 Sep '19

12 Sep '19

Hi, While creating a R-package, based on my R client-implementation, I found that the binding function was malfunctioning. After rewriting that function, the following is accepted: query_txt <- "declare variable $name external; for $i in 1 to 5 return element { $name } { $i }" query_obj_1 <- Query(Sess, query_txt) success <- query_obj_1$queryObject$Bind("name", "number") print(query_obj_1$queryObject$ExecuteQuery()) results in: "<number>1</number>" "<number>2</number>" "<number>3</number>" "<number>4</number>" "<number>5</number>" When I change the line 'success <- query_obj_1$queryObject$Bind("name", "number")' in success <- query_obj_1$queryObject$Bind("name", "number", "xs:integer") the following error is produced: "[XPST0081] No namespace declared for '\002xs:integer'." What types are accepted? Ben

1 0

Stack overflow instead of syntax error
by Buddy Kresge 11 Sep '19

11 Sep '19

We have some code generation and it produced a syntax error. However, running 9.2.4 in production, the server gave a stack overflow tail recursion instead of "Cannot convert xs:integer to xs:date: 1970." (like the GUI did). Here was the error reported: java.lang.RuntimeException: org.basex.core.BaseXException: Stack Overflow: Try tail recursion? Here is the snippet of code that originally cause the 'stack overflow'. The issue is the date is not in quotes. function($endDate as xs:date?) as xs:Boolean { let $intervalBeginDate := xs:date(2019-09-30) let $intervalEndDate := xs:date(2019-09-30) return ($endDate ge $intervalBeginDate) and ($endDate le $intervalEndDate) }; However, when I fix the code generation, everything is fine (see new code below). Just seems "strange" that this would cause the server to throw a 'stack overflow' error. function($endDate as xs:date?) as xs:Boolean { let $intervalBeginDate := xs:date('2019-09-30') let $intervalEndDate := xs:date('2019-09-30') return ($endDate ge $intervalBeginDate) and ($endDate le $intervalEndDate) }; Buddy

2 2

Re: [basex-talk] GUI, visualization panels
by Giuseppe G. A. Celano 10 Sep '19

10 Sep '19

Hi, no, I just open the GUI and then an XML document. Thanks. > On Sep 9, 2019, at 2:52 PM, Alexander Holupirek <alex(a)holupirek.de> wrote: > > Hi Giuseppe, > > just gave it a try using basex 9.2.4 with java version "1.8.0_121" on macOS 10.14.6. > I was not able to reproduce the issue so far. > > One thing, did you open a database first? Visualization are only available on (opened) databases (not in-memory documents, such as XML read from filesystem, http, ...) > > Kind regards, > Alex > >> On 7. Sep 2019, at 15:56, Giuseppe G. A. Celano <celano(a)informatik.uni-leipzig.de> wrote: >> >> 1.8.0_171 >> >> Thanks. >> >> >> >>> On Sep 7, 2019, at 1:43 PM, Alexander Holupirek <alex(a)holupirek.de> wrote: >>> >>> Hi Giuseppe, >>> >>> what java version do you use? >>> >>> Kind regards, >>> Alex >>> >>>> Am 07.09.2019 um 11:42 schrieb Giuseppe G. A. Celano <celano(a)informatik.uni-leipzig.de>: >>>> >>>> Hi, >>>> >>>> Apparently, I cannot access the visualization panels in the GUI on my Mac (10.13.6) (they cannot be selected): am I missing anything? Thanks. >>>> >>>> Best, >>>> Giuseppe >>> >> > >

1 0

Passing through entities unchanged when serializing
by Andreas Mixich 09 Sep '19

09 Sep '19

Hi, when serializing a string, that contains literal XML with entities, how do I pass through those entities unchanged? Example: let $input := "Lorem ipsum ' dolor sit amet " return serialize($input) results in: Lorem ipsum dolor sit amet, ' consectetur adipisicing elit. but I want: Lorem ipsum dolor sit amet, ' consectetur adipisicing elit. -- Minden jót, all the best, Alles Gute, Andreas Mixich

3 7

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

BaseX-Talk September 2019