Implementation of fn:collection

> Hi Guys, > > I was trying to use the fn:collectino function today, but I am having > trouble understanding the implementation. From past experience, > fn:collection($uri) (where $uri points to a document with a list of docs > within it) has returned a sequence of all the documents. It appears that > BaseX does not implement it this way. > > I have a database where I am storing a large number of documents, and I > would like to have several collections inside which are subsets of these > documents. I need to open these subsets quickly, and I thought fn:collection > would allow me to do this. Is there any way to easily open these subsets > using fn:collection (or a similar high performance function)? As of right > now I am solving the problem by using a for loop to traverse the documents, > but this is not fast enough for my needs. > > Thanks, > > Jeremy > > _______________________________________________ > BaseX-Talk mailing list > BaseX-Talk at mailman.uni-konstanz.de > https://mailman.uni-konstanz.de/mailman/listinfo/basex-talk >

> Is there a method to > open a subset of documents (not distinguishable by path) in the database > with performance similar to calling db:open($db_name)?

> Yes, sorry I should have specified the criteria. I have a list of a subset > of the documents in the database that need to be opened (I can store this > list in any form necessary), but I am experiencing performance problems > since I need to iterate over the list in order to filter or choose which > documents to open. > > Thanks, > > Jeremy > > > On Thu, Oct 17, 2013 at 10:18 AM, Christian Grün <christian.gruen at gmail.com> > wrote: >> >> Hi Jeremy, >> >> > Is there a method to >> > open a subset of documents (not distinguishable by path) in the database >> > with performance similar to calling db:open($db_name)? >> >> Is there any criteria regarding the documents you want to open? If you >> simply want to choose the first 10 documents, you could try a position >> filter: >> >> collection("db")[position() = 1 to 10] >> >> Talking about performance: fn:collection and db:open are based on the >> same code, so there shouldn’t be any difference. >> >> Hope this helps, >> Christian > >

> Hi Jeremy, > > it’s somewhat surprising, but XQuery itself has no database semantics. > This is the reason why the implementation of fn:collection is pretty > much implementation-defined [1]. In BaseX, the function checks if the > specified URI matches a database. If negative, the URI will be > resolved against the file system. I’m not sure why it didn’t turn out > to do so in your scenario, but I’m pretty sure you’ll find some > answers in our Wiki article on BaseX databases [2]. If not, feel free > to give us some more feedback. > > Christian > > [1] http://www.w3.org/TR/xpath-functions/#func-collection > [2] http://docs.basex.org/wiki/Databases > ___________________________ > > > Hi Guys, > > > > I was trying to use the fn:collectino function today, but I am having > > trouble understanding the implementation. From past experience, > > fn:collection($uri) (where $uri points to a document with a list of docs > > within it) has returned a sequence of all the documents. It appears that > > BaseX does not implement it this way. > > > > I have a database where I am storing a large number of documents, and I > > would like to have several collections inside which are subsets of these > > documents. I need to open these subsets quickly, and I thought > fn:collection > > would allow me to do this. Is there any way to easily open these subsets > > using fn:collection (or a similar high performance function)? As of right > > now I am solving the problem by using a for loop to traverse the > documents, > > but this is not fast enough for my needs. > > > > Thanks, > > > > Jeremy > > > > _______________________________________________ > > BaseX-Talk mailing list > > BaseX-Talk at mailman.uni-konstanz.de > > https://mailman.uni-konstanz.de/mailman/listinfo/basex-talk > > >

> Hi Jeremy, > > > Is there a method to > > open a subset of documents (not distinguishable by path) in the database > > with performance similar to calling db:open($db_name)? > > Is there any criteria regarding the documents you want to open? If you > simply want to choose the first 10 documents, you could try a position > filter: > > collection("db")[position() = 1 to 10] > > Talking about performance: fn:collection and db:open are based on the > same code, so there shouldn’t be any difference. > > Hope this helps, > Christian >

> Hi Jeremy, > > if the list is more or less arbitrary, then you’ll indeed have to > browse all your documents in order to find the ones that are relevant > you. One approach could be to specify a filtering predicate: > > let $paths := ("a.xml", "b.xml") > return db:open("db")[db:path(.) = $paths] > > If this is too slow, string comparisons can be sped up by using a map, > as recently proposed on this list: > > let $paths := ("a.xml", "b.xml") > let $map := map:new( $paths ! { . : true() }) > return db:open("db")[$map(db:path(.))] > > How many documents are stored in your database? > > Best, > Christian > ___________________________ > > > Yes, sorry I should have specified the criteria. I have a list of a > subset > > of the documents in the database that need to be opened (I can store this > > list in any form necessary), but I am experiencing performance problems > > since I need to iterate over the list in order to filter or choose which > > documents to open. > > > > Thanks, > > > > Jeremy > > > > > > On Thu, Oct 17, 2013 at 10:18 AM, Christian Grün < > christian.gruen at gmail.com> > > wrote: > >> > >> Hi Jeremy, > >> > >> > Is there a method to > >> > open a subset of documents (not distinguishable by path) in the > database > >> > with performance similar to calling db:open($db_name)? > >> > >> Is there any criteria regarding the documents you want to open? If you > >> simply want to choose the first 10 documents, you could try a position > >> filter: > >> > >> collection("db")[position() = 1 to 10] > >> > >> Talking about performance: fn:collection and db:open are based on the > >> same code, so there shouldn’t be any difference. > >> > >> Hope this helps, > >> Christian > > > > >