Hi again, Daniel, So use a factory method to construct the item, and always return the same one.
- Dennis Krügler Daniel wrote: > Hi Dennis, > > Thanks for your interest in this issue! > > Actually we considered your proposed approach, but didn't try > it out, because we *assumed*, that jibx would allocate a new item instance > for each read item of the sequence. If this could be prevented, your > proposal would be cleaner, I think. The prevention of that reallocation of > items is important in our use cases, because each item usually contains > a lot of sub items. > > Daniel > > >> Hi Daniel, >> >> That's interesting, I hadn't thought anyone would have this >> type of situation. But I don't understand the problem with >> using <collection>s. >> You can *define* something as a collection in JiBX binding >> terms without actually *having* a collection in your Java >> object; just define an add-method for the collection, which >> can throw away the processed items. >> >> - Dennis >> >> Dennis M. Sosnoski >> SOA, Web Services, and XML >> Training and Consulting >> http://www.sosnoski.com - http://www.sosnoski.co.nz Seattle, >> WA +1-425-296-6194 - Wellington, NZ +64-4-298-6117 >> >> >> >> Krügler Daniel wrote: >> >>> Hello! >>> >>> I wonder, why the structural attribute "allow-repeats" is >>> >> limited to unordered groups! >> >>> Indeed, we use this attribute very often, because we have >>> >> the usecase >> >>> where a client program imports xml files (via jibx) which contain >>> large and ordered(!) sequences of the same item type each into the >>> same item representation (i.e. class) instance and use this item's >>> post-set method to transfer the instance to a server-side >>> >> database. We >> >>> *cannot* use collections here, because of its large memory >>> >> overhead in >> >>> our cases (The client itself doesn't need them). Our >>> >> solution leads to >> >>> a quite effective block-wise data transfer and our post-set >>> >> listener can use any caching strategy it likes, e.g. it can >> use a small buffer of only some item instances to cache them >> before sending the bunch to the server. >> >>> The sole drawback is, that unordered reading is less >>> >> effective than ordered reading. >> >>> Are there any chances to extend "allow-repeats" for ordered groups? >>> From my point of view this limitation seems rather >>> >> artificial, doesn't it? >> >>> Greetings from Bremen, >>> >>> Daniel Krügler >>> >>> >>> _______________________________________________ >>> jibx-users mailing list >>> [email protected] >>> https://lists.sourceforge.net/lists/listinfo/jibx-users >>> >>> >>> >> _______________________________________________ >> jibx-users mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/jibx-users >> >> > > > _______________________________________________ > jibx-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/jibx-users > > _______________________________________________ jibx-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jibx-users
