From: Leo S. <le...@gn...> - 2005-12-20 14:35:18
|
Es begab sich aber zu der Zeit 19.12.2005 17:21, da Christiaan Fluit schrieb: > Some recently gathered insights on this issue: > > *** one simple workaround that completely makes this a non-issue is to > store all extracted information in temporary containers ... hm, too complicated, the RDFContainer should spare us this work > *** Another workaround is to create several RDFContainer instances, > each with a different described URI but *sharing* the same Repository. > Each DataObject then has its own RDFContainer with a stable described > URI. This fixes the behaviour of the Map-like methods: you can at any > time retrieve the metadata of a DataObject and inspect or alter its > contents. I expect this will also make my dependencies on the > underlying Repository in the IMAP code go away. better > > - I have no idea yet whether there are any consequences for concurrent > access on the repository. probably no problem. crawling is usually done in one thread anyway (multiple threads don't help here, ie when crawling MS-Outlook they are impossible anyway) Leo |