On 16/04/2009 5:47 PM, Martin Pfeifle wrote: > Hi, > we store proprietary organized data in blobs within an sqlite database. > Assume you have a table mydata(id, attr1, attr2, blob). > The page size of the database is 1k.
This is the *default* page size. You can change it before you first put data in the db file. > If you now store blobs in your table which are larger than 1k, > sqlite uses overflow pages and the content of the blobs is distributed > to several pages, right? Not just blobs; if your row won't fit in a page for any reason, excess row contents are put into one or more overflow pages. > Are the pages on which the blobs are stored > consecutive or can they be distributed arbitrarily within the database file? I imagine it would use pages off the list of free pages in preference to extending the size of the file ... try inspecting the source code. > Can two records share the same overflow page, or is an overflow page dedicated to one record? My reading of the relevant section of http://www.sqlite.org/fileformat.html suggests the answer is dedicated ... have a look and see if you agree. BTW there's lots of interesting bed-time reading in there :-) > Is there a formal definition of the fragmentation value which is reported by > sqlite3_analyzer? This was discussed briefly in this forum within the last few days ... AFAIK the only formal definition you will find is in the analyzer source code. HTH, John _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
