Hello everyone, We are facing a problem, or strictly speaking, we will face in the not so far future regarding the number of elements invenio can hold. Standard invenio tables use "unsigned medium int" type columns to store record indexes. This means the highest value will be around 16Mill (exactly 16777215) and we plan to store at least 3 or 4 times that quantity of records. The solution is quite straightforward, change all columns used to store record ids to say: int type (spanning to 4Gigs) should be enough. The problem is to implement it, since invenio has hundreds of tables and we don't know upfront which of them need to be altered. In the worse of cases we can change every column whose name be like 'id' or 'rec_id' in every table. The problems with this approach are obvious, it could take even more time that necessary, we could waste space by increasing size to columns that don't need to be changed or we could miss necessary columns, some cases are obvious like bibrec* tables but some others are not. So, have anyone faced such problem? In such case, what solution have you implemented. Even not having it, any ideas?
Thanks to everyone. -- Mauricio Acebal Senior Software Engineer Frontiers <http://www.frontiersin.org/> Centro de Empresas - UPM Campus de Montegancedo 28223 Pozuelo de Alarcón Madrid
