I think we could use the in-memory database to store read-only documents (our repository), a use disk-persisted database for the resulting documents.

In term of requirements, I'd prefer Marc (added to dest list) to clearly explain what is expected, but I think attribute and node-name index should be enough.

As this, we could avoid the update functionnalty on in-memory database.

Best regards,

Christopeh

Le 2015-07-16 10:06, Christian Grün a écrit :

Hi Christophe,

is there a way to put all the databases in memory ? We have a 30Gb database,
serving as repository for XInclude-like resolution, and I want to avoid disk
access (I work with Marc Messeant...)

Using special JVM, I can allocate many memory, and so want to work only in
memory. Is this possible ?

In the early days, BaseX has been a pure main-memory database. Today,
some main-memory features are still available. For example, you can
set MAINMEM to true and create a new database completely in main
memory.

You can't open existing databases in memory, though. We are currently
looking for sponsors to realize this feature [1]. Feel free to give us
a note if you are interested? And what are your requirements: do you
also need updates, index structures, ...?

Best,
Christian

[1] https://github.com/BaseXdb/basex/issues/948