Is it correct to think that the data resides in files and is only accessed as needed?

[EricGT]

The question may seem odd at first but having worked with SWI-Prolog for some time, currently all of the data for use with Prolog queries needs to be loaded into memory to be used. I have a need that could eventually need access to more data than can be loaded into memory on a standard PC with 16GB of memory. Even going to 128GB may not be enough. Also I am not concerned with how long the queries take, so paying the price in time to access the data from hard drives is not an issue.

Does Logica work like a normal SQL database where the data can reside on hard drives and is only accessed as needed? Or does all of the data need to be loaded into memory before a query can be run? Or something else?

[EvgSkv]

Thanks for your question!

That's correct! Logica delegates execution of the program to a database engine, which means data can be stored on disks and is accessed as needed.

BigQuery, which is currently the only fully-supprted Logica's back-end, is Google Datawarehouse which is capable executing heavy queries over datasets of hundreds of terabytes. In this case the data is stored in a distributed file system. A reasonably complex logic program over a dataset of under 1 terabyte is likely to run under a minute.

If you want to try Logica on datasets of a few gigabytes then Google BigQuery free quota of 1TB would be sufficient for quite a few experimental runs.

A lot of functionality is already supported for PostgreSQL, which makes it possible to run queries inside a standard PC. In this case the data would be stored on the local disk and accessed as needed as well.

Please let me know if you have further questions.

[EricGT]

Thanks for the quick response!

If you want to try Logica on datasets of a few gigabytes then Google BigQuery free quota of 1TB would be sufficient for quite a few experimental runs.

I see BigQuery public datasets and Google Cloud Free Program which while nice have limits that convert into a pay plan; sorry can't do that at present. Will save for a latter day.

However I do know enough about SWI-Prolog and PostgreSQL (SWI-Prolog connecting to PostgreSQL via ODBC) to possibly give this a trial run.

FYI - While not exactly the same as Logica there is a related SWI-Prolog pack. See: Constraint Query Language A high level interface to SQL databases

Also in the light of using SWI-Prolog with SQL databases, the following is useful. Wiki Discussion: Prolog in the mind set of SQL

I wrote most of the Wiki article and Wiki discussion in looking for answers to my problem so Logica is something I do plan to understand in more detail, but I am not the creator of CQL.

Thanks.

[EvgSkv]

The decision to proceed with PostgreSQL makes sense.

You might be able to run queries on BigQuery without transition to pay plan. E.g. here it is stated that you can use BigQuery on public datasets with billing disabled. And COVID-19 datasets are available to query for free until September 15th, 2021.

Having said that, your rationale for using PostgreSQL engine is very clear. An example of running a Logica program on PostgreSQL can be found here. You can modify and run it interactively in CoLab.
Let me know if you have any questions on this.

Thank you for the links about connection of SWI-Prolog with PostgreSQL, I'll study that.