Replies: 3 comments 9 replies
-
|
In the context of Prolog, database refers to the clause database: 7.5 Database The database is the set of user-defined procedures which currently exist during execution. The complete database is the collection of procedures with respect to which execution is performed. Each procedure is: ... The database can be managed with Of the ACID properties, the clause database should satisfy A (atomicity) in all Prolog implementations: A change of the database is either made fully or not at all. Definitely not satisfied by Scryer Prolog at the moment is D (durability), because changes to the Prolog database are currently stored only in volatile memory. It would be possible to store these changes also in durable memory. C and I, i.e., consistency and isolation in the context of sequences of operations, may require additional primitives for synchronization that must still be added to Scryer Prolog. |
Beta Was this translation helpful? Give feedback.
-
|
Prolog is a language, not a DBMS system. I feel bad even making the comparison, this is no fault of Prolog, I could make the same comparison with Python or any other general purpose programming language. But it doesn't feature nearly anything you would want in a DBMS! 2-phase commits etc The only way to add information and persist it would be to write to the source code and restart the application! |
Beta Was this translation helpful? Give feedback.
-
|
SQL is also a language, and yet we use it to control a DBMS right? From the listed features, are there any that could not or should not take place automatically when the clause database is changed? |
Beta Was this translation helpful? Give feedback.
-
|
It's fine if this is a socratic exercise, and I think Scryer/Prolog would be a fine tool for implementing such things (that's the goal!), but that's why I was wondering if anyone knew of any prior work or papers in the area. Perhaps especially with deep indexing of terms. But if XML can be indexed, I assume Prolog can too. |
Beta Was this translation helpful? Give feedback.
-
|
Maybe this helps as a starting point for such considerations: https://eu.swi-prolog.org/pldoc/man?section=persistency |
Beta Was this translation helpful? Give feedback.
-
|
Ha! I saw that and tried to port it a few months ago but it required also porting several dependent libraries and there were enough syntax differences that I took a break from it for awhile :P |
Beta Was this translation helpful? Give feedback.
-
|
There is something called Triplestore it isn't popular anymore after death of "semantic web" and "semantic desktop", but I read somewhere that it is still used in biology and in some corporate environments. |
Beta Was this translation helpful? Give feedback.
-
|
Also as far as I understand Datalog uses forward-chaining inferences: it tries to populate its database with everything that can be proved, which after indexing gives huge performance boost. Which is in contrast to Prolog which uses different default execution strategy, but you can implement "datalog" in Prolog and export it as database, I even saw some projects not so long ago... |
Beta Was this translation helpful? Give feedback.
-
Datalog is the "functor-free subset of pure Prolog", so storing predicates and such would violate that property |
Beta Was this translation helpful? Give feedback.
-
I'm increasingly interested in the idea of supporting Prolog code in databases -- as in a dedicated data-structure for Prolog terms. a text string in a text column in a relational database seems stupid. datalog isn't quite what I'm looking for either as being able to query for structure in the same way Prolog does (but with more aggressive indexing) seems ideal.
has anyone heard of anything like this or can anyone suggest good search terms?
Beta Was this translation helpful? Give feedback.
All reactions