So far we have a GUI for entering simple queries via a user interface.
This query interface could be extended to allow more complex queries, however:
This is a difficult implementation; and would result in a complex interface
It is not clear how feasible such an interface is for queries against multiple Datasets
this might not be what mathematicians want, they know SQL and in certain cases it might be much simpler to be aware of the code and "just" type out an SQL query
I believe that we do not want to give direct SQL (database) access to users to retain some kind of control what kind of queries are being sent.
For this reason we should develop an extended query language an expose this to users.
This language:
should allow everything the interface currently does
allow combining queries across multiple datasets and objects
be SQL-inspired (i.e. similar syntax), so that it can be easily adopted
abstract away over specific details, e.g. when JOINing multiple datasets, the obvious ids to join on should be used
compile into real database queries
be extended into a true tetrapodal query language in the future (that could be compiled into MMT)
So far we have a GUI for entering simple queries via a user interface. This query interface could be extended to allow more complex queries, however:
SQLand in certain cases it might be much simpler to be aware of the code and "just" type out an SQL queryI believe that we do not want to give direct SQL (database) access to users to retain some kind of control what kind of queries are being sent.
For this reason we should develop an extended query language an expose this to users. This language:
JOINing multiple datasets, the obviousids to join on should be usedcompileinto real database queries