You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Make sure the temp DB survives a restart of the broker
FuncTest for all possible combinations on tenants already there and not, and make sure the db data is never lost
Misc
Merge the feature branch to develop and continue the work in "develop"
GeoProperty - both as attribute and as sub-attribute
Make Mintaka run locally
Connection Pool
How to deal with SQL injection?
there is a function for this, part of the psql driver but it uses malloc ... :(
We need to urlencode every string value of all attrs+subattrs, and also their names and probably more stuff.
So, we'd potentially end up with thousands of calls to malloc for the URL-encoding ...
We DON'T WANT THAT - we'll have to use something else - preferably a home made routine
Check the null value of attributes for deletion - some requests support this feature
Valgrind to make sure we have no leaks nor errors
It should be enough with a single timestamp in the DB tables (createdAt, modifiedAt, deletedAt)
the opMode field tells us which timestamp it is.
Do we need unitCode, observedAt inside the 'attributes' table? I'd like to see them in the sub-attributes table - to avoid copies!
Performance:
Avoid cloning the request tree - put back "id" and "type" instead (for batch requests)
Make service routines expand entity type and attribute names - on expansion from inside TRoE routines
Done
Prepare the functest framework for temporal
New CLI -tmp for the functest framework - to run all functests for temporal (and nothing more)
FT help function postgresCmd to query postgres db tables to check that the operations do what they should
Untested implementation for creation of DB tables on startup
DB Change: naming of tables and fields:
Tables:
entities
attributes
sub-attributes
pgDrop and pgCreate in functest framework - to start the functest from an empty state
Merge the feature/temporal branch with develop
Change of the DB table layout to be able to support temporal representation of entities
Use only one value_string field + AttributeType field to interpret the string_value field
Change almost all the names of the table fields
Change the name of the db tables to: { entities, attributes, subAttributes }
Use the exact same database names (tenants with prefix) in postgres as we use in mongo
deletedAt (to mark an entity/attribute/sub-attribute to be deleted in a point in time)
Prepared for TRANSACTIONS
Prepared for CONNECTION POOL
Short functions, with descriptive names
New DB tables
Get to the same degree of "supported features" that we had in the PoC
it was simply faster and easier to rewrite the whole thing than to try to fix the PoC
Initial implementation of the new architecture
DB Change: default tenant must be named "orion", not "orion_ld" - aligned with orion (and this prefix configurable via CLI)
Support for Attributes of type "Relationship"
Support for Attributes of type "String Property"
Support for Attributes of type "Integer Property"
Support for Attributes of type "Float Property"
CLI options for postgres db params (IP, port, user, password) - must test IP and port !
Meeting with Andres + preparation + report
travis (github actions) - with much appreciated help from @wistefan , to support postgres for the functests for TRoE (Temporal Representation of Entities)
Merge all the work (while waiting for github actions) to feature/temporal
Transactions - group all the SQL commands over the three DB tables into one single transaction (so that all fail if one of them fails)
----------------- Dec 10 - reported in CTO meeting minutes -------------------------
ToDo for the implementation of injection of temporal representation of entities
Tenants:
Misc
We need to urlencode every string value of all attrs+subattrs, and also their names and probably more stuff.
So, we'd potentially end up with thousands of calls to malloc for the URL-encoding ...
We DON'T WANT THAT - we'll have to use something else - preferably a home made routine
Do we need unitCode, observedAt inside the 'attributes' table? I'd like to see them in the sub-attributes table - to avoid copies!
Performance:
Done
-tmp
for the functest framework - to run all functests for temporal (and nothing more)postgresCmd
to query postgres db tables to check that the operations do what they should----------------- Dec 10 - reported in CTO meeting minutes -------------------------
unitCode
only for Number PropertiesThe text was updated successfully, but these errors were encountered: