#Needs reroll

The latest patch was applied successfully and the issue was resolved to my project.

Updated the cherry-picked patch to rc15

prem suthar → made their first commit to this issue’s fork.

Please Review the code and let me know if i missed anything.
Thanks!!

This needs a reroll from #152 , and it should be converted to a merge request.

@daffie well, that's why I suggest using events, that gives us the possibility to extend/override how they are processed. At least conceptually, the MongoDB driver could implement a subscriber/listener for the ExecuteMethodEnsuringSchemaEvent, make it so that it processes the event before core's SchemaRequestSubscriber does, do its alternative processing of the schema request, and stop propagation of the event so that core no longer processes it afterwards. Of course, needs to be proven in practice.

Wrongly assigned to issue to @acbramley

@mondrake: It looks like a beautiful solution. My problem is that the solution will not work with MongoDB. The solution relies on that the database will throw an exception when the table does not exists. However that is not what MongoDB does. MongoDB just creates a table when one does not exists on an insert. A table in MongoDB is called a collection. It will be a table with no validation added to make sure that it only allows data in the right form being added. See: https://www.mongodb.com/docs/manual/reference/method/db.collection.inser....

The reason I started to work on this issue was to add a hook to allow the database driver to change the table schema. The database driver for MongoDB stores timestamps as MongoDB\BSON\UTCDateTime instead of an integer value as is done by the by core supported databases.

If we could make it to work for MongoDB that would be great. I will need some time to think how I can make it to work with MongoDB. If you have some ideas or suggestions, then please speak up.

Filed 📌 Explore injecting the event dispatcher in the database service Active as a possible follow up.

Now Flood\DatabaseBackend implemenrs #166, and if this is OK we need to convert the rest where possible (i.e. calls to dynamic queries extending Query).

Pausing now to let reviewing happen.

MR!10403 is now doing #163, deprecates/converts all uses of ::ensureTableExists() and ::catchException(), and is green on all databases. A review at this stage would be helpful.

#166 is just a piece of cake away with this underlying layer, so I will push that forward.

Needs MR

Needs MR

Needs MR

Needs MR

Needs MR for proposed changes.

I'm not too happy keeping an unbound amount of state inside the page cache middleware. The original approach in #3029373: PageCache::getCacheId() sometimes uses different cache IDs during get and set, causing unnecessary cache misses for some requests → would be compatible with n main requests (look at the patches before comment #13.