thebarty / meteor-denormalized-views Goto Github PK

because of this issue #4

i need to manually refresh docs if the source collection has changed. However if the doc does not exist in the target view collection, it does not get inserted, because it asumes that the doc is already there.

Solution would be to do an upsert there instead of an update.

It seems that the package adds code to meteor client. It should be server-only.

also it should depend on simpleSchema if possible (maybe take newer version from npm instead of atmosphere-version)

This package looks like a god-send... but I'm a bit confused as to what I'm doing wrong. Hoping you might be able to point me in the right direction and clarify a couple of issues.

So, the end goal is to get tabular search working across multiple collections. We have a Responses collection, referencing both a contact and survey collection.

At the moment, my view looks like this...

 DenormalizedViews.addView({
        identifier: ident,
        sourceCollection: Responses,
        viewCollection: ResponsesView,
        pick: ['created', 'comment', 'score', 'contact', 'survey'],
        sync: {
            contactCache: (response) => {
                return Contacts.findOne({_id: response.contact});
            },
            surveyCache: (response) => {
                return Surveys.findOne({_id: response.survey});
            }
        },
        postSync: {
            responseContact: (response) => {
                const contact = response.contactCache.email;
                return contact;
            },
            responseSurvey: (response) => {
                const survey = response.surveyCache.name;
                return survey;
            }
        }
    });

I've also set up two refreshByCollections. Think I'm missing something, but just can't get anything beyond an empty cursor. To be frank, I haven't dug around too much in the code, so I'm probably just missing something really basic and hoping for a magic bullet. Two things I wasn't exactly sure about were the role of the identifier and the params on refreshByCollection, for example (doc, docPrevious, userId). Any pointers would be greatly appreciated! Thanks.

collection-hooks do not work in all occasions, e.g. if a collection is changed from another instance.

using cursor.observe should be more stable then collection-hooks. This also works if the source collection was changed in the mongodb directly

Hi @thebarty,

I find this package interesting, and as I see that you are actively working on it, I wonder if you are familiar with the Event sourcing architecture and with CQRS? The idea to have distinctions collections for writing and reading and a unidirectional data flow in between matches perfectly theses architecture principles, though the concepts naming is a little bit different—for example a “denormalized view” would be a “materialized view”.

So I just wanted to open a discussion on that, maybe these architectures could help to write the documentation or even to shape work in progress APIs.

Hoping this package isn't entirely abandoned, as it seems to be the key to fixing our broken Tabular tables that don't search or column-sort across collection joins.

How do you pass a filter at runtime using this package? As we would want the filter to run based on the user and their application context (e.g. showing a table of a user's posts from a specific chat room using a chatroomId, etc.) We would need that to run dynamically for all our users across all their chatrooms. In other words, a user goes into a chat room, goes to a tab that has a tabular table on it, and the table lists their posts from that chatroom.

We don't want or need to have the denormalized view run on every single chat post in the entire collection. Only on those that match the filter.

The knee-jerk solution seems to call a method that calls (and defines) addView at runtime upon visiting the table's template. But then how do we wait for it? How do we destroy it? Would this leak/build up memory on the server?

Having the same issue with the denormalize package which seems to run by default on the entire collection. Not a subset defined by a query.