Log file Viewer
#mongodb logs for Sunday the 24th of April, 2016
[00:35:52] <kexmex> how come mongodb doesn't overwrite password in arguments so that it's not seen in PS?
[04:41:31] <Ryzzan> good tutorial teaching how to populate and then aggregate ($group) population result (is it possible?)
[05:52:38] <Ryzzan> good tutorial teaching how to populate and then aggregate ($group) population result (is it possible?)... anyone
[09:08:20] <Avocher> Hello, I've made too many collections is there a way to merge two or more collections?
[09:09:28] <Derick> there is no automated way. You need to read all docs, and import them into the other collection
[11:22:15] <kurushiyama> yopp: you mean you have a document like { Foo: {Bar:"Baz"}} and you want an index on "Foo.Bar"?
[11:26:47] <kurushiyama> yopp: As you noted, you'd have to have the names in your code in order to access the values...
[11:27:45] <kurushiyama> yopp: Believe me or not. This will get you into trouble and slow and in the long run hard to maintain code.
[11:28:41] <kurushiyama> yopp: Well, as you seem to know better than I do, I answer your question: You will need to create an index for each key
[11:32:44] <kurushiyama> yopp: As I said, you'd have to create an index for each unique token in this case _as you use a value as a key_...
[11:39:03] <yopp> so you need to read and then write back on application level. with eventId as a key you don't need that.
[11:39:44] <kurushiyama> yopp: I do not know how to explain it different. If in doubt, you are of course right, which was why you asked for help in the first place.
[11:42:16] <yopp> anyway, that's a not a big deal, I can denormalize data and store ids in separate field or even collection. My fight is to keep the record footprint as small as possible, as mongo can efficiently deal with billions of small records. Overhead is just too big.
[15:10:00] <Ryzzan> how to populate and then aggregate ($group) population result (is it possible?)... anyone
[17:03:18] <palaska> i am trying to get a document by matching a status string and updated_at newer than a date
[19:06:22] <Yatekii> when I have let's say a mailbox and some mails belonging to it. would you do a key on the mailbox object which holds a list of all mail objects or would you do a separate collection for the mails?
[19:08:43] <kurushiyama> Yatekii: Good question. I have hardly ever needed one. Much less an array of subdocs. I can give you an artificial example, though: http://blog.mahlberg.io/blog/2015/11/05/data-modelling-for-mongodb/#why-would-one-use-embedding
[19:10:32] <kurushiyama> Yatekii: As per your use case, I'd have an additional field named folder, which would be a materialized path by definition. with the recipients email and the path, you should be good.
[19:16:03] <Yatekii> another question: how would you delete an object that has no name, id and nothing like that except for an attribute thats a huge list of other objects?
[19:16:26] <Yatekii> (I accidentially have a nameless object because I didn't know that I have to specifiy EVERYTHING in the update query)
[19:17:40] <kurushiyama> Yatekii: I'd probably use an aggregation to find the docs with the highest number of elements for said attribute.
[19:19:43] <kurushiyama> "Give a man a fish, and you feed him for a day. Teach a man _how_ to fish, and you feed him for a lifetime."
[19:20:32] <kurushiyama> Yatekii: I could of course give you the aggregation, but you'll learn more when you google for "mongodb $size aggregation"
[19:26:04] <kurushiyama> Yatekii: However, empty string is searchable. I could have used "dsljkasldjadoqjwoidöjaklösdj" as key
[21:01:40] <Ryzzan> how to populate and then aggregate ($group) population result (is it possible?)... anyone
[21:40:33] <StephenLynx> http://mongodb.github.io/node-mongodb-native/2.1/api/Collection.html#insertMany
[21:41:13] <StephenLynx> http://mongodb.github.io/node-mongodb-native/2.1/api/Collection.html#insertOne
[21:41:18] <StephenLynx> http://mongodb.github.io/node-mongodb-native/2.1/api/Collection.html#aggregate
[21:48:44] <Ryzzan> let's say i have a document like this { _id: {type: Number}, fk : [{ _creator : {type: Number, ref : "SomeCollection" } }] }
[21:50:12] <Ryzzan> then i would like to get data (populate) from "SomeCollection" (_creator would work as a foreign key) using a valur to related to this other collection
[21:50:38] <kurushiyama> Ryzzan: The problem with Mongoose is that it forces you to see MongoDB as an SQLish or document store. Neither could be farer from the truth
[21:53:50] <Ryzzan> StephenLynx: but when i'm dealing with data from different collection... what would be the best strategy?
[21:54:19] <kurushiyama> Ryzzan: It is conceptually problematic, to say the least, and StephenLynx can tell you some details about performance. populate basically is a good way to keep you from proper data modelling, further decreasing performance. Try it over a sharded cluster. It wreaks havox on UX.
[21:56:31] <kurushiyama> So you can just do a db.movies.find({name"Hitchikers Guide to the Galaxy"})
[21:56:32] <StephenLynx> also that. often you can use embedded data to represent 1 to many relations.
[21:57:41] <kurushiyama> Now, let us assume you have displayed your movie, and you want details on Bill Nighe
[21:58:37] <kurushiyama> Ryzzan: Get your use cases right. Derive the questions on your data from them. Now, start to model for optimized UX.
[21:59:33] <kurushiyama> Ryzzan: But somebody clicks on Bill Nighe, then you can load Bill Nighe's details.