Log file Viewer
#mongodb logs for Saturday the 20th of February, 2016
[03:46:38] <kalx> Hi all. Is there a way to query a document to see if my query input array contains all the values that are in a document's array value? So like $.all except matching in the other direction.
[03:47:00] <kalx> eg. document is: { tags: ['b','c','d'] }. I want to query with ['a','b','c','d','e'] and find documents where all the document's tags are in my input array
[03:53:11] <Boomtime> however, you can use aggregation to do it today; https://docs.mongodb.org/manual/reference/operator/aggregation/setIsSubset/#exp._S_setIsSubset
[04:16:54] <kalx> @Boomtime: Tested $setIsSubset, worked but only in the context of an aggregate call. I'm trying to do this within a findOneAndDelete call (atomic grab+delete for perfomance reasons). Any idea if that's possible? Seems not, but researching in meantime
[04:41:33] <kalx> @Boomtime: Just to share. Found a workaround but will likely destroy performance for larger collections. As a find condition -> {my_field: {$not: {$elemMatch: {$nin: ['a','b','c','d','e']}}}} . Too many negatives makes it confusing... essentially, finds any document with my_field array where none of the array elements can satisfy not being in the input list.
[04:42:28] <kalx> so not happy about it (even the other one $setIsSubset likely would have been not good performance wise)
[04:48:53] <Boomtime> @kalx: yeah, i know of that workaround and it suffers the problem you recognized - it cannot use an index, literally there isn't one that can be used - the aggregation pipeline _can_ - it won't be terribly efficient, but it should be much better than the $not/$elemMatch/$nin abomination
[05:07:44] <Trinity> i'm working with mongo in node.js and i'm trying to use a string variable for a key in the document
[05:10:34] <Trinity> shell shows that the key 'throttleKey' is being created/updated instead of 'myIndex'
[05:14:55] <Boomtime> @Trinity: you can't do what you want in JSON - you'll need to use a real javascript array to create it
[05:15:20] <Boomtime> right now you are describing the update using JSON syntax - which will assume that all keys are literals
[05:16:01] <Boomtime> create a javascript object instead, even if that means instantiating all other members using JSON first, then adding one more entry using javascript array syntax
[05:16:04] <kalx> @Boomtime: Unfortunately, aggregate leads to other trouble. Need to find+delete or mark processed at a rapid pace which means atomicity is best (Findanddelete or findandupdate). Aggregate would lead to a multistep process which is ultimately doomed.
[05:17:52] <Boomtime> i mean, they can do what they say on the packet - which is assure atomicity, but so can you
[05:18:57] <Boomtime> if you find a document that matches your special query and then use an update which specifies the _id you can hit only that document - even if there is a chance of an ABA problem, you can just use another field as a sequencer to avoid it
[05:24:14] <kalx> Issue is more to avoid multiple processors from identifying the same task to process. Sure, we can easily prevent multiple from grabbing the same record, but then one fails and has to re-identify + grab again. With very high concurrency, it's easy to run into perf problems there.
[05:28:19] <Boomtime> so don't re-identify - use a cursor - then loop 'update' until you get one that succeeds - reads are much cheaper than writes, so a miss on the read portion of an efficient update doesn't matter so much
[05:28:46] <Boomtime> also, given that in order to 'miss' one other _must_ have succeeded, the system must be making progress
[12:33:19] <cdunklau> i don't get this... i'm trying to do the update data part of the mongodb getting started thing (https://docs.mongodb.org/getting-started/node/update/), here's my code https://gist.github.com/cdunklau/7eef9f4a3280007bf3ff
[12:56:56] <cdunklau> aha. i didn't have an assert.equal(err, null) so the error wasn't getting outputted
[13:06:34] <cdunklau> looks like there's something up with my mongodb package. when i do "npm version mongodb" it says it can't find the package.json but when i look in the node_modules/mongodb i find it, and it shows me this: https://gist.github.com/8e0610dadbd39be40f01
[13:08:49] <kurushiyama> synapse: ... ...Had we? You were asking for other poor lads spending their Friday... I was the other one ;)
[13:09:27] <synapse> As I said I done some tests. I have a small db with under 100 entries, and I also messed about with using the "restaurants" collection example dataset which is very big. The app works fine with the former and on the latter the whole app gets killed on like 4 queries
[13:11:44] <synapse> This is just pure JS server side using simple mongodb driver and AJAX client side, no frameworks or libraries
[13:12:40] <synapse> So basically my questions is A) is this normal behaviour for an app that has no server side result restrictions?
[13:14:40] <synapse> Well I am shocked to hear that because I have no idea then why the app would die when returning large responses
[13:14:54] <kurushiyama> If just a few queries are killing you... you might want to have a look at the queries and the according indices.
[13:15:28] <kurushiyama> synapse: And you definetly want to use a cursor and not load thw hole result set ;)
[13:16:46] <olivervscreeper> Hi - I'm trying to get the PHP Mongo driver installed. I've installed the MongoDB driver (not Mongo), and phpinfo() shows it as enabled, but if I try to create a connection with new Mongo() it says that the class is not found. Any ideas? [13:09] == Cannot send to channel: #mongodb
[13:17:08] <kurushiyama> you should have an index on those two fields (order matters!). So you need to check wether an according index exists.
[13:19:15] <kurushiyama> without an index, mongod would need to read each and every document from disk to determine wethe the document matches.
[13:19:16] <olivervscreeper> My issue's on StackOverflow if that explains it better: http://stackoverflow.com/questions/35514409/uncaught-error-class-mongoclient-not-found
[13:21:33] <kurushiyama> synapse: ok. Since persumably you do not have an index on "name", think for a sec what this query does...
[13:21:44] <synapse> kurushiyama In simple terms can you describe how I'd alter the query to include indexing
[13:22:18] <kurushiyama> synapse: You have to understand first. I teach you how to fish, not simply give you one.
[13:23:01] <kurushiyama> synapse: So, what happens is that each and every document is read from disk and a regex is applied then...
[13:23:32] <kurushiyama> synapse: (a bit simplified, since there are RAM caches, but take it as a good rule of thumb)
[13:25:44] <kurushiyama> synapse: An index basically is a specialized data structure which can be searched efficiently and will be kept in RAM, if at all possible. It holds the values of the indexed field(s) and references the position of the document in the data files.
[13:30:02] <kurushiyama> When you do a query on an indexed field, the match is done against the index, and only then the matching docs are read from the disc, (vastly) reducing latency (RAM is orders of magnitude faster than any disk)
[13:31:15] <synapse> Ahh so basically if the query checks a significant amount of docs more than what is returned eg, checked 10,000 docs to return 2 results, this is a good example where an index would be ideal
[13:32:16] <synapse> I believe that's what my other query was doing, checking like a million docs to return 2 results lol
[13:32:49] <kurushiyama> synapse: So, now you go figure on how to create a proper index for your query. aka: read the acording docs!
[13:38:20] <synapse> an index is basically an extra entry on the key:value that incorporates a built in limitation?
[13:47:58] <kurushiyama> synapse: after your index is built, the query looks exactly the same, but mongod will utilize the index to make the query faster ;)
[13:57:45] <kurushiyama> synapse: Ok, a basic: Identify the questions you have on your data, model your data ___according to your questions___ (as opposed to model by entities) and add the indexes for the fields your query. That's about all there is to it (details and intricacies ommited) and this is what about 90% of the people ranting about MongoDB get wrong.
[13:59:06] <kurushiyama> synapse: I wouldnt. If you change the index (NoSQL is pretty fluid), the hint will go wrong.
[14:08:30] <synapse> I know when i tested on the shell without .hint it still used the index but for some reason in the code i can only assume it wasn't using it
[14:10:13] <kurushiyama> synapse: Yeah. Absolutely. I'd rather change my job for a promising career as a loaf of bread.
[14:10:41] <synapse> If I had a little timer to the onkeyup AJAX query so it doesn't send off queries every single time it should be a lot more solid
[14:13:12] <kurushiyama> synapse: Single-threaded, no type safety and callback hell. I tried node _very_ shortly.
[14:15:14] <synapse> kurushiyama callback hell is being phased out with the use of chained methods / promises :)
[14:31:41] <kurushiyama> synapse: Rolls Royce? Not really, since it lacks mature frameworks, imho. The better comparison would be Ferrari – it is __fast__. And you compile static binaries. Deployment can be done by copying a (one) file from one machine to the next, if done properly.
[14:33:12] <kurushiyama> synapse: Where it excels is API building. This is where Go really rocks. Think of a Rolls Royce with a formula 1 engine and the capabilities of using it.
[14:34:22] <StephenLynx> 1- callback hell is just as bad as any other uncontrolled nesting. no differen than nesting too many loops or conditions
[14:34:40] <StephenLynx> 2- single threaded isn't an issue because of how you have the cluster native module to use all cores of the cpu
[14:35:07] <StephenLynx> 3- neither is type safety because you have unambiguous operators such as '==='
[14:35:58] <kurushiyama> StephenLynx: It still is concurrent instead of parallel. Agreed, node does an impressive job, but still it is single threaded ;) And type safety enforcing during runtime is not exactly the problem. ;)
[14:38:54] <kurushiyama> StephenLynx: Well, let's say you have like 4 users and 4 cores. With true parallelism, those would be processed in parallel. With concurrency, each worker from time to time yields. It is not that this would not work, but it makes less good use of the resources. When it comes to scaling out, the difference will become noticeable sooner or later.
[14:39:43] <StephenLynx> they will be processed in parallel because each one would be processed by a different process running on a different core.
[14:40:28] <StephenLynx> now if it were processing 8 requests on 4 cores, now you have 4 in parallel and the other 4 concurrently.
[14:40:49] <kurushiyama> StephenLynx: Huh? Wait a sec. As far as I know, v8 is inherently singlethreaded.
[14:41:34] <StephenLynx> 1 master process and as many worker processes, usually you spawn 1 worker for each core
[14:43:24] <kurushiyama> StephenLynx: So, if I get you right (paraphrasing to check if I got you right): You have a single master node process, dispatching request to worker processes, for which one is spawned per CPU?
[14:44:15] <cdunklau> i'm following the getting started guide for node (https://docs.mongodb.org/getting-started/node/update/) and I've hit a wall with the update portion. code and error is here: https://gist.github.com/cdunklau/7eef9f4a3280007bf3ff
[14:44:53] <StephenLynx> you spawn the worker processes manually, though. so 1 per core is just what you usually want.
[14:45:44] <kurushiyama> StephenLynx: Interesting. One point less on my long list "Why not to do node..." ;) Thanks for the info! But I think cdunklau needs a node specialist, now ;)
[14:48:58] <kurushiyama> synapse: Nah, that's runtime. Last but not least – I used to be a sysadmin. I like go for the easy way to have a natively compiled application including all resources deployed by merely copying a file.
[14:49:49] <synapse> cdunklau $currentDate should be 'currentDate' as it's a field not an in-built mongo option on line 13 of pastebin
[14:50:22] <StephenLynx> and deploying is not an issue on scripted software when you use git to deploy.
[14:53:21] <cdunklau> synapse: currentDate isn't a field in the document, or did you mean something else?
[14:57:36] <cdunklau> synapse: right i'm just curious why you suggested that $currentDate should be currentDate (before you realized it was valid)
[14:58:03] <cdunklau> synapse: did you think i just made a typo in trying to replace an embedded document or something?
[14:59:27] <synapse> I didn't think it was a recognised mongo field, but it's got an issue with that line
[15:11:04] <cdunklau> synapse: it's all good now, i removed the currentDate thing because it doesn't work
[15:12:54] <cdunklau> the original one meaning the one i sent ~30 min ago or the one i sent a while before
[15:14:27] <synapse> cdunklau the syntax you used is wrong, you don't chain a callback function on the end of the query like that
[15:15:21] <cdunklau> http://mongodb.github.io/node-mongodb-native/2.0/api/Collection.html?_ga=1.217822224.729576732.1455824695#replaceOne
[15:16:02] <synapse> I tend to build the query as it would be sent on the shell cdunklau and then put it inside a function with a callback
[15:16:34] <cdunklau> yeah i'm *super* new to all this. i imagine i'll figure out or read about better strategies
[15:16:38] <synapse> because the query works fine on my shell I tried it on my restaurants collection cdunklau
[15:17:38] <synapse> Yes your error pointed to that line so your mongo version / node versions are important. I've had many errors with version issues
[15:17:40] <cdunklau> synapse: i really appreciate you going into depth on this, but i have a feeling i communicated the status or something badly
[15:21:51] <Tachaikowsky> If something is written to mongodb using mongoose and secret key, can I use mongodb plugin now to read it without a secretkey?
[15:22:20] <cdunklau> the problem was originally difficult to figure out because the example with updates doesn't have the error === null assertion. i don't know any better so i took it out of my code. then i just got undefined in the console and was confused
[15:22:51] <cdunklau> the examples before (querying and such) have assert.equal(err, null) in the callback
[15:23:17] <Tachaikowsky> kurushima, in my case - the key is a 'secretword' that is in my config.js (I am working with node.js)
[15:24:58] <cdunklau> i'm learning mongo and node with the hopes of moving from tech support to development at my company
[15:25:03] <kurushiyama> cdunklau: You might likely run into driver problems, unless you are very careful etc.
[15:25:19] <synapse> cdunklau that syntax looks so odd to me, where the function was tagged onto the query, I'm a beginner too but that does look strange as
[15:26:09] <cdunklau> synapse: seemed natural enough to me. i'm just getting started with mongo, but i've played with node and other async stuff to become fairly comfortable with that style
[15:27:36] <synapse> ahh so updateOne in the mongo driver api docs must define an addition callback as a parameter, something you wouldn't have on the shell of course
[15:31:46] <kurushiyama> cdunklau: As long as you assume localhost as being safe, sure. Though I wouldnt
[15:33:03] <kurushiyama> cdunklau: Sorry ;) Well, as for your playing around, I'd use a docker image for a more recent version of MongoDB. And you might want to have a look at MongoU
[15:37:22] <synapse> Tachaikowsky just taking a random stab at this as it's not something I know much about but I assume the secret key is only there to validate a session cookie on the client, I don't see why it would affect your ability to read a document from the db?
[15:38:29] <kurushiyama> StephenLynx: Do you know the times when men were real men and compiled their OS on their own?
[15:40:27] <kurushiyama> StephenLynx: I dont get the point of bower. In which way is it different than to have a git submodule?
[15:45:07] <cdunklau> so there's a lot of mongo hate, some of it's based on old issues that have been fixed, but the arguments about data-you-think-isnt-relational-until-it-is seem to be pretty reasonable. I'm sure you guys have to deal with the hate a lot in here, so i'll try to keep it specific and opinion-free.... What are the use cases where mongo really shines?
[15:45:28] <kurushiyama> StephenLynx: Infinite Monkey theorem. Sounds about right... I could not describe how much I hate frontend development.
[15:45:56] <StephenLynx> large datasets of data that are not too related and don't follow a strict schema, cdunklau
[15:46:07] <kurushiyama> cdunklau: Many people say MongoDB can not be used for financial transactions.
[15:46:59] <StephenLynx> well, if your system depends deeply on transactions, mongo will be not too optimal. because you will have to implement the transactions on software code.
[15:48:06] <kurushiyama> cdunklau: A booking would be {_id:someId, deductFrom: acctId1, transferTo: acctId2, amount:200.00}
[15:49:05] <synapse> I used to use postgres but I recently switched to mongo purely for the experience of using it with node, the JSON format is handy but I think postgres supports that too
[15:49:16] <kurushiyama> cdunklau: ever wondered why it takes even directly connected cash machines a while to find out wether you can get some money? Because they will do an aggregation on the bookings.
[15:49:40] <kurushiyama> StephenLynx: But that's an example you often see in the context "You cant do this with Mongo"...
[15:50:35] <kurushiyama> cdunklau: To make a long story short: There are _very_ few things you absolutely can not do with MongoDB. Everything else is prone to sensible data modelling.
[15:51:14] <kurushiyama> StephenLynx: Actualy, the format of bookings with a referall to accounts is how banks exchange bookings ;)
[15:51:15] <StephenLynx> if your implementation is X tool is a flawed hack, you might as well be using Y tool.
[15:51:27] <Tachaikowsky> synapse, to confirm - YES. It works. The secret key is associated with the plugin, and the db details can be directly accessed from mongoshell
[15:52:30] <synapse> Tachaikowsky thanks for the update, that's how I expected it to work but as I say it's just on instinct as I don't know anything about mongoose. Secret key's on the server side are used to validate the clients session
[15:53:07] <Tachaikowsky> StephenLynx, yes I am moving away from mongoose - exactly why these questions are being asked :)
[15:53:11] <kurushiyama> StephenLynx: I get you. It was just an example. Transactions imho are overestimated. From my experience, if you need a transaction spanning multiple tables, that's a good candidate for an according document in MongoDB.
[15:53:32] <synapse> I was told fiercely to not use MongoDB but the truth is I didn't really understand why the said person was so adamant about it
[15:53:42] <cdunklau> kurushiyama: well i was wondering about the transaction/data integrity thing too, but i don't know enough about it to really know one way or the other
[15:53:44] <Tachaikowsky> synapse, I just want to confirm - any query that is executable on mongoshell can also be executed through node.js without much modification?
[15:53:57] <cdunklau> kurushiyama: i'm more interested in things i do understand, that is, relations
[15:55:06] <kurushiyama> cdunklau: Well, I have _started_ a series of blog posts. Need to finish the second article this WE... never seem to find the time.
[16:00:33] <kurushiyama> cdunklau: If you enjoyed it, please comment. Most important: Please add what you want to see in the future, so I can focus. Wrinting on data modelling for MongoDB can easily fill a book.
[16:07:27] <cdunklau> kurushiyama: i can't think of anything to contribute... this is so new to me i can't even have an opinion :)
[16:09:17] <cdunklau> and i don't have any formal education, so i can't even really talk about this kind of stuff from a theoretical point of vier
[16:19:49] <Tachaikowsky> db.users.find({"_id": ObjectId("req.decoded.id")}).update{ $push: { arrayOfFriends: req.body.friend } }
[16:27:45] <cdunklau> Tachaikowsky: http://mongodb.github.io/node-mongodb-native/2.0/api/Collection.html#findOneAndUpdate ?
[16:32:35] <StephenLynx> on node you just pass your callback as the last parameter. or if you wish to perform multiple operations on the cursor, like sort and limit, you call toArray() as the latest function and pass the callback for it
[16:33:23] <StephenLynx> blabla.find({query},{projection, if desired}).sort({field and direction}).toArray(function gotArray(error, data){});
[16:35:39] <StephenLynx> performing multiple queries fetching one document at once on multiple cursors.
[16:36:13] <StephenLynx> usually I fetch all as a single array if I know there isn't too much data or use one cursor at a time if I know there might be too much data.
[16:36:44] <StephenLynx> using a single cursor and getting one document at once usually isn't worth.
[16:37:00] <StephenLynx> too much code and your cursor will timeout if your operations take too long
[17:50:33] <synapse> StephenLynx: http://pastebin.com/BZCciPrc .. so I have a private method _searchUsers which never gets called externally
[17:53:57] <StephenLynx> actually, findOne wouldn't work there, nvm. but a find followed by toarray would
[17:54:36] <synapse> but I don't see how an error can happen if the query is filtered and processed prior to going in
[17:56:26] <StephenLynx> and I think you could also use $text. don`t know your use case, you should check if it works better than regex for that.
[17:56:50] <synapse> my function StephenLynx that gets used inside the core of the app doesn't take a db argument, as that's handled on the module page I posted. I just do searchUsers(query, .....
[17:59:14] <synapse> StephenLynx I stress tested it earlier and it was dying so I knew it needed improving that's why I joined the channel to ask about it, I was told to add indexing which I did
[17:59:51] <StephenLynx> change so you only connect once and get a reference to the collections once.
[18:01:31] <StephenLynx> notice how I just provide functions that return the result of .collection()
[18:01:48] <StephenLynx> when the other modules start they call these functions and store the result.
[18:03:47] <synapse> so as this was inside ajax, it's literally connect/dc every time key is pressed!
[18:09:04] <StephenLynx> things that only happen every other bluemoon but only if you are wearing a rainbow wig on a bathtub made out of cheese
[18:09:49] <StephenLynx> so now you have to check if your bathtub is made out of cheese or porcelain and can`t just assume the later
[18:12:06] <StephenLynx> any software that handles external input works based on assumptions to some degree.
[18:13:46] <StephenLynx> the more usage a software gets, more of these unpredictable and undesirable behaviour is revealed.
[18:13:55] <synapse> I totally agree I was lazy on the error handling but it wouldn't be that way if it was a fully working app I'm literally just trying to get off the ground with mongo and node ya know?
[18:15:19] <StephenLynx> i have myself some code that would easily break if the input were some unexpected stuff.
[18:16:11] <synapse> I have to admit StephenLynx, Node has been a peculiar language for me to get to grips with as it's very different to traditional OOP or procedural languages I've used before. I found coming up with a way to organise the code a minefield
[18:19:46] <synapse> The problem is not so much me, but what everyone else does you know? I'f I have to depend on libs I kinda need to follow how they're built
[18:20:26] <synapse> I had to be able to understand them when I read them to know what they were doing, and they all make use of the proto chain
[18:20:43] <StephenLynx> you dont. you just have to understand what they spit out on their callback or their return.
[18:21:07] <synapse> So I ended up implementing it in my own code not because I had to but because I figured well if libs are built this way it can't be that bad?
[18:22:33] <synapse> The problem for new comers is besides from official docs, where do we go for consensus on "the right way"
[18:24:55] <kurushiyama> Regarding obfuscation... $_=q(s%(.*)%$_=qq(\$_=q($1),$1),print%e),s%(.*)%$_=qq(\$_=q($1),$1),print%e
[18:29:59] <synapse> The problem comes though as I said, lets say I end up in an environment that means I need to use JS but it's insisted I do it the "conventional" way.... I'd be lost
[18:33:29] <synapse> it's np Tachaikowsky but I really am not an expert with node or mongo but I will help if I can
[18:35:22] <synapse> StephenLynx do you like Mongo? I here people on freenode with real split opinions on it's viability
[18:36:34] <StephenLynx> its performance, feature rich, intuitive and got tools to easily able one to scale its deploy.
[18:41:29] <synapse> StephenLynx what do you think is essential steps to deploying a node http server
[18:43:57] <StephenLynx> https://gitgud.io/LynxChan/LynxChan/blob/master/aux/lynxchan.conf https://gitgud.io/LynxChan/LynxChan/blob/master/aux/lynxchan.rc https://gitgud.io/LynxChan/LynxChan/blob/master/aux/lynxchan.systemd
[18:45:15] <StephenLynx> and then I provide a install script https://gitgud.io/LynxChan/LynxChan/blob/master/aux/root-setup.sh
[18:56:47] <StephenLynx> where any template or FE file, such as js and css live on a separate project.
[18:57:39] <synapse> I have these sites bookmarked so I'll come back to them when i start working on my project again
[19:21:22] <StephenLynx> since I get shit done (and shit few could, mind you) they just let me be :v
[19:23:26] <StephenLynx> the problem is market saturation of code monkeys and people not knowing how to hire
[19:23:32] <synapse> You had that virus recently and set the army in to fight the mosquito, we had it on our news here in England
[19:28:39] <StephenLynx> its because of them we have homophobic religious dicks running our human rights comitee.
[19:29:47] <synapse> you don't know about heretics? It was a period in history where anyone who questioned the religious doctrine in society was tortured :)
[19:32:04] <StephenLynx> they are poor because of shitty politicians and they are the ones that keep electing these.
[19:32:23] <StephenLynx> and as a byproduct we get less personal freedoms because of religious fanatics.
[19:32:48] <synapse> I often feel that way here that it seems like the majority of people don't think or question anything and they end up voting in the governments
[19:46:57] <vak> how to pass an index hint to an aggregate() ? there is no info on this in docs and a call like this: db.c.aggregate(...).hint() leads to TypeError: Object #<Object> has no method 'hint'
[19:48:23] <Tachaikowsky> synapse - quick question, is the only way to verify whether something exists is to do a find, then pass output to array, then check length??
[19:51:11] <StephenLynx> if you can skip code if you didn't find anything, just check the final array length
[19:55:39] <StephenLynx> findOne is also faster because it will stop traversing the collection on the first match.
[19:57:14] <StephenLynx> it just uses a limit(1) and returns you the document instead of an array of documents
[19:57:28] <StephenLynx> so is more handy than using a find().limit(1), because that will give you an array
[19:59:25] <Tachaikowsky> StephenLynx, basically I am seeing if the person I am trying to add has me on their blocked list.
[19:59:39] <synapse> Tachaikowsky http://stackoverflow.com/questions/7658228/mongodb-count-is-very-slow-how-do-we-refine-work-around-with-it
[20:00:41] <Tachaikowsky> StephenLynx, whats the best way to find if I figure in their blocked list?
[20:01:23] <synapse> Tachaikowsky if you're using that query in the pastebin then just check the length of the result (my opinion)
[20:02:18] <StephenLynx> the other one is finding of arrayOfBlockedUsers has the user you are trying to see if its blocked
[20:02:49] <synapse> so it doesn't just check for those entries and if a record comes back the person is blocked?
[20:03:24] <Tachaikowsky> That's why I am using req.body.id for _id and req.decoded.id for $elemMatch
[20:04:23] <StephenLynx> in that case the value of arrayOfBlockedUsers on user won`t be what I expect
[20:05:59] <synapse> That's what I meant by the way of a true or false result, it's either blocked or isn't
[20:07:33] <Tachaikowsky> If the user doesn't exist at all, I cannot be on their blocked list right?
[20:09:12] <synapse> Tachaikowsky why not pass the result of that back into the function that checks the blocking depending if the user is found first
[20:10:11] <StephenLynx> if you try to access user.arrayOfblocked and user isn`t defined, you will crash.
[20:12:08] <synapse> That db query Tachaikowsky is literally only going to return either 0 or 1 results though right, or wrong?
[20:14:09] <StephenLynx> 1: no document, 2: document with no field for array of blocked users, 3: document with field for array of blocked users
[20:15:33] <synapse> if he added an arrayOfBlockedUsers: true in there that would assert it needs to exist right StephenLynx?
[20:22:24] <StephenLynx> callbacks its just a mean of returning value where you execute something instead of returning a value.
[20:23:53] <StephenLynx> in C you wouldn`t be able to declare it like that, though, because js got closures.
[20:28:02] <synapse> but Tachaikowsky take StephenLynx's advice check both the error and result exisit and if neither handle it all accordingly :)
[20:54:39] <synapse> Tachaikowsky well there are many problems with how that's constructed, one thing you should know is never place anything after a return, return will return you from the function, ntohing after will be executed
[20:56:29] <synapse> Second that will cause a connection to the db every single time the query is run, in a real world setting it will kill your app ( literally had this problem myself and learned the hardway), you need to create a connection cache / pointer something which I've yet to implement myself but StephenLynx just provided me with a well written project code in node js which I will take a look at to learn
[20:58:22] <synapse> The idea behind async code is that you can get on with other tasks whilst something is still being executed i.e it's non-blocking, eg we don't wait to read in a 10gb file before doing anything we get on with other things that's where callbacks and events come in Tachaikowsky
[20:59:57] <synapse> Tachaikowsky when defining functions, you write them with a callback as you saw earlier what me and stephen posted
[21:03:12] <Tachaikowsky> <synapse> All it is is a function which takes another function as an argument
[21:14:47] <synapse> Tachaikowsky Basically when you define a function you specifiy a bunch of work to do as you would any other, the function solves a task
[21:16:40] <synapse> instead of running a return value; to the caller, we run a function (the callback) that's placed as an argument with the results of our work as the parameters of the function we callback.
[21:16:55] <synapse> Tachaikowsky By building your entire app this way, your app will be completely async
[21:23:22] <synapse> lets say you have a function that needs to read in a 10GB file and do some complicated stuff with the bytes it read in....
[21:23:51] <synapse> Are you going to wait for that to occur and then give the result back to the caller?
[21:24:56] <synapse> No... we'll make that file handler function a callback function so we can write other functions that do other tasks in our code that will also be callbacks. Our app will consist of these types of functions, our becomes async
[21:26:54] <synapse> If you look at the node.js API, look at the Filesystem API you will see there are two functions that read in files that node offers you to use. You can choose a syncronous .readFile() function or an async version of the same thing. They both achieve the same task, the difference is how the results are handled
[21:39:52] <synapse> Tachaikowsky if you don't understand something just ask.. http://pastebin.com/ZesL9Mic
[21:45:58] <StephenLynx> since you are executing callback you are returning what callback returns upon execution with those two parameters.
[21:50:09] <synapse> And I never had someone hold my hand through it when i was learning so you are lucky :)