|
well, Im just beginning to play with it for an upcoming series on Go(lang), mainly because I wish to use JSON 'documents', and I think for non SQL work it is quite popular - on looking at the examples Im putting together, it is certainly 'easier' to change the structure of a document, rather than redefine a SQL relation table
As with sooo many things, one chooses the right tool for the job - some tools can be used as a hammer to achieve a task, that doesn't necessarily mean that it is correct to do so
A recent company I worked for were inclined to store large XML blobs in relational tables with other information - their life would have been much easier not to have done so, but evolution is rarely logical
|
|
|
|
|
It's one of the most popular databases and the most popular NoSQL database.
Source: DB-Engines Ranking - popularity ranking of database management systems[^]
No idea why Oracle is at #1 though, must be some kind of Stockholm syndrome thing
Anyway, I've used MongoDB in the past and it's very easy to just write some code and store it in the database.
Except that when you add a property you always need to think about whether that property is backwards compatible.
You have to do that with SQL too, of course, but there you are more forced to think about it.
It's VERY easy to just add a new property of type int, but when a MongoDB document doesn't have that property you're going to have a runtime mapping exception so you'll have to make it int? instead, etc.
Also, forget normalization.
If you have a highly normalized data structure, MongoDB is going to make life harder.
So really think about what data you can store with redundancy and what data you can't.
Also, I found querying a MongoDB a lot harder, but that's probably only because I'm not used to the syntax
I'm not used to seeing you use popular technology by the way, unless you're going to use this for your LALR X-PARSE B-TREE REGEX EXPRESSION COMPILER +5
|
|
|
|
|
so in the "middle" of the "middle field". Nice position
Press F1 for help or google it.
Greetings from Germany
|
|
|
|
|
This is a nice write-up and has some valuable points.
I would also add that yes, MongoDB is quite popular based on the fact that it had an IPO and you can buy stock in the company: MongoDB current stock price[^]
That may sound a bit ridiculous but I mean it does show a bit of consumer and investor confidence and that the company itself seems to be solid.
|
|
|
|
|
Sander Rossel wrote: you're going to have a runtime mapping exception
Only if you're using a strongly typed language. Javascript (or even TypeScript) with NodeJS does not have that problem, although I do agree you still need to think about it.
To alcohol! The cause of, and solution to, all of life's problems - Homer Simpson
Our heads are round so our thoughts can change direction - Francis Picabia
|
|
|
|
|
Yeah, it's not a problem if you use NodeJS, but I'm assuming honey is using C#
Of course you still can't just add number2 after some weeks of production (and always set it for new records) and then somewhere do number1 + number2, assuming they both have a value.
Even if number1 always has a value, number2 only gets a value at some point in time, unless you do a collection update like you'd do in a SQL database.
When I worked with MongoDB we used to do that because updating the entire collection is easier than updating the entire software
For the record, I didn't think we needed MongoDB, SQL would've been fine, especially when you want schema integrity anyway, but it was the decision of our architect.
At least it gave me a chance to work with MongoDB in a production environment.
|
|
|
|
|
CouchDB is another alternative, but MongoDB is the one I hear mentioned most often.
Having said that, a lot of those mentions are on infosec news, when someone's left their MongoDB store up, unprotected, on Amazon S3 and the data in it has been compromised... That's not MongoDB's fault, of course
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
|
|
|
|
|
Stuart Dootson wrote: That's not MongoDB's fault, of course Well, in part.
Their default setting is (was?) unprotected and open for all to see
That's a very dubious default and one that many people failed to change!
|
|
|
|
|
Sander Rossel wrote: Their default setting is (was?) unprotected and open for all to see
The reports I've heard were implying more that the S3 permissions had been defaulted, so anyone could access (and download) the datastore rather than accessing the associated MongoDB server.
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
|
|
|
|
|
I may be thinking of another incident where researchers just tried the default MongoDB port on thousands of servers and found that for many of those servers the port was open and the database was exposed.
|
|
|
|
|
|
|
Quote: MongoDB is a general purpose, document-based, distributed database built for modern application developers and for the cloud era.
However, they also say:
Quote: No database makes you more productive.
So I guess simple text files are the future.
|
|
|
|
|
MongoDB - where GUIDs are displayed in the wrong order just to punish you.
2/5 would not use again.
|
|
|
|
|
I wasn't asking for a review
Real programmers use butterflies
|
|
|
|
|
|
Because 1.x existed prior to generics we have issues of legacy object models not implementing IList<T> and instead simply exposing hard typed indexer properties.
Normally, you'd just get the generic parameters of the generic IEnumerable<T> interface, but because some object trees were created prior to 2.0 - like the CodeDOM they don't have them.
This makes determining the element type of a typed list extremely difficult. The problem is that your alternative is the indexer property which isn't a member of a hard interface, so you have to select the appropriate indexer property from the properties on that type. There might be this[string name] in there too, for example. There is no contract however, so there are no guarantees. This isn't especially robust.
Which means, the obvious solution is to first try to get it using the generic interfaces, and if they aren't available, then we fall back to the less robust method above.
This is not ideal, and it requires maybe a page of code to handle all the scenarios.
Microsoft didn't put generics into 1.x I think because of time constraints, and if so they should have waited, IMO.
Edit: For anyone interested I just posted a tip that solves this problem. How To Get A Collection Element Type Using Reflection in C#[^]
Real programmers use butterflies
modified 5-May-20 15:08pm.
|
|
|
|
|
I agree.
V1 should never have seen the light of day.
And sub-teams working on different areas should have communicated, and there should have been a core team to determine commonalities.
I mean, seriously!?
public ArgumentException (string message, string paramName);
public ArgumentOutOfRangeException (string paramName, string message);
public ArgumentNullException (string paramName, string message);
modified 5-May-20 14:38pm.
|
|
|
|
|
Amen!
Real programmers use butterflies
|
|
|
|
|
Diversity is our strength!
|
|
|
|
|
"Keep 'em guessing" supports the training industry.
|
|
|
|
|
Do you mean there are three argument Exceptions (which I can understand) or that message and paramName are switched in two of them?
|
|
|
|
|
I suspect option 2.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
The inconsistency in parameter order.
|
|
|
|
|
1.x was just "beta" to play around with, it was never intended for production code, but *hands up* the company I worked for at the time did indeed use it in production. We used sockets a fair bit to communication with UNIX systems and also with banking applications, and .net had socket classes built in so we were keen to migrate to it ASAP.
|
|
|
|