Search Results

Search found 5295 results on 212 pages for 'transaction scope'.

Page 31/212 | < Previous Page | 27 28 29 30 31 32 33 34 35 36 37 38  | Next Page >

  • How to access method variables from within an anonymous function in JavaScript?

    - by Hussain
    I'm writing a small ajax class for personal use. In the class, I have a "post" method for sending post requests. The post method has a callback parameter. In the onreadystatechange propperty, I need to call the callback method. Something like this: this.requestObject.onreadystatechange = function() { callback(this.responseText); } However, I can't access the callback variable from within the anonomous function. How can I bring the callback variable into the scope of the onreadystatechange anonomous function? edit: Here's the full code so far: function request() { this.initialize = function(errorHandeler) { try { try { this.requestObject = new XDomainRequest(); } catch(e) { try { this.requestObject = new XMLHttpRequest(); } catch (e) { try { this.requestObject = new ActiveXObject("Msxml2.XMLHTTP"); //newer versions of IE5+ } catch (e) { this.requestObject = new ActiveXObject("Microsoft.XMLHTTP"); //older versions of IE5+ } } } } catch(e) { errorHandeler(); } } this.post = function(url,data) { var response;var escapedData = ""; if (typeof data == 'object') { for (i in data) { escapedData += escape(i)+'='+escape(data[i])+'&'; } escapedData = escapedData.substr(0,escapedData.length-1); } else { escapedData = escape(data); } this.requestObject.open('post',url,true); this.requestObject.setRequestHeader("Content-type", "application/x-www-form-urlencoded"); this.requestObject.setRequestHeader("Content-length", data.length); this.requestObject.setRequestHeader("Connection", "close"); this.requestObject.onreadystatechange = function() { if (this.readyState == 4) { // call callback function } } this.requestObject.send(data); }

    Read the article

  • named_scope + average is causing the table to be specified more then once in the sql query run on po

    - by hadees
    I have a named scopes like so... named_scope :gender, lambda { |gender| { :joins => {:survey_session => :profile }, :conditions => { :survey_sessions => { :profiles => { :gender => gender } } } } } and when I call it everything works fine. I also have this average method I call... Answer.average(:rating, :include => {:survey_session => :profile}, :group => "profiles.career") which also works fine if I call it like that. However if I were to call it like so... Answer.gender('m').average(:rating, :include => {:survey_session => :profile}, :group => "profiles.career") I get... ActiveRecord::StatementInvalid: PGError: ERROR: table name "profiles" specified more than once : SELECT avg("answers".rating) AS avg_rating, profiles.career AS profiles_career FROM "answers" LEFT OUTER JOIN "survey_sessions" survey_sessions_answers ON "survey_sessions_answers".id = "answers".survey_session_id LEFT OUTER JOIN "profiles" ON "profiles".id = "survey_sessions_answers".profile_id INNER JOIN "survey_sessions" ON "survey_sessions".id = "answers".survey_session_id INNER JOIN "profiles" ON "profiles".id = "survey_sessions".profile_id WHERE ("profiles"."gender" = E'm') GROUP BY profiles.career Which is a little hard to read but says I'm including the table profiles twice. If I were to just remove the include from average it works but it isn't really practical because average is actually being called inside a method which gets passed the scoped. So there is some times gender or average might get called with out each other and if either was missing the profile include it wouldn't work. So either I need to know how to fix this apparent bug in Rails or figure out a way to know what scopes were applied to a ActiveRecord::NamedScope::Scope object so that I could check to see if they have been applied and if not add the include for average.

    Read the article

  • Design: How to declare a specialized memory handler class

    - by Michael Dorgan
    On an embedded type system, I have created a Small Object Allocator that piggy backs on top of a standard memory allocation system. This allocator is a Boost::simple_segregated_storage< class and it does exactly what I need - O(1) alloc/dealloc time on small objects at the cost of a touch of internal fragmentation. My question is how best to declare it. Right now, it's scope static declared in our mem code module, which is probably fine, but it feels a bit exposed there and is also now linked to that module forever. Normally, I declare it as a monostate or a singleton, but this uses the dynamic memory allocator (where this is located.) Furthermore, our dynamic memory allocator is being initialized and used before static object initialization occurs on our system (as again, the memory manager is pretty much the most fundamental component of an engine.) To get around this catch 22, I added an extra 'if the small memory allocator exists' to see if the small object allocator exists yet. That if that now must be run on every small object allocation. In the scheme of things, this is nearly negligable, but it still bothers me. So the question is, is there a better way to declare this portion of the memory manager that helps decouple it from the memory module and perhaps not costing that extra isinitialized() if statement? If this method uses dynamic memory, please explain how to get around lack of initialization of the small object portion of the manager.

    Read the article

  • Emulating Test::More::done_testing - what is the most idiomatic way?

    - by DVK
    I have to build unit tests for in environment with a very old version of Test::More (perl5.8 with $Test::More::VERSION being '0.80') which predates the addition of done_testing(). Upgrading to newer Test::More is out of the question for practical reasons. And I am trying to avoid using no_tests - it's generally a bad idea not catching when your unit test dies prematurely. What is the most idiomatic way of running a configurable amount of tests, assuming no no_tests or done_testing() is used? Details: My unit tests usually take the form of: use Test::More; my @test_set = ( [ "Test #1", $param1, $param2, ... ] ,[ "Test #1", $param1, $param2, ... ] # ,... ); foreach my $test (@test_set) { run_test($test); } sub run_test { # $expected_tests += count_tests($test); ok(test1($test)) || diag("Test1 failed"); # ... } The standard approach of use Test::More tests => 23; or BEGIN {plan tests => 23} does not work since both are obviously executed before @tests is known. My current approach involves making @tests global and defining it in the BEGIN {} block as follows: use Test::More; BEGIN { our @test_set = (); # Same set of tests as above my $expected_tests = 0; foreach my $test (@tests) { my $expected_tests += count_tests($test); } plan tests = $expected_tests; } our @test_set; # Must do!!! Since first "our" was in BEGIN's scope :( foreach my $test (@test_set) { run_test($test); } # Same sub run_test {} # Same I feel this can be done more idiomatically but not certain how to improve. Chief among the smells is the duplicate our @test_test declarations - in BEGIN{} and after it.

    Read the article

  • Associative Array / Object can't be read in functions

    - by Matrym
    At the very beginning of the javascript file, I have: var lbp = {}; lbp.defaults = { minLength: 40 }; I can successfully alert it afterwards, with: alert(lbp.defaults.minLength); But as soon as I put it inside a function, when I alert, I get "Undefined". What gives, and how do I avoid this? Is it absolutely necessary to pass this variable into each function, for example, by doing: function(lbp) { alert(lbp.defaults.minLength); } I would have thought that defining it first, it would attain global scope and not be required to be passed in? Thanks in advance for enlightening me :) ==================================== EDIT: The problem seems like it might be my initialize function is itself defined within lbp. Is there any way to use this function var, and still use lbp vars inside it? lbp.initialize = function() { alert(lbp.defaults.minLength); }; The full bit of code looks like this: <script type="text/javascript"> var lbp = { defaults: { minLength: 40 } }; lbp.initialize = function() { alert(lbp.defaults.minLength); }; window.onload = lbp.initialize; </script>

    Read the article

  • Change a finder method w/ parameters to an association

    - by Sai Emrys
    How do I turn this into a has_one association? (Possibly has_one + a named scope for size.) class User < ActiveRecord::Base has_many :assets, :foreign_key => 'creator_id' def avatar_asset size = :thumb # The LIKE is because it might be a .jpg, .png, or .gif. # More efficient methods that can handle that are OK. ;) self.assets.find :first, :conditions => ["thumbnail = '#{size}' and filename LIKE ?", self.login + "_#{size}.%"] end end EDIT: Cuing from AnalogHole on Freenode #rubyonrails, we can do this: has_many :assets, :foreign_key => 'creator_id' do def avatar size = :thumb find :first, :conditions => ["thumbnail = ? and filename LIKE ?", size.to_s, proxy_owner.login + "_#{size}.%"] end end ... which is fairly cool, and makes syntax a bit better at least. However, this still doesn't behave as well as I would like. Particularly, it doesn't allow for further nice find chaining (such that it doesn't execute this find until it's gotten all its conditions). More importantly, it doesn't allow for use in an :include. Ideally I want to do something like this: PostsController def show post = Post.get_cache(params[:id]) { Post.find(params[:id], :include => {:comments => {:users => {:avatar_asset => :thumb}} } ... end ... so that I can cache the assets together with the post. Or cache them at all, really - e.g. get_cache(user_id){User.find(user_id, :include => :avatar_assets)} would be a good first pass. This doesn't actually work (self == User), but is correct in spirit: has_many :avatar_assets, :foreign_key => 'creator_id', :class_name => 'Asset', :conditions => ["filename LIKE ?", self.login + "_%"] (Also posted on Refactor My Code.)

    Read the article

  • Limiting a search to records from last_request_at...

    - by bgadoci
    I am trying to figure out how to display a count for records that have been created in a table since the last_request_at of a user. In my view I am counting the notes of a question with the following code: <% unless @questions.empty? %> <% @questions.each do |question| %> <%= h(question.notes.count) %> end end This is happening in the /views/users/show.html.erb file. Instead of counting all the notes for the question, I would only like to count the notes that have been created since the users last_request_at datetime. I don't neccessarily want to scope notes to display this 'new notes' count application wide, just simply in this one instance. To accomplish I am assuming I need to create a variable in the User#show action and call it in the view but not really sure how to do that. Other information you may need: class Note < ActiveRecord::Base belongs_to :user belongs_to :question end class Question < ActiveRecord::Base has_many :notes, :dependent => :destroy belongs_to :user end

    Read the article

  • Which of the following Java coding fragments is better?

    - by Simon
    This isn't meant to be subjective, I am looking for reasons based on resource utilisation, compiler performance, GC performance etc. rather than elegance. Oh, and the position of brackets doesn't count, so no stylistic comments please. Take the following loop; Integer total = new Integer(0); Integer i; for (String str : string_list) { i = Integer.parse(str); total += i; } versus... Integer total = 0; for (String str : string_list) { Integer i = Integer.parse(str); total += i; } In the first one i is function scoped whereas in the second it is scoped in the loop. I have always thought (believed) that the first one would be more efficient because it just references an existing variable already allocated on the stack, whereas the second one would be pushing and popping i each iteration of the loop. There are quite a lot of other cases where I tend to scope variables more broadly than perhaps necessary so I thought I would ask here to clear up a gap in my knowledge. Also notice that assignment of the variable on initialisation either involving the new operator or not. Do any of these sorts of semi-stylistic semi-optimisations make any difference at all?

    Read the article

  • Variable Scoping in a method and its persistence in C++

    - by de costo
    Consider the following public method that adds an integer variable to a vector of ints(private member) in a class in C++. KoolMethod() { int x; x = 10; KoolList.Add(x); } Vector<int>KoolList; But is this a valid addition to a vector ??? Upon calling the method, it creates a local variable. The scope of this local variable ends the moment the execution control leaves the method. And since this local variable is allocated on a stack(on the method call), any member of KoolList points to an invalid memory location in deallocated stack which may or may not contain the expected value of x. Is this an accurate description of above mechanism ?? Is there a need for creating an int in heap storage using "new" operator everytime a value needs to be added to the vector like described below ????: KoolMethod() { int *x = new int(); *x = 10; KoolList.Add(x); } Vector<int*>KoolList;

    Read the article

  • C vs C++ function questions

    - by james
    I am learning C, and after starting out learning C++ as my first compiled language, I decided to "go back to basics" and learn C. There are two questions that I have concerning the ways each language deals with functions. Firstly, why does C "not care" about the scope that functions are defined in, whereas C++ does? For example, int main() { donothing(); return 0; } void donothing() { } the above will not compile in a C++ compiler, whereas it will compile in a C compiler. Why is this? Isn't C++ mostly just an extension on C, and should be mostly "backward compatible"? Secondly, the book that I found (Link to pdf) does not seem to state a return type for the main function. I check around and found other books and websites and these also commonly do not specify return types for the main function. If I try to compile a program that does not specify a return type for main, it compiles fine (although with some warnings) in a C compiler, but it doesn't compile in a C++ compiler. Again, why is that? Is it better style to always specify the return type as an integer rather than leaving it out? Thanks for any help, and just as a side note, if anyone can suggest a better book that I should buy that would be great!

    Read the article

  • NHibernate transactions randomly not rolled back

    - by cbp
    I have a suite of integration tests that run inside transactions. Sometimes it seems that NHibernate transactions are not being correctly rolled back. I can't work out what causes this. Here is a slightly simplified overview of the base class that these integration test fixtures run with: public class IntegrationTestFixture { private TransactionScope _transactionScope; private ConnectionScope _connectionScope; [TestFixtureSetUp] public virtual void TestFixtureSetUp() { var session = NHibernateSessionManager.SessionFactory.OpenSession(); CallSessionContext.Bind(session); _connectionScope = new ConnectionScope(); _transactionScope = new TransactionScope(); } [TestFixtureTearDown] public virtual void TestFixtureTearDown() { _transactionScope.Dispose(); _connectionScope.Dispose(); var session = CurrentSessionContext.Unbind(SessionFactory); session.Close(); session.Dispose(); } } A call to the TransactionScope's commit method is never made, therefore how is it possible that data still ends up in the database?

    Read the article

  • MS SQL Server BEGIN/END vs BEGIN TRANS/COMMIT/ROLLBACK

    - by Rich
    I have been trying to find info on the web about the differences between these statements, and it seems to me they are identical but I can't find confirmation of that or any kind of comparison between the two. What is the difference between doing this: BEGIN -- Some update, insert, set statements END and doing this BEGIN TRANS -- Some update, insert, set statements COMMIT TRANS ? Note that there is only the need to rollback in the case of some exception or timeout or other general failure, there would not be a conditional reason to rollback.

    Read the article

  • Automatic e-mail processing

    - by Jon Harrop
    I'd like to write a .NET application in F# to automate some of the processing of my e-mails. For example, when an order comes in my program might compute a new htpasswd from the e-mail's contents, upload it to our web server and reply to the customer with login details. How do people do this? I've tried Outlook 2007 automation but it just prompts the user for security and my attempts to get it to stop doing this have failed so I cannot automate anything with it. Is there a .NET-friendly e-mail client I can use more easily? This has been so tedious that I'm seriously considering writing my own .NET-friendly e-mail client...

    Read the article

  • Why qry.post executed with asynchronous mode?

    - by Ryan
    Recently I met a strange problem, see code snips as below: var sqlCommand: string; connection: TADOConnection; qry: TADOQuery; begin connection := TADOConnection.Create(nil); try connection.ConnectionString := 'Provider=Microsoft.Jet.OLEDB.4.0;Data Source=Test.MDB;Persist Security Info=False'; connection.Open(); qry := TADOQuery.Create(nil); try qry.Connection := connection; qry.SQL.Text := 'Select * from aaa'; qry.Open; qry.Append; qry.FieldByName('TestField1').AsString := 'test'; qry.Post; beep; finally qry.Free; end; finally connection.Free; end; end; First, Create a new access database named test.mdb and put it under the directory of this test project, we can create a new table named aaa in it which has only one text type field named TestField1. We set a breakpoint at line of "beep", then lunch the test application under ide debug mode, when ide stops at the breakpoint line (qry.post has been executed), at this time we use microsoft access to open test.mdb and open table aaa you will find there are no any changes in table aaa, if you let the ide continue running after pressing f9 you can find a new record is inserted in to table aaa, but if you press ctrl+f2 to terminate the application at the breakpoint, you will find the table aaa has no record been inserted, but in normal circumstance, a new record should be inserted in to the table aaa after qry.post executed. who can explain this problem , it troubles me so long time. thanks !!! BTW, the ide is delphi 2010, and the access mdb file is created by microsoft access 2007 under windows 7

    Read the article

  • "Undoing deletes" in webapplication?

    - by Industrial
    Hi everybody, I have seen more and more of the websites that offers a undo option after pressing a delete button. How is the logic done behind the button? Is the item deleted by javascript and "dissapears" from the users screen and a scheduled delete added, that gives the user time to undo it or how does it work? What are the other options to offer the users an undo feature?

    Read the article

  • Why doesn't access database update?

    - by Ryan
    Recently I met a strange problem, see code snips as below: var sqlCommand: string; connection: TADOConnection; qry: TADOQuery; begin connection := TADOConnection.Create(nil); try connection.ConnectionString := 'Provider=Microsoft.Jet.OLEDB.4.0;Data Source=Test.MDB;Persist Security Info=False'; connection.Open(); qry := TADOQuery.Create(nil); try qry.Connection := connection; qry.SQL.Text := 'Select * from aaa'; qry.Open; qry.Append; qry.FieldByName('TestField1').AsString := 'test'; qry.Post; beep; finally qry.Free; end; finally connection.Free; end; end; First, Create a new access database named test.mdb and put it under the directory of this test project, we can create a new table named aaa in it which has only one text type field named TestField1. We set a breakpoint at line of "beep", then lunch the test application under ide debug mode, when ide stops at the breakpoint line (qry.post has been executed), at this time we use microsoft access to open test.mdb and open table aaa you will find there are no any changes in table aaa, if you let the ide continue running after pressing f9 you can find a new record is inserted in to table aaa, but if you press ctrl+f2 to terminate the application at the breakpoint, you will find the table aaa has no record been inserted, but in normal circumstance, a new record should be inserted in to the table aaa after qry.post executed. who can explain this problem , it troubles me so long time. thanks !!! BTW, the ide is delphi 2010, and the access mdb file is created by microsoft access 2007 under windows 7

    Read the article

  • db2 jdbc driver does not release table locks

    - by as
    situation: We have a web service running on tomcat accessing DB2 database on AS400, we are using JTOPEN drivers for JNDI connections handled by tomcat. For handling transactions and access to database we are using Spring. For each select system takes JDBC connection from JNDI (i.e. from connection pool), does selection, and in the end it closes ResultSet, Statement and releases Connection in that order. That passes fine, shared lock on table dissappears. When we want to do update the same way as we did with select (exception on ResultSet object, we don't have one in such situation), after releasing Connection to JNDI lock on table stays. If we put maxIdle=0 for number of connections in JNDI configuration, this problem disappears, but this degrades performances, we have cca 100 users online on that service, we need few connections to be alive in pool. What do you suggest?

    Read the article

  • Spring @Transactional Annotation Best Practice

    - by Thomas Einwaller
    We are currently discussing the Best Practice for placing the @Transactional annotations in our code. Do you place the @Transactional in the DAO classes and/or their methods or is it better to annotate the Service classed which are calling using the DAO objects? Or does it make sense to annotate both "layers"?

    Read the article

  • django multiprocess problem

    - by iKiR
    I have django application, running under lighttpd via fastcgi. FCGI running script looks like: python manage.py runfcgi socket=<path>/main.socket method=prefork \ pidfile=<path>/server.pid \ minspare=5 maxspare=10 maxchildren=10 maxrequests=500 \ I use SQLite. So I have 10 proccess, which all work with the same DB. Next I have 2 views: def view1(request) ... obj = MyModel.objects.get_or_create(id=1) obj.param1 = <some value> obj.save () def view2(request) ... obj = MyModel.objects.get_or_create(id=1) obj.param2 = <some value> obj.save () And If this views are executed in two different threads sometimes I get MyModel instance in DB with id=1 and updated either param1 or param2 (BUT not both) - it depends on which process was the first. (of course in real life id changes, but sometimes 2 processes execute these two views with same id) The question is: What should I do to get instance with updated param1 and param2? I need something for merging changes in different processes. One decision is create interprocess lock object but in this case I will get sequence executing views and they will not be able to be executed simultaneously, so I ask help

    Read the article

  • How do I delete in Django? (mysql transactions)

    - by alex
    If you are familiar with Django, you know that they have a Authentication system with User model. Of course, I have many other tables that have a Foreign Key to this User model. If I want to delete this user, how do I architect a script (or through mysql itself) to delete every table that is related to this user? My only worry is that I can do this manually...but if I add a table , but I forget to add that table to my DELETE operation...then I have a row that links to a deleted, non-existing User.

    Read the article

  • SQL Server "Long running transaction" performance counter: why no workee?

    - by Sleepless
    Please explain to me the following observation: I have the following piece of T-SQL code that I run from SSMS: BEGIN TRAN SELECT COUNT (*) FROM m WHERE m.[x] = 123456 or m.[y] IN (SELECT f.x FROM f) SELECT COUNT (*) FROM m WHERE m.[x] = 123456 or m.[y] IN (SELECT f.x FROM f) COMMIT TRAN The query takes about twenty seconds to run. I have no other user queries running on the server. Under these circumstances, I would expect the performance counter "MSSQL$SQLInstanceName:Transactions\Longest Transaction Running Time" to rise constantly up to a value of 20 and then drop rapidly. Instead, it rises to around 12 within two seconds and then oscillates between 12 and 14 for the duration of the query after which it drops again. According to the MS docs, the counter measures "The length of time (in seconds) since the start of the transaction that has been active longer than any other current transaction." But apparently, it doesn't. What gives?

    Read the article

  • Saving child collections with NHibernate

    - by Ben
    Hi, I am in the process or learning NHibernate so bare with me. I have an Order class and a Transaction class. Order has a one to many association with transaction. The transaction table in my database has a not null constraint on the OrderId foreign key. Order class: public class Order { public virtual Guid Id { get; set; } public virtual DateTime CreatedOn { get; set; } public virtual decimal Total { get; set; } public virtual ICollection<Transaction> Transactions { get; set; } public Order() { Transactions = new HashSet<Transaction>(); } } Order Mapping: <class name="Order" table="Orders"> <cache usage="read-write"/> <id name="Id"> <generator class="guid"/> </id> <property name="CreatedOn" type="datetime"/> <property name="Total" type="decimal"/> <set name="Transactions" table="Transactions" lazy="false" inverse="true"> <key column="OrderId"/> <one-to-many class="Transaction"/> </set> Transaction Class: public class Transaction { public virtual Guid Id { get; set; } public virtual DateTime ExecutedOn { get; set; } public virtual bool Success { get; set; } public virtual Order Order { get; set; } } Transaction Mapping: <class name="Transaction" table="Transactions"> <cache usage="read-write"/> <id name="Id" column="Id" type="Guid"> <generator class="guid"/> </id> <property name="ExecutedOn" type="datetime"/> <property name="Success" type="bool"/> <many-to-one name="Order" class="Order" column="OrderId" not-null="true"/> Really I don't want a bidirectional association. There is no need for my transaction objects to reference their order object directly (I just need to access the transactions of an order). However, I had to add this so that Order.Transactions is persisted to the database: Repository: public void Update(Order entity) { using (ISession session = NHibernateHelper.OpenSession()) { using (ITransaction transaction = session.BeginTransaction()) { session.Update(entity); foreach (var tx in entity.Transactions) { tx.Order = entity; session.SaveOrUpdate(tx); } transaction.Commit(); } } } My problem is that this will then issue an update for every transaction on the order collection (regardless of whether it has changed or not). What I was trying to get around was having to explicitly save the transaction before saving the order and instead just add the transactions to the order and then save the order: public void Can_add_transaction_to_existing_order() { var orderRepo = new OrderRepository(); var order = orderRepo.GetById(new Guid("aa3b5d04-c5c8-4ad9-9b3e-9ce73e488a9f")); Transaction tx = new Transaction(); tx.ExecutedOn = DateTime.Now; tx.Success = true; order.Transactions.Add(tx); orderRepo.Update(order); } Although I have found quite a few articles covering the set up of a one-to-many association, most of these discuss retrieving of data and not persisting back. Many thanks, Ben

    Read the article

  • Of transactions and Mongo

    - by Nuri Halperin
    Originally posted on: http://geekswithblogs.net/nuri/archive/2014/05/20/of-transactions-and-mongo-again.aspxWhat's the first thing you hear about NoSQL databases? That they lose your data? That there's no transactions? No joins? No hope for "real" applications? Well, you *should* be wondering whether a certain of database is the right one for your job. But if you do so, you should be wondering that about "traditional" databases as well! In the spirit of exploration let's take a look at a common challenge: You are a bank. You have customers with accounts. Customer A wants to pay B. You want to allow that only if A can cover the amount being transferred. Let's looks at the problem without any context of any database engine in mind. What would you do? How would you ensure that the amount transfer is done "properly"? Would you prevent a "transaction" from taking place unless A can cover the amount? There are several options: Prevent any change to A's account while the transfer is taking place. That boils down to locking. Apply the change, and allow A's balance to go below zero. Charge person A some interest on the negative balance. Not friendly, but certainly a choice. Don't do either. Options 1 and 2 are difficult to attain in the NoSQL world. Mongo won't save you headaches here either. Option 3 looks a bit harsh. But here's where this can go: ledger. See, and account doesn't need to be represented by a single row in a table of all accounts with only the current balance on it. More often than not, accounting systems use ledgers. And entries in ledgers - as it turns out – don't actually get updated. Once a ledger entry is written, it is not removed or altered. A transaction is represented by an entry in the ledger stating and amount withdrawn from A's account and an entry in the ledger stating an addition of said amount to B's account. For sake of space-saving, that entry in the ledger can happen using one entry. Think {Timestamp, FromAccountId, ToAccountId, Amount}. The implication of the original question – "how do you enforce non-negative balance rule" then boils down to: Insert entry in ledger Run validation of recent entries Insert reverse entry to roll back transaction if validation failed. What is validation? Sum up the transactions that A's account has (all deposits and debits), and ensure the balance is positive. For sake of efficiency, one can roll up transactions and "close the book" on transactions with a pseudo entry stating balance as of midnight or something. This lets you avoid doing math on the fly on too many transactions. You simply run from the latest "approved balance" marker to date. But that's an optimization, and premature optimizations are the root of (some? most?) evil.. Back to some nagging questions though: "But mongo is only eventually consistent!" Well, yes, kind of. It's not actually true that Mongo has not transactions. It would be more descriptive to say that Mongo's transaction scope is a single document in a single collection. A write to a Mongo document happens completely or not at all. So although it is true that you can't update more than one documents "at the same time" under a "transaction" umbrella as an atomic update, it is NOT true that there' is no isolation. So a competition between two concurrent updates is completely coherent and the writes will be serialized. They will not scribble on the same document at the same time. In our case - in choosing a ledger approach - we're not even trying to "update" a document, we're simply adding a document to a collection. So there goes the "no transaction" issue. Now let's turn our attention to consistency. What you should know about mongo is that at any given moment, only on member of a replica set is writable. This means that the writable instance in a set of replicated instances always has "the truth". There could be a replication lag such that a reader going to one of the replicas still sees "old" state of a collection or document. But in our ledger case, things fall nicely into place: Run your validation against the writable instance. It is guaranteed to have a ledger either with (after) or without (before) the ledger entry got written. No funky states. Again, the ledger writing *adds* a document, so there's no inconsistent document state to be had either way. Next, we might worry about data loss. Here, mongo offers several write-concerns. Write-concern in Mongo is a mode that marshals how uptight you want the db engine to be about actually persisting a document write to disk before it reports to the application that it is "done". The most volatile, is to say you don't care. In that case, mongo would just accept your write command and say back "thanks" with no guarantee of persistence. If the server loses power at the wrong moment, it may have said "ok" but actually no written the data to disk. That's kind of bad. Don't do that with data you care about. It may be good for votes on a pole regarding how cute a furry animal is, but not so good for business. There are several other write-concerns varying from flushing the write to the disk of the writable instance, flushing to disk on several members of the replica set, a majority of the replica set or all of the members of a replica set. The former choice is the quickest, as no network coordination is required besides the main writable instance. The others impose extra network and time cost. Depending on your tolerance for latency and read-lag, you will face a choice of what works for you. It's really important to understand that no data loss occurs once a document is flushed to an instance. The record is on disk at that point. From that point on, backup strategies and disaster recovery are your worry, not loss of power to the writable machine. This scenario is not different from a relational database at that point. Where does this leave us? Oh, yes. Eventual consistency. By now, we ensured that the "source of truth" instance has the correct data, persisted and coherent. But because of lag, the app may have gone to the writable instance, performed the update and then gone to a replica and looked at the ledger there before the transaction replicated. Here are 2 options to deal with this. Similar to write concerns, mongo support read preferences. An app may choose to read only from the writable instance. This is not an awesome choice to make for every ready, because it just burdens the one instance, and doesn't make use of the other read-only servers. But this choice can be made on a query by query basis. So for the app that our person A is using, we can have person A issue the transfer command to B, and then if that same app is going to immediately as "are we there yet?" we'll query that same writable instance. But B and anyone else in the world can just chill and read from the read-only instance. They have no basis to expect that the ledger has just been written to. So as far as they know, the transaction hasn't happened until they see it appear later. We can further relax the demand by creating application UI that reacts to a write command with "thank you, we will post it shortly" instead of "thank you, we just did everything and here's the new balance". This is a very powerful thing. UI design for highly scalable systems can't insist that the all databases be locked just to paint an "all done" on screen. People understand. They were trained by many online businesses already that your placing of an order does not mean that your product is already outside your door waiting (yes, I know, large retailers are working on it... but were' not there yet). The second thing we can do, is add some artificial delay to a transaction's visibility on the ledger. The way that works is simply adding some logic such that the query against the ledger never nets a transaction for customers newer than say 15 minutes and who's validation flag is not set. This buys us time 2 ways: Replication can catch up to all instances by then, and validation rules can run and determine if this transaction should be "negated" with a compensating transaction. In case we do need to "roll back" the transaction, the backend system can place the timestamp of the compensating transaction at the exact same time or 1ms after the original one. Effectively, once A or B visits their ledger, both transactions would be visible and the overall balance "as of now" would reflect no change.  The 2 transactions (attempted/ reverted) would be visible , since we do actually account for the attempt. Hold on a second. There's a hole in the story: what if several transfers from A to some accounts are registered, and 2 independent validators attempt to compute the balance concurrently? Is there a chance that both would conclude non-sufficient-funds even though rolling back transaction 100 would free up enough for transaction 117 (some random later transaction)? Yes. there is that chance. But the integrity of the business rule is not compromised, since the prime rule is don't dispense money you don't have. To minimize or eliminate this scenario, we can also assign a single validation process per origin account. This may seem non-scalable, but it can easily be done as a "sharded" distribution. Say we have 11 validation threads (or processing nodes etc.). We divide the account number space such that each validator is exclusively responsible for a certain range of account numbers. Sounds cunningly similar to Mongo's sharding strategy, doesn't it? Each validator then works in isolation. More capacity needed? Chop the account space into more chunks. So where  are we now with the nagging questions? "No joins": Huh? What are those for? "No transactions": You mean no cross-collection and no cross-document transactions? Granted - but don't always need them either. "No hope for real applications": well... There are more issues and edge cases to slog through, I'm sure. But hopefully this gives you some ideas of how to solve common problems without distributed locking and relational databases. But then again, you can choose relational databases if they suit your problem.

    Read the article

< Previous Page | 27 28 29 30 31 32 33 34 35 36 37 38  | Next Page >