Search Results

Search found 2767 results on 111 pages for 'concept'.

Page 100/111 | < Previous Page | 96 97 98 99 100 101 102 103 104 105 106 107  | Next Page >

  • How to perform add/update of a model object that contains EntitySet

    - by David Liddle
    I have a similar concept to the SO questions/tags scenario however am trying to decide the best way of implementation. Tables Questions, QuestionTags and Tags Questions QuestionTags Tags --------- ------------ ---- QID QID TID QName TID TName When adding/updating a question I have 2 textboxes. The important part is a single textbox that allows users to enter in multiple Tags separated by spaces. I am using Linq2Sql so the Questions model has an EntitySet of QuestionTags with then link to Tags. My question is regarding the adding/updating of Questions (part 1), and also how to best show QuestionTags for a Question (part 2). Part 1 Before performing an add/update, my service layer needs to deal with 3 scenarios before passing to their respective repositories. Insert Tags that do not already exist Insert/Update Question Insert QuestionTags - when updating need to remove existing QuestionTags Here is my code below however started to get into a bit of a muddle. I've created extension methods on my repositories to get Tags WithNames etc. public void Add(Question q, string tags) { var tagList = tags.Split(new string[] { " " }, StringSplitOptions.RemoveEmptyEntries).ToList(); using (DB.TransactionScope ts = new DB.TransactionScope()) { var existingTags = TagsRepository.Get() .WithName(tagList) .ToList(); var newTags = (from t in tagList select new Tag { TName = t }).Except(existingTags, new TagsComparer()).ToList(); TagsRepository.Add(newTags); //need to insert QuestionTags QuestionsRepository.Add(q); ts.Complete(); } } Part 2 My second question is, when displaying a list of Questions how is it best to show their QuestionTags? For example, I have an Index view that shows a list of Questions in a table. One of the columns shows an image and when the user hovers over it shows the list of Tags. My current implementation is to create a custom ViewModel and show a List of QuestionIndexViewModel in the View. QuestionIndexViewModel { Question Question { get; set; } string Tags { get; set; } } However, this seems a bit clumsy and quite a few DB calls. public ViewResult Index() { var model= new List<QuestionIndexViewModel>(); //make a call to get a list of questions //foreach question make a call to get their QuestionTags, //to be able to get their Tag names and then join them //to form a single string. return View(model); } Also, just for test purposes using SQL Profiler, I decided to iterate through the QuestionTags entity set of a Question in my ViewModel however nothing was picked up in Profiler? What would be the reason for this?

    Read the article

  • WndProc(ref Message m), Prevent minimize Games, Send key strokes.

    - by Stanomatic
    Overview: I am going to create a touch application that interfaces with games and other apps. This concept is similar to the app found on touch-buddy.com but I will be using C# and WPF instead of how the application is written in Perl. I have a few challenges I would like to evaluate. The touch-buddy app uses two approaches while interacting with games; 1. Client mode (Same machine runs both game and touch-buddy). 2. Server / Client mode where a separate box sends commands to the game machine. The reason I believe for this method was to circumvent the issue with games minimizing. In Client only mode I am faced with the issue where I touch a screen OTHER than the main screen where the game is viewed and then the game minimizes. Not all games have this behavior but I would like to conquer the games that do minimize and prevent it. Is it possible to keep a game front and center Focused and prevent minimizing utilizing C# WndProc(ref Message m)? I have been experimenting with WndProc(ref Message m) where I created a win form and when I press minimize on my own Win form and it will close an instance of notepad. This proves to me that I can capture a message, prevent that message from bubbling up and then send a message to another application. I then tried to click on notepad with my touch screen and keep my win form application in focus and not minimize. At this point I am unsuccessful. I need more time understanding message codes. Is this the right approach? Can it be done? Should I look at other libraries such as Windows Automation? Key input is my other concern. What is the best way to send key strokes to other apps/games. Should I tap into DirectX, use some kind of send key, Automation Framework? Can any of these handle the multiple key strokes that some simulation games require? I appreciate any links and or insight you may have. If you have gone down this path for any reason I would love to hear your comments. Stan

    Read the article

  • Schema to support dynamic properties

    - by Johan Fredrik Varen
    Hi people. I'm working on an editor that enables its users to create "object" definitions in real-time. A definition can contain zero or more properties. A property has a name a type. Once a definition is created, a user can create an object of that definition and set the property values of that object. So by the click of a mouse-button, the user should ie. be able to create a new definition called "Bicycle", and add the property "Size" of type "Numeric". Then another property called "Name" of type "Text", and then another property called "Price" of type "Numeric". Once that is done, the user should be able to create a couple of "Bicycle" objects and fill in the "Name" and "Price" property values of each bike. Now, I've seen this feature in several software products, so it must be a well-known concept. My problem started when I sat down and tried to come up with a DB schema to support this data structure, because I want the property values to be stored using the appropriate column types. Ie. a numeric property value is stored as, say, an INT in the database, and a textual property value is stored as VARCHAR. First, I need a table that will hold all my object definitions: Table obj_defs id | name | ---------------- 1 | "Bicycle" | 2 | "Book" | Then I need a table for holding what sort of properties each object definition should have: Table prop_defs id | obj_def_id | name | type | ------------------------------------ 1 | 1 | "Size" | ? | 2 | 1 | "Name" | ? | 3 | 1 | "Price" | ? | 4 | 2 | "Title" | ? | 5 | 2 | "Author" | ? | 6 | 2 | "ISBN" | ? | I would also need a table that holds each object: Table objects id | created | updated | ------------------------------ 1 | 2011-05-14 | 2011-06-15 | 2 | 2011-05-14 | 2011-06-15 | 3 | 2011-05-14 | 2011-06-15 | Finally, I need a table that will hold the actual property values of each object, and one solution is for this table to have one column for each possible value type, such as this: Table prop_vals id | prop_def_id | object_id | numeric | textual | boolean | ------------------------------------------------------------ 1 | 1 | 1 | 27 | | | 2 | 2 | 1 | | "Trek" | | 3 | 3 | 1 | 1249 | | | 4 | 1 | 2 | 26 | | | 5 | 2 | 2 | | "GT" | | 6 | 3 | 2 | 159 | | | 7 | 4 | 3 | | "It" | | 8 | 5 | 3 | | "King" | | 9 | 6 | 4 | 9 | | | If I implemented this schema, what would the "type" column of the prop_defs table hold? Integers that each map to a column name, varchars that simply hold the column name? Any other possibilities? Would a stored procedure help me out here in some way? And what would the SQL for fetching the "name" property of object 2 look like?

    Read the article

  • Patterns: Local Singleton vs. Global Singleton?

    - by Mike Rosenblum
    There is a pattern that I use from time to time, but I'm not quite sure what it is called. I was hoping that the SO community could help me out. The pattern is pretty simple, and consists of two parts: A singleton factory, which creates objects based on the arguments passed to the factory method. Objects created by the factory. So far this is just a standard "singleton" pattern or "factory pattern". The issue that I'm asking about, however, is that the singleton factory in this case maintains a set of references to every object that it ever creates, held within a dictionary. These references can sometimes be strong references and sometimes weak references, but it can always reference any object that it has ever created. When receiving a request for a "new" object, the factory first searches the dictionary to see if an object with the required arguments already exits. If it does, it returns that object, if not, it returns a new object and also stores a reference to the new object within the dictionary. This pattern prevents having duplicative objects representing the same underlying "thing". This is useful where the created objects are relatively expensive. It can also be useful where these objects perform event handling or messaging - having one object per item being represented can prevent multiple messages/events for a single underlying source. There are probably other reasons to use this pattern, but this is where I've found this useful. My question is: what to call this? In a sense, each object is a singleton, at least with respect to the data it contains. Each is unique. But there are multiple instances of this class, however, so it's not at all a true singleton. In my own personal terminology, I tend to call the factory method a "global singleton". I then call the created objects "local singletons". I sometimes also say that the created objects have "reference equality", meaning that if two variables reference the same data (the same underlying item) then the reference they each hold must be to the same exact object, hence "reference equality". But these are my own invented terms, and I am not sure that they are good ones. Is there standard terminology for this concept? And if not, could some naming suggestions be made? Thanks in advance...

    Read the article

  • Rails learn's confusion

    - by Steve
    This is a beginner's rails learning confusion. When I learn rails, from time to time, I feel frustrated on rails' principle "Convention over Configuration". Rails uses heavily on conventions. A lot of them are just naming conventions. If I forget a convention, I will either use the wrong naming and get unexpected result or get things magically done but don't understand how. Sometimes, I think of configuration. At least configuration lists everything clearly and nothing is in fog. In rails, there seems a hidden, dark contract between you and the machine. If you follow the contract, you communicate well. But a beginner usually forgets items listed on the contract and this usually leads to confusion. That's why when I first pick up rails, I feel like it is somehow difficult to learn. Besides, there are many other things that could be new to a learner, such as using git, using plugins from community, using RESTful routing style, using RSpec. All these are new and come together in learning ruby and rails. This definitely adds up difficulties for a beginner. In contrast, if you learn php, it wouldn't be that bad. You can forget many things and focus on learning php itself. You don't need to learn database handling if you know SQL already(in rails, you need to learn a whole new concept migration), you don't have to learn a new decent unit test(in rails, usually they teach RSpec along the way because rails is agile and you should learn test-driven development in the early learning stage), you don't have to learn a new version control(in rails, you will be taught about git anyway), you don't have to use complicated plugins(in rails, they usually use third-party plugins in textbook examples! what the hell? why not teach how to do a simplified similar thing in rails?), you don't have to worry RESTful style. All in all, when I learn php, I learn it quick and soon I start to write things myself. Learning php is similar to learning C/java. It tastes like those traditional languages. When I learn rails, it is more difficult. And I need to learn ruby as well (I believe many of you learn ruby just because of rails). Does anyone have the similar feeling as I have? How do you overcome it and start to master rails? Hints will be welcomed. Thank you.

    Read the article

  • Stepping into Ruby Meta-Programming: Generating proxy methods for multiple internal methods

    - by mstksg
    Hi all; I've multiply heard Ruby touted for its super spectacular meta-programming capabilities, and I was wondering if anyone could help me get started with this problem. I have a class that works as an "archive" of sorts, with internal methods that process and output data based on an input. However, the items in the archive in the class itself are represented and processed with integers, for performance purposes. The actual items outside of the archive are known by their string representation, which is simply number_representation.to_s(36). Because of this, I have hooked up each internal method with a "proxy method" that converts the input into the integer form that the archive recognizes, runs the internal method, and converts the output (either a single other item, or a collection of them) back into strings. The naming convention is this: internal methods are represented by _method_name; their corresponding proxy method is represented by method_name, with no leading underscore. For example: class Archive ## PROXY METHODS ## ## input: string representation of id's ## output: string representation of id's def do_something_with id result = _do_something_with id.to_i(36) return nil if result == nil return result.to_s(36) end def do_something_with_pair id_1,id_2 result = _do_something_with_pair id_1.to_i(36), id_2.to_i(36) return nil if result == nil return result.to_s(36) end def do_something_with_these ids result = _do_something_with_these ids.map { |n| n.to_i(36) } return nil if result == nil return result.to_s(36) end def get_many_from id result = _get_many_from id return nil if result == nil # no sparse arrays returned return result.map { |n| n.to_s(36) } end ## INTERNAL METHODS ## ## input: integer representation of id's ## output: integer representation of id's def _do_something_with id # does something with one integer-represented id, # returning an id represented as an integer end def do_something_with_pair id_1,id_2 # does something with two integer-represented id's, # returning an id represented as an integer end def _do_something_with_these ids # does something with multiple integer ids, # returning an id represented as an integer end def _get_many_from id # does something with one integer-represented id, # returns a collection of id's represented as integers end end There are a couple of reasons why I can't just convert them if id.class == String at the beginning of the internal methods: These internal methods are somewhat computationally-intensive recursive functions, and I don't want the overhead of checking multiple times at every step There is no way, without adding an extra parameter, to tell whether or not to re-convert at the end I want to think of this as an exercise in understanding ruby meta-programming Does anyone have any ideas? edit The solution I'd like would preferably be able to take an array of method names @@PROXY_METHODS = [:do_something_with, :do_something_with_pair, :do_something_with_these, :get_many_from] iterate through them, and in each iteration, put out the proxy method. I'm not sure what would be done with the arguments, but is there a way to test for arguments of a method? If not, then simple duck typing/analogous concept would do as well.

    Read the article

  • Where are the function literals in c++?

    - by academicRobot
    First of all, maybe literals is not the right term for this concept, but its the closest I could think of (not literals in the sense of functions as first class citizens). The idea is that when you make a conventional function call, it compiles to something like this: callq <immediate address> But if you make a function call using a function pointer, it compiles to something like this: mov <memory location>,%rax callq *%rax Which is all well and good. However, what if I'm writing a template library that requires a callback of some sort with a specified argument list and the user of the library is expected to know what function they want to call at compile time? Then I would like to write my template to accept a function literal as a template parameter. So, similar to template <int int_literal> struct my_template {...};` I'd like to write template <func_literal_t func_literal> struct my_template {...}; and have calls to func_literal within my_template compile to callq <immediate address>. Is there a facility in C++ for this, or a work around to achieve the same effect? If not, why not (e.g. some cataclysmic side effects)? How about C++0x or another language? Solutions that are not portable are fine. Solutions that include the use of member function pointers would be ideal. I'm not particularly interested in being told "You are a <socially unacceptable term for a person of low IQ>, just use function pointers/functors." This is a curiosity based question, and it seems that it might be useful in some (albeit limited) applications. It seems like this should be possible since function names are just placeholders for a (relative) memory address, so why not allow more liberal use (e.g. aliasing) of this placeholder. p.s. I use function pointers and functions objects all the the time and they are great. But this post got me thinking about the don't pay for what you don't use principle in relation to function calls, and it seems like forcing the use of function pointers or similar facility when the function is known at compile time is a violation of this principle, though a small one.

    Read the article

  • What is an overloaded operator in c++?

    - by Jeff
    I realize this is a basic question but I have searched online, been to cplusplus.com, read through my book, and I can't seem to grasp the concept of overloaded operators. A specific example from cplusplus.com is: // vectors: overloading operators example #include <iostream> using namespace std; class CVector { public: int x,y; CVector () {}; CVector (int,int); CVector operator + (CVector); }; CVector::CVector (int a, int b) { x = a; y = b; } CVector CVector::operator+ (CVector param) { CVector temp; temp.x = x + param.x; temp.y = y + param.y; return (temp); } int main () { CVector a (3,1); CVector b (1,2); CVector c; c = a + b; cout << c.x << "," << c.y; return 0; } from http://www.cplusplus.com/doc/tutorial/classes2/ but reading through it I'm still not understanding them at all. I just need a basic example of the point of the overloaded operator (which I assume is the "CVector CVector::operator+ (CVector param)"). There's also this example from wikipedia: Time operator+(const Time& lhs, const Time& rhs) { Time temp = lhs; temp.seconds += rhs.seconds; if (temp.seconds >= 60) { temp.seconds -= 60; temp.minutes++; } temp.minutes += rhs.minutes; if (temp.minutes >= 60) { temp.minutes -= 60; temp.hours++; } temp.hours += rhs.hours; return temp; } from "http://en.wikipedia.org/wiki/Operator_overloading" The current assignment I'm working on I need to overload a ++ and a -- operator. Thanks in advance for the information and sorry about the somewhat vague question, unfortunately I'm just not sure on it at all.

    Read the article

  • What are some things you'd like fresh college grads to know?

    - by bradhe
    So I proposed this to the Reddit community and I'd like to get SO's perspective on this. This is pretty much the copypasta of what I put there. I was thinking about this last night and thought it would be neat to compile a list. I'm still a pretty fresh college grad -- been in industry for 2 years -- but I think that I might have a few interesting things to lend. You don't know as much as you think you do. Somehow, college students think they know a lot more than they do (or maybe that was just me). Likewise, they think they can do more than they actually can. You should fairly assess your skills. QA people are not out to get you. Humans introduce bugs to code. It's not (nescessarily) a personal reflection on you and your skills if your code has a bug and it's caught by the QA/testing team. Listen to your senior (developers). They are not actually fuddy duddies who don't know about the new L337 hax in Ruby (okay, sometimes they are, but still...). They have a wealth of knowledge that you can learn from and it's in your best interest to do so. You will most likely not be doing what you want to for a while. This is mostly true in the corporate world -- startups are a different matter. Also, this is due to more than just the economy, man! Junior devs need to earn their keep, so to speak. Everyone wants to be lead dev on the next project and there are a lot of people in line ahead of you! For every elite developer there are 100 average developers. Joel Spolsky, I'm looking at you. Somehow this concept of ninja coders has really ingrained itself in our culture. While I encourage you to be the best you can be don't be disappointed if people aren't writing blog posts about you in the near future. Anyone else have anything they would see added to this list?

    Read the article

  • Why boost::recursive_mutex is not working as expected?

    - by Kjir
    I have a custom class that uses boost mutexes and locks like this (only relevant parts): template<class T> class FFTBuf { public: FFTBuf(); [...] void lock(); void unlock(); private: T *_dst; int _siglen; int _processed_sums; int _expected_sums; int _assigned_sources; bool _written; boost::recursive_mutex _mut; boost::unique_lock<boost::recursive_mutex> _lock; }; template<class T> FFTBuf<T>::FFTBuf() : _dst(NULL), _siglen(0), _expected_sums(1), _processed_sums(0), _assigned_sources(0), _written(false), _lock(_mut, boost::defer_lock_t()) { } template<class T> void FFTBuf<T>::lock() { std::cerr << "Locking" << std::endl; _lock.lock(); std::cerr << "Locked" << std::endl; } template<class T> void FFTBuf<T>::unlock() { std::cerr << "Unlocking" << std::endl; _lock.unlock(); } If I try to lock more than once the object from the same thread, I get an exception (lock_error): #include "fft_buf.hpp" int main( void ) { FFTBuf<int> b( 256 ); b.lock(); b.lock(); b.unlock(); b.unlock(); return 0; } This is the output: sb@dex $ ./src/test Locking Locked Locking terminate called after throwing an instance of 'boost::lock_error' what(): boost::lock_error zsh: abort ./src/test Why is this happening? Am I understanding some concept incorrectly?

    Read the article

  • Where are the function literals c++?

    - by academicRobot
    First of all, maybe literals is not the right term for this concept, but its the closest I could think of (not literals in the sense of functions as first class citizens). The idea is that when you make a conventional function call, it compiles to something like this: callq <immediate address> But if you make a function call using a function pointer, it compiles to something like this: mov <memory location>,%rax callq *%rax Which is all well and good. However, what if I'm writing a template library that requires a callback of some sort with a specified argument list and the user of the library is expected to know what function they want to call at compile time? Then I would like to write my template to accept a function literal as a template parameter. So, similar to template <int int_literal> struct my_template {...};` I'd like to write template <func_literal_t func_literal> struct my_template {...}; and have calls to func_literal within my_template compile to callq <immediate address>. Is there a facility in C++ for this, or a work around to achieve the same effect? If not, why not (e.g. some cataclysmic side effects)? How about C++0x or another language? Solutions that are not portable are fine. Solutions that include the use of member function pointers would be ideal. I'm not particularly interested in being told "You are a <socially unacceptable term for a person of low IQ>, just use function pointers/functors." This is a curiosity based question, and it seems that it might be useful in some (albeit limited) applications. It seems like this should be possible since function names are just placeholders for a (relative) memory address, so why not allow more liberal use (e.g. aliasing) of this placeholder. p.s. I use function pointers and functions objects all the the time and they are great. But this post got me thinking about the don't pay for what you don't use principle in relation to function calls, and it seems like forcing the use of function pointers or similar facility when the function is known at compile time is a violation of this principle, though a small one.

    Read the article

  • Unable to get set intersection to work

    - by chavanak
    Sorry for the double post, I will update this question if I can't get things to work :) I am trying to compare two files. I will list the two file content: File 1 File 2 "d.complex.1" "d.complex.1" 1 4 5 5 48 47 65 21 d.complex.10 d.complex.10 46 6 21 46 109 121 192 192 TI am trying to compare the contents of the two file but not in a trivial way. I will explain what I want with an example. If you observe the file content I have typed above, the d.complex.1 of file_1 has "5" similar to d.complex.1 in file_2; the same d.complex.1 in file_1 has nothing similar to d.complex.10 in file_2. What I am trying to do is just to print out those d.complex. which has nothing in similar with the other d.complex. Consider the d.complex. as a heading if you want. But all I am trying is compare the numbers below each d.complex. and if nothing matches, I want that particular d.complex. from both files to be printed. If even one number is present in both d.complex. of both files, I want it to be rejected. My Code: The method I chose to achieve this was to use sets and then do a difference. Code I wrote was: first_complex=open( "file1.txt", "r" ) first_complex_lines=first_complex.readlines() first_complex_lines=map( string.strip, first_complex_lines ) first_complex.close() second_complex=open( "file2.txt", "r" ) second_complex_lines=second_complex.readlines() second_complex_lines=map( string.strip, second_complex_lines ) second_complex.close() list_1=[] list_2=[] res_1=[] for line in first_complex_lines: if line.startswith( "d.complex" ): res_1.append( [] ) res_1[-1].append( line ) res_2=[] for line in second_complex_lines: if line.startswith( "d.complex" ): res_2.append( [] ) res_2[-1].append( line ) h=len( res_1 ) k=len( res_2 ) for i in res_1: for j in res_2: print i[0] print j[0] target_set=set ( i ) target_set_1=set( j ) for s in target_set: if s not in target_set_1: if s[0] != "d": print s The above code is giving an output like this (just an example): d.complex.1.dssp d.complex.1.dssp 1 48 65 d.complex.1.dssp d.complex.10.dssp 46 21 109 What I would like to have is: d.complex.1 d.complex.1 (name from file2) d.complex.1 d.complex.10 (name from file2) I am sorry for confusing you guys, but this is all that is required. I am so new to python so my concept above might be flawed. Also I have never used sets before :(. Can someone give me a hand here?

    Read the article

  • Swing: How do I run a job from AWT thread, but after a window was layed out?

    - by java.is.for.desktop
    My complete GUI runs inside the AWT thread, because I start the main window using SwingUtilities.invokeAndWait(...). Now I have a JDialog which has just to display a JLabel, which indicates that a certain job is in progress, and close that dialog after the job was finished. The problem is: the label is not displayed. That job seems to be started before JDialog was fully layed-out. When I just let the dialog open without waiting for a job and closing, the label is displayed. The last thing the dialog does in its ctor is setVisible(true). Things such as revalidate(), repaint(), ... don't help either. Even when I start a thread for the monitored job, and wait for it using someThread.join() it doesn't help, because the current thread (which is the AWT thread) is blocked by join, I guess. Replacing JDialog with JFrame doesn't help either. So, is the concept wrong in general? Or can I manage it to do certain job after it is ensured that a JDialog (or JFrame) is fully layed-out? Simplified algorithm of what I'm trying to achieve: Create a subclass of JDialog Ensure that it and its contents are fully layed-out Start a process and wait for it to finish (threaded or not, doesn't matter) Close the dialog I managed to write a reproducible test case: EDIT Problem from an answer is now addressed: This use case does display the label, but it fails to close after the "simulated process", because of dialog's modality. import java.awt.*; import javax.swing.*; public class _DialogTest2 { public static void main(String[] args) throws Exception { SwingUtilities.invokeAndWait(new Runnable() { final JLabel jLabel = new JLabel("Please wait..."); @Override public void run() { JFrame myFrame = new JFrame("Main frame"); myFrame.setDefaultCloseOperation(JFrame.EXIT_ON_CLOSE); myFrame.setSize(750, 500); myFrame.setLocationRelativeTo(null); myFrame.setVisible(true); JDialog d = new JDialog(myFrame, "I'm waiting"); d.setModalityType(Dialog.ModalityType.APPLICATION_MODAL); d.add(jLabel); d.setSize(300, 200); d.setLocationRelativeTo(null); d.setVisible(true); SwingUtilities.invokeLater(new Runnable() { @Override public void run() { try { Thread.sleep(3000); // simulate process jLabel.setText("Done"); } catch (InterruptedException ex) { } } }); d.setVisible(false); d.dispose(); myFrame.setVisible(false); myFrame.dispose(); } }); } }

    Read the article

  • Controlling the USB from Windows

    - by b-gen-jack-o-neill
    Hi, I know this probably is not the easiest thing to do, but I am trying to connect Microcontroller and PC using USB. I dont want to use internal USART of Microcontroller or USB to RS232 converted, its project indended to help me understand various principles. So, getting the communication done from the Microcontroller side is piece of cake - I mean, when I know he protocol, its relativelly easy to implement it on Micro, becouse I am in direct control of evrything, even precise timing. But this is not the case of PC. I am not very familiar with concept of Windows handling the devices connected. In one of my previous question I ask about how Windows works with devices thru drivers. I understood that for internal use of Windows, drivers must have some default set of functions available to OS. I mean, when OS wants to access HDD, it calls HDD driver (which is probably internal in OS), with specific "questions" so that means that HDD driver has to be written to cooperate with Windows, to have write function in the proper place to be called by the OS. Something similiar is for GPU, Even DirectX, I mean DirectX must call specific functions from drivers, so drivers must be written to work with DX. I know, many functions from WinAPI works on their own, but even "simple" window must be in the end written into framebuffer, using MMIO to adress specified by drivers. Am I right? So, I expected that Windows have internal functions, parts of WinAPI designed to work with certain comonly used things. To call manufacturer-designed drivers. But this seems to not be entirely true becouse Windows has no way to communicate thru Paralel port. I mean, there is no function in the WinAPI to work with serial port, but there are funcions to work with HDD,GPU and so. But now there comes the part I am getting very lost at. So, I think Windows must have some built-in functions to communicate thru USB, becouse for example it handles USB flash memory. So, is there any WinAPI function designed to let user to operate USB thru that function, or when I want to use USB myself, do I have to call desired USB-driver function myself? Becouse all you need to send to USB controller is device adress and the infromation right? I mean, I don´t have to write any new drivers, am I right? Just to call WinAPI function if there is such, or directly call original USB driver. Does any of this make some sense?

    Read the article

  • Finding open contiguous blocks of time for every day of a month, fast

    - by Chris
    I am working on a booking availability system for a group of several venues, and am having a hard time generating the availability of time blocks for days in a given month. This is happening server-side in PHP, but the concept itself is language agnostic -- I could be doing this in JS or anything else. Given a venue_id, month, and year (6/2012 for example), I have a list of all events occurring in that range at that venue, represented as unix timestamps start and end. This data comes from the database. I need to establish what, if any, contiguous block of time of a minimum length (different per venue) exist on each day. For example, on 6/1 I have an event between 2:00pm and 7:00pm. The minimum time is 5 hours, so there's a block open there from 9am - 2pm and another between 7pm and 12pm. This would continue for the 2nd, 3rd, etc... every day of June. Some (most) of the days have nothing happening at all, some have 1 - 3 events. The solution I came up with works, but it also takes waaaay too long to generate the data. Basically, I loop every day of the month and create an array of timestamps for each 15 minutes of that day. Then, I loop the time spans of events from that day by 15 minutes, marking any "taken" timeslot as false. Remaining, I have an array that contains timestamp of free time vs. taken time: //one day's array after processing through loops (not real timestamps) array( 12345678=>12345678, // <--- avail 12345878=>12345878, 12346078=>12346078, 12346278=>false, // <--- not avail 12346478=>false, 12346678=>false, 12346878=>false, 12347078=>12347078, // <--- avail 12347278=>12347278 ) Now I would need to loop THIS array to find continuous time blocks, then check to see if they are long enough (each venue has a minimum), and if so then establish the descriptive text for their start and end (i.e. 9am - 2pm). WHEW! By the time all this looping is done, the user has grown bored and wandered off to Youtube to watch videos of puppies; it takes ages to so examine 30 or so days. Is there a faster way to solve this issue? To summarize the problem, given time ranges t1 and t2 on day d, how can I determine the remaining time left in d that is longer than the minimum time block m. This data is assembled on demand via AJAX as the user moves between calendar months. Results are cached per-page-load, so if the user goes to July a second time, the data that was generated the first time would be reused. Any other details that would help, let me know. Edit Per request, the database structure (or the part that is relevant here) *events* id (bigint) title (varchar) *event_times* id (bigint) event_id (bigint) venue_id (bigint) start (bigint) end (bigint) *venues* id (bigint) name (varchar) min_block (int) min_start (varchar) max_start (varchar)

    Read the article

  • Is it possible to create a C++ factory system that can create an instance of any "registered" object

    - by chrensli
    Hello, I've spent my entire day researching this topic, so it is with some scattered knowledge on the topic that i come to you with this inquiry. Please allow me to describe what I am attempting to accomplish, and maybe you can either suggest a solution to the immediate question, or another way to tackle the problem entirely. I am trying to mimic something related to how XAML files work in WPF, where you are essentially instantiating an object tree from an XML definition. If this is incorrect, please inform. This issue is otherwise unrelated to WPF, C#, or anything managed - I solely mention it because it is a similar concept.. So, I've created an XML parser class already, and generated a node tree based on ObjectNode objects. ObjectNode objects hold a string value called type, and they have an std::vector of child ObjectNode objects. The next step is to instantiate a tree of objects based on the data in the ObjectNode tree. This intermediate ObjectNode tree is needed because the same ObjectNode tree might be instantiated multiple times or delayed as needed. The tree of objects that is being created is such that the nodes in the tree are descendants of a common base class, which for now we can refer to as MyBase. Leaf nodes can be of any type, not necessarily derived from MyBase. To make this more challenging, I will not know what types of MyBase derived objects might be involved, so I need to allow for new types to be registered with the factory. I am aware of boost's factory. Their docs have an interesting little design paragraph on this page: o We may want a factory that takes some arguments that are forwarded to the constructor, o we will probably want to use smart pointers, o we may want several member functions to create different kinds of objects, o we might not necessarily need a polymorphic base class for the objects, o as we will see, we do not need a factory base class at all, o we might want to just call the constructor - without #new# to create an object on the stack, and o finally we might want to use customized memory management. I might not be understanding this all correctly, but that seems to state that what I'm trying to do can be accomplished with boost's factory. But all the examples I've located, seem to describe factories where all objects are derived from a base type. Any guidance on this would be greatly appreciated. Thanks for your time!

    Read the article

  • Are there any CMS editors out there which users can populate locked down HTML templates with content

    - by Deep
    Hi there, We work in email marketing, creating HTML/TEXT emails for clients. In essence we design HTML email templates for our clients. Clients then post us content (via a form) to populate these templates before we send them out. Right now we do this manually, basically cutting and pasting the content from their submitted form into the relevant parts of the template, which is time consuming and particularly mind-numbing. What we're looking for (and have so far been unable to find) is a simple system which will allow us to capture this client content in a sort of WYSIWYG HTML format. Basically they populate a locked down version of the template, entering text where necessary, before submitting to us. This is our most basic requirement, and a friend of mine kindly demo'd a proof of concept here: http://advantageone.co.uk/mbe/ Note: If you click on a text area in the body of the template, an editor pop ups. Now what we are looking for a CMS editor out there which can be easily adapted to do the above and the following for our end clients? User login View previously submitted campaigns that they have created and edit these Create new - selecting from template (assigned to their user/client id), perhaps being able to add new rows to the template. And have these HTML templates locked down so they can only edit what they're allowed too (like in the demo above), and perhaps make some areas required. Perhaps have a simple workflow or approval built in Allow us to lock submitted campaigns after a point so they can't be further edited, and as administrators view all campaigns from all users Be so incredibly simple, with any extraneous functionality switched off Essentially an extremley simple stripped down CMS, but we use the outputted HTML for sending out as an email, rather than publishing onto the web. Now to the actual dilemma: we're looking for something really simple, and the above sounds like a CMS. But we haven't been able to find anything that already does, or can be easily adapted to do this. Everything is either too complex, or simple and inflexible. We're sure there must be something off the shelf available, rather than us coding something ourselves. But we've kind of got stuck. Does anyone know of a system, or could recommend a system that can do the above out of the box, or with a few days tweaking? Forgive me if this is a little disjointed, if I'm being incredibly dopey and there is something out there please let me know! Kind regards, Dp.

    Read the article

  • How to check total cache size using a program

    - by user1888541
    so I'm having some trouble creating a program to measure cache size in C. I understand the basic concept of going about this but I'm still having trouble figuring out exactly what I am doing wrong. Basically, I create an array of varying length (going by power of 2s) and access each element in the array and put it in a dummy variable. I go through the array and do this around 1000 times to negate the "noise" that would otherwise occur if I only did it once to get an accurate measurement for time. Then, I look for the size that causes a big jump in access time. Unfortunately, this is where I am having my problem, I don't see this jump using my code and clearly I am doing something wrong. Another thing is that I used /proc/cpuinfo to check the cache and it said the size was 6114 but that was not a power of 2. I was told to go by powers of 2 to figure out the cache can anyone explain why this is? Here is the just of my code...I will post the rest if need be { struct timeval start; struct timeval end; // int n = 1; // change this to test different sizes int array_size = 1048576*n; // I'm trying to check the time "manually" first before creating a loop for the program to do it by itself this is why I have a separate "n" variable to increase the size char x = 0; int i =0, j=0; char *a; a =malloc(sizeof(char) * (array_size)); gettimeofday(&start,NULL); for(i=0; i<1000; i++) { for(j=0; j < array_size; j += 1) { x = a[j]; } } gettimeofday(&end,NULL); int timeTaken = (end.tv_sec * 1000000 + end.tv_usec) - (start.tv_sec *1000000 + start.tv_usec); printf("Time Taken: %d \n", timeTaken); printf("Average: %f \n", (double)timeTaken/((double)array_size); }

    Read the article

  • Is there really a need for encryption to have true wireless security? [closed]

    - by Cawas
    I welcome better key-wording here, both on tags and title. I'm trying to conceive a free, open and secure network environment that would work anywhere, from big enterprises to small home networks of just 1 machine. I think since wireless Access Points are the most, if not only, true weak point of a Local Area Network (let's not consider every other security aspect of having internet) there would be basically two points to consider here: Having an open AP for anyone to use the internet through Leaving the whole LAN also open for guests to be able to easily read (only) files on it, and even a place to drop files on Considering these two aspects, once everything is done properly... What's the most secure option between having that, or having just an encrypted password-protected wifi? Of course "both" would seem "more secure". But it shouldn't actually be anything substantial. That's the question, but I think it may need more elaborating on. If you don't think so, please feel free to skip the next (long) part. Elaborating more on the two aspects ... I've always had the feeling using any kind of the so called "wireless security" methods is actually a bad design. I'm talking mostly about encrypting and pass-phrasing (which are actually two different concepts), since I won't even consider hiding SSID and mac filtering. I understand it's a natural way of thinking. With cable networking nobody can access the network unless they have access to the physical cable, so you're "secure" in the physical way. In a way, encrypting is for wireless what building walls is for the cables. And giving pass-phrases would be adding a door with a key. But the cabling without encryption is also insecure. If someone plugin all the data is right there. So, while I can see the use for encrypting data, I don't think it's a security measure in wireless networks. It's wasting resources for too little gain. I believe we should encrypt only sensitive data regardless of wires. That's already done with HTTPS, so I don't really need to encrypt my torrents, for instance. They're torrents, they are meant to be freely shared! As for using passwords, they should be added to the users, always. Not to wifi. For securing files, truly, best solution is backup. Sure all that doesn't happen that often, but I won't consider the most situations where people just don't care. I think there are enough situations where we actually use passwords on our OS users, so let's go with that in mind. I keep promoting the Fonera concept as an instance. It opens up a free wifi port, if you choose so, and anyone can connect to the internet through that, without having any access to your LAN. It also uses a QoS which will never let your bandwidth drop from that public usage. That's security, and it's open. But it's lacking the second aspect. I'll probably be bashed for promoting the non-usage of WPA 2 with AES or whatever, but I wanted to know from more experienced (super) users out there: what do you think?

    Read the article

  • creating a heirarchy of terminals or workspaces

    - by intuited
    <rant This question occurred to me ('occurred' meaning 'whispered seductively in my ear for the 100th time') while using GNU-screen, so I'll make that my example. However this is a much more general question about user interfaces and what I perceive as a flawmissing feature in every implementation I've yet seen. I'm wondering if there is some way to create a heirarchy/tree of terminals in a screen session. EG I'd like to have something like 1 bash 1.1 bash 1.2 bash 2 bash 3 bash 3.1 bash 3.1.1 bash 3.1.2 bash It would be good if the terminals could be labelled instead of having to be navigated to via some arrangement that I suspect doesn't exist. So then you could jump to one using eg ^A:goto happydays or ^A:goto dykstra.angry. So to generalize that: Firefox, Chrome, Internet Explorer, gnome-terminal, roxterm, konsole, yakuake, OpenOffice, Microsoft Office, Mr. Snuffaluppagus's Funtime Carousel™, and Your Mom's Jam Browser™ all offer the ability to create a flat set of tabs containing documents of an identical nature: web pages, terminals, documents, fun rideable animals, and jams. GNU-screen implements the same functionality without using tabs. Linux and OS/X window managers provide the ability to organize windows into an array of workspaces, which amounts to again, the same deal. Over the past few years, this has become a more or less ubiquitous concept which has been righteously welcomed into the far reaches of the computer interface funfest. Heavy users of these systems quickly encounter a problem with it: the set of entities is flat. In the case of workspaces, an option may be available to create a 2d array. However none of these applications furnish their users with the ability to create heirarchies, similar to filesystem directory structures, containing instances of their particular contained type. I for one am consistently bothered by this, and am wondering if the community can offer some wisdom as to why this has not happened in any of the foremost collections of computational functionality our culture has yet produced. Or if perhaps it has and I'm just an ignorant savage. I'd like to be able to not only group things into a tree structure, but also to create references (aka symbolic links, aka pointers) from one part of the structure to another, as well as apply properties (eg default directory, colorscheme, ...) recursively downward from a given node. I see no reason why we shouldn't be able to save these structures as known sessions, and apply tags to particular instances. So then you can sort through them by tag, find them by name, or just use the arrow keys (with an appropriate modifier) to move left or right and in or out of a given level. Another key combo would serve to create a branch in the place of the current terminal/webpage/lifelike statue/spreadsheet/spreadsheet sheet/presentation/jam and move that entity into the new branch, then create a fresh one as a sibling to it: a second leaf node within the same branch node. They would get along well. I find it a bit astonishing that this hasn't happened yet, and the only reason I can venture as a guess is that the creators of these fine systems do not consider such functionality to be useful to a significant portion of their userbase. I posit that the probability that that such an assumption would be correct is pretty low. On the other hand, given the relative ease with which such structures can be implemented using modern libraries/languages, it doesn't seem likely that difficulty of implementation would be a major roadblock. If it could be done in 1972 or whenever within the constraints of a filesystem driver, it should be relatively painless to implement in 2010 in a fullblown application. Given that all of these systems are capable of maintaining a set of equivalent entities, it seems unlikely that a major infrastructure overhaul would be necessary in order to enable a navigable heirarchy of them. </rant Mostly I'm just looking to start up a discussion and/or brainstorming on this topic. Any ideas, examples, criticism, or analysis are quite welcome. * Mr. Snuffaluppagus's Funtime Carousel is a registered trademark of Children's Television Workshop Inc. * Your Mom's Jam Browser is a registered trademark of Your Mom Inc.

    Read the article

  • LDAP object class violation: attribute ou not allowed in suffix?

    - by Paramaeleon
    I am about to set up a LDAP directory. It is used as a tool to communicate user permissions from a web application to WebDav file system access, e.g. adding a user to the web platform shall allow login to the file system with the same credentials. There are no other usages intended. Following this German tutorial which encourages the use of the attributes c, o, ou etc. over dc, I configured the following suffix and root: suffix "ou=webtool,o=myOrg,c=de" rootdn "cn=ldapadmin,ou=webtool,o=myOrg,c=de" Server starts and I can connect to it by LDAP Admin, which reports “LDAP error: Object lacks”. Well, there aren’t any objects yet. I now want to create the root and admin elements from shell. I created an init.ldif file: dn: ou=webtool,o=myOrg,c=de objectclass: dcObject objectclass: organization dc: webtool o: webtool dn: cn=ldapadmin,ou=webtool,o=myOrg,c=de objectclass: organizationalRole cn: ldapadmin Trying to load the file runs into an error, telling me that ou is not allowed: server:~ # ldapadd -x -D "cn=ldapadmin,ou=webtool,o=myOrg,c=de" -W -f init.ldif Enter LDAP Password: adding new entry "ou=webtool,o=myOrg,c=de" ldap_add: Object class violation (65) additional info: attribute 'ou' not allowed I am not using ou anywhere except in the suffix, so the question: Isn’t it allowed here? What is allowed here? Here is my answer. I am not allowed to post it as answer for 8 hours, so don’t mind that it is part of the question by now. I will move it outside some day, if I don’t forget to do so. There are numberous dependencies for the creation of elements, and error messages are rather confusing if you don’t know of the concept. The objectclass isn’t necessarily dcObject for the databases’ root node, as it is likely to guess when you read several tutoriales. Instead, it must correspond to the object’s type: Here, for a name starting with ou=, it must be organizationalUnit. I found this piece of information in these tables [Link removed due to restriction: Oops! Your edit couldn't be submitted because: We're sorry, but as a spam prevention mechanism, new users can only post a maximum of two hyperlinks. Earn more than 10 reputation to post more hyperlinks. Link is below]. Further on, the object class dictates which properties must and can be added in the record. Here, organizationalUnit must have an ou: entry and must not have neither dc: nor o: entry. The healthy init.ldif file looks like that: dn: ou=webtool,o=myOrg,c=de objectclass: organizationalUnit ou: LDAP server for my webtool dn: cn=ldapadmin,ou=webtool,o=myOrg,c=de objectclass: organizationalRole cn: ldapadmin Note: The page also states: “While many objectClasses show no MUST attributes you must (ouch) follow any hierarchy […] to determine if this is the really case.” I thought that would mean my root record would have to provide the must fields for c= and o= (c: and o:, respectively) but this isn’t the case. Link in answer is (1): http :// www (dot) zytrax (dot) com/books/ldap/ape/ "Appendix E: LDAP - Object Classes and Attributes"

    Read the article

  • Wear and tear on server hard drive from filesystem polling by PHP script

    - by jackie
    So I'm working on a discussion platform, and various clients will visit http://host/thread.php, which will render the discussion thread to date in addition to a form to submit a new post. When a new post is submitted, I would like all of the other clients with browser windows open to have it appear in near-real-time. One of the constraints of my script is that it may not use a DBMS and it must stay in the filesystem. Additionally, I can't use any PECL/PEAR extensions like inotify or anything like that for IPC. The flow will look like this: Client A requests thread.php and the thread is so far empty, but nonetheless it opens a Server-Side Event at eventPusher.php. Client B does the same. Client A fills out a post in the form and and submits (POSTs) it to subHandler.php. ??? (subHandler stores the new submission into the main thread storefile which gets read from when a fresh, new client requests thread.php, in addition to somehow signalling to the continually-running eventPusher event-source that a new comment was posted and that it should echo the event-json to the client. How, exactly, it will send this signal I'm yet unsure of, but there are a few options that I've thought of -- this is the crux of the question, so see below for more clarification) eventPusher.php happily pushes the new event to the client and it shows up soon after it was originally submitted on all clients who have the page open's screens. Now for the #4 missing-link mystery-step, I see a few problems. I mean, either way, eventPusher is gonna be doing a while loop of some sort -- it's gonna be polling something, I think that much is clear. (If that's a bad assumption please do let me know.) Now, the simplest way would be subHandler gets invoked on the form submission, writes it to the main store in addition to newComments.xml, then exits without doing anything else. Then eventPusher checks in newComments.xml every X seconds (by the way, what would be a reasonable time interval here?) and if it finds something then it emits an event to the client. Now, my fear with this is that the server's hard drive will have to constantly start spinning up. Maybe this isn't the case, perhaps it would just get cached in RAM and the linux kernel would take care of this transparently such that filesystem access doesn't actually engage the device because the kernel knows that that particular file hasn't changed since last read. * idea #2: I have no idea how to go about this, but perhaps there is a variable scope that gets stored in general RAM on the system which can be read by any process. Like if we mega-exported a bash variable so that $new_post is normally false but it gets toggled to true by subHandler, and then back to flase once it's pushed to the client. I doubt there's such a variable scope in PHP directly, but I struggle with the concept of variable scope, I just can't seem to understand it no matter what I read on it. * idea #3: eventPusher queries ps in its whileloop for another instance of itself. If there's not another eventPusher active then it's highly unlikely that new comments will be getting submitted. It's okay if this only works =90% of the time, it doesn't need to be completely foolproof. * idea #4: eventPusher queries DMESG to see if that file's been written to recently. So to sum everything up, I need to have inter-php-script-communication in near-real-time that will work on a standard mod_php shared hosting setup without any elevated privileges, PHP addon modules, or other system adjustments that can't be done from the PHP script itself at runtime. With*out* spinning up the drive more than a few times. No SQL servers either. Apologies if my english isn't the best, I'm still trying to improve on it.

    Read the article

  • Network update solutions for a company of ~20 (5 local, 15 remote)?

    - by Margaret
    Hi all This is probably going to be a bit up in the air, because we're still in the "reaching towards solutions" phase, but I figured I'd see what you guys had to say. Plus I honestly know very little about systems and what is good and bad pratice. My organisation has always more or less worked on the concept of local machines; since it primarily employed contractors who were working from home, each of those people was largely responsible for their own machine and backup procedures and the like. We're now expanding, though we're still reasonably small (we're up to about 20 staff members). Most people still work remotely, but we have a central office where about five people are working. But we're getting large enough that we're starting to think it would be a good idea to have a central file server, and things like that - if someone gets hit by a bus, we want someone else to know where to look for the files to continue their work. A lot of the people who work for us remotely work on projects for other companies as well, so I don't want to force them to log in to our server whenever they're on a network. But I do want to make connection to be as painless as possible to do so, to improve utilisation. The other thing is that we're getting more people who would like to remote into the office server and do their work there. Our current remote connection application is an SSH install that allows people access to the network; the problem is, it's a black box to me, and I've never understood how to even connect to it (despite supposedly being de facto sysadmin). Thus far I've been able to bounce questions about how to get it working to the guy who does know it well, but he's leaving the company soon. So we probably need a solution for this that I actually understand. We were knocking around the idea of implementing a VPN with some form of remote desktop, and someone mentioned that this was largely a matter of purchasing a router capable of it; I'm not sure of the truth of that statement. This is what we have in the office: Two shiny new i7 servers, each running Windows Server 2008. Precise eventual layout is still being debated, a little, but the current suggestion is that one is primary database crunching, while the other is a warm backup of the databases, along with running Reporting Services. They currently have SQL Server 2008 installed on them, which is being connected to via the 'sa' account. We're hoping to make each person use their own account (preferably one tied to the 'central' password we set up, so we can use Windows Authentication). An older server, running XP Pro, that we are currently using as a test bed for a project that requires access to older versions of software. This machine is also being used to take backups, but I'm thinking of moving that functionality elsewhere. A spare desktop from a guy who left the company (XP Pro). We're thinking of bumping up the hard disk space and using it as the magical file server that's going to solve one particular everything. Assorted desktops, laptops, etc, at least one for each person in the office (mix of Win XP and Win 7; occasionally a person who normally works remotely might drop in to the office and bring a laptop bearing Vista, but it's pretty rare). All are set up as local user accounts at the moment; I don't know if it's the best arrangement. Purchasing more hardware is not a big problem, but we figure we might as well make use of what we've got first. Is Active Directory a big magic wand that's going to solve all the world's problems? Is there some other arrangement we should be looking to instead?

    Read the article

  • Multi-tenant ASP.NET MVC – Introduction

    - by zowens
    I’ve read a few different blogs that talk about multi-tenancy and how to resolve some of the issues surrounding multi-tenancy. What I’ve come to realize is that these implementations overcomplicate the issues and give only a muddy implementation! I’ve seen some really illogical code out there. I have recently been building a multi-tenancy framework for internal use at eagleenvision.net. Through this process, I’ve realized a few different techniques to make building multi-tenant applications actually quite easy. I will be posting a few different entries over the issue and my personal implementation. In this first post, I will discuss what multi-tenancy means and how my implementation will be structured.   So what’s the problem? Here’s the deal. Multi-tenancy is basically a technique of code-reuse of web application code. A multi-tenant application is an application that runs a single instance for multiple clients. Here the “client” is different URL bindings on IIS using ASP.NET MVC. The problem with different instances of the, essentially, same application is that you have to spin up different instances of ASP.NET. As the number of running instances of ASP.NET grows, so does the memory footprint of IIS. Stack Exchange shifted its architecture to multi-tenancy March. As the blog post explains, multi-tenancy saves cost in terms of memory utilization and physical disc storage. If you use the same code base for many applications, multi-tenancy just makes sense. You’ll reduce the amount of work it takes to synchronize the site implementations and you’ll thank your lucky stars later for choosing to use one application for multiple sites. Multi-tenancy allows the freedom of extensibility while relying on some pre-built code.   You’d think this would be simple. I have actually seen a real lack of reference material on the subject in terms of ASP.NET MVC. This is somewhat surprising given the number of users of ASP.NET MVC. However, I will certainly fill the void ;). Implementing a multi-tenant application takes a little thinking. It’s not straight-forward because the possibilities of implementation are endless. I have yet to see a great implementation of a multi-tenant MVC application. The only one that comes close to what I have in mind is Rob Ashton’s implementation (all the entries are listed on this page). There’s some really nasty code in there… something I’d really like to avoid. He has also written a library (MvcEx) that attempts to aid multi-tenant development. This code is even worse, in my honest opinion. Once I start seeing Reflection.Emit, I have to assume the worst :) In all seriousness, if his implementation makes sense to you, use it! It’s a fine implementation that should be given a look. At least look at the code. I will reference MvcEx going forward as a comparison to my implementation. I will explain why my approach differs from MvcEx and how it is better or worse (hopefully better).   Core Goals of my Multi-Tenant Implementation The first, and foremost, goal is to use Inversion of Control containers to my advantage. As you will see throughout this series, I pass around containers quite frequently and rely on their use heavily. I will be using StructureMap in my implementation. However, you could probably use your favorite IoC tool instead. <RANT> However, please don’t be stupid and abstract your IoC tool. Each IoC is powerful and by abstracting the capabilities, you’re doing yourself a real disservice. Who in the world swaps out IoC tools…? No one!</RANT> (It had to be said.) I will outline some of the goodness of StructureMap as we go along. This is really an invaluable tool in my tool belt and simple to use in my multi-tenant implementation. The second core goal is to represent a tenant as easily as possible. Just as a dependency container will be a first-class citizen, so will a tenant. This allows us to easily extend and use tenants. This will also allow different ways of “plugging in” tenants into your application. In my implementation, there will be a single dependency container for a single tenant. This will enable isolation of the dependencies of the tenant. The third goal is to use composition as a means to delegate “core” functions out to the tenant. More on this later.   Features In MvcExt, “Modules” are a code element of the infrastructure. I have simplified this concept and have named this “Features”. A feature is a simple element of an application. Controllers can be specified to have a feature and actions can have “sub features”. Each tenant can select features it needs and the other features will be hidden to the tenant’s users. My implementation doesn’t require something to be a feature. A controller can be common to all tenants. For example, (as you will see) I have a “Content” controller that will return the CSS, Javascript and Images for a tenant. This is common logic to all tenants and shouldn’t be hidden or considered a “feature”; Content is a core component.   Up next My next post will be all about the code. I will reveal some of the foundation to the way I do multi-tenancy. I will have posts dedicated to Foundation, Controllers, Views, Caching, Content and how to setup the tenants. Each post will be in-depth about the issues and implementation details, while adhering to my core goals outlined in this post. As always, comment with questions of DM me on twitter or send me an email.

    Read the article

  • Windows 7 Phone Database – Querying with Views and Filters

    - by SeanMcAlinden
    I’ve just added a feature to Rapid Repository to greatly improve how the Windows 7 Phone Database is queried for performance (This is in the trunk not in Release V1.0). The main concept behind it is to create a View Model class which would have only the minimum data you need for a page. This View Model is then stored and retrieved rather than the whole list of entities. Another feature of the views is that they can be pre-filtered to even further improve performance when querying. You can download the source from the Microsoft Codeplex site http://rapidrepository.codeplex.com/. Setting up a view Lets say you have an entity that stores lots of data about a game result for example: GameScore entity public class GameScore : IRapidEntity {     public Guid Id { get; set; }     public string GamerId {get;set;}     public string Name { get; set; }     public Double Score { get; set; }     public Byte[] ThumbnailAvatar { get; set; }     public DateTime DateAdded { get; set; } }   On your page you want to display a list of scores but you only want to display the score and the date added, you create a View Model for displaying just those properties. GameScoreView public class GameScoreView : IRapidView {     public Guid Id { get; set; }     public Double Score { get; set; }     public DateTime DateAdded { get; set; } }   Now you have the view model, the first thing to do is set up the view at application start up. This is done using the following syntax. View Setup public MainPage() {     RapidRepository<GameScore>.AddView<GameScoreView>(x => new GameScoreView { DateAdded = x.DateAdded, Score = x.Score }); } As you can see, using a little bit of lambda syntax, you put in the code for constructing a single view, this is used internally for mapping an entity to a view. *Note* you do not need to map the Id property, this is done automatically, a view model id will always be the same as it’s corresponding entity.   Adding Filters One of the cool features of the view is that you can add filters to limit the amount of data stored in the view, this will dramatically improve performance. You can add multiple filters using the fluent syntax if required. In this example, lets say that you will only ever show the scores for the last 10 days, you could add a filter like the following: Add single filter public MainPage() {     RapidRepository<GameScore>.AddView<GameScoreView>(x => new GameScoreView { DateAdded = x.DateAdded, Score = x.Score })         .AddFilter(x => x.DateAdded > DateTime.Now.AddDays(-10)); } If you wanted to further limit the data, you could also say only scores above 100: Add multiple filters public MainPage() {     RapidRepository<GameScore>.AddView<GameScoreView>(x => new GameScoreView { DateAdded = x.DateAdded, Score = x.Score })         .AddFilter(x => x.DateAdded > DateTime.Now.AddDays(-10))         .AddFilter(x => x.Score > 100); }   Querying the view model So the important part is how to query the data. This is done using the repository, there is a method called Query which accepts the type of view as a generic parameter (you can have multiple View Model types per entity type) You can either use the result of the query method directly or perform further querying on the result is required. Querying the View public void DisplayScores() {     RapidRepository<GameScore> repository = new RapidRepository<GameScore>();     List<GameScoreView> scores = repository.Query<GameScoreView>();       // display logic } Further Filtering public void TodaysScores() {     RapidRepository<GameScore> repository = new RapidRepository<GameScore>();     List<GameScoreView> todaysScores = repository.Query<GameScoreView>().Where(x => x.DateAdded > DateTime.Now.AddDays(-1)).ToList();       // display logic }   Retrieving the actual entity Retrieving the actual entity can be done easily by using the GetById method on the repository. Say for example you allow the user to click on a specific score to get further information, you can use the Id populated in the returned View Model GameScoreView and use it directly on the repository to retrieve the full entity. Get Full Entity public void GetFullEntity(Guid gameScoreViewId) {     RapidRepository<GameScore> repository = new RapidRepository<GameScore>();     GameScore fullEntity = repository.GetById(gameScoreViewId);       // display logic } Synchronising The View If you are upgrading from Rapid Repository V1.0 and are likely to have data in the repository already, you will need to perform a synchronisation to ensure the views and entities are fully in sync. You can either do this as a one off during the application upgrade or if you are a little more cautious, you could run this at each application start up. Synchronise the view public void MyUpgradeTasks() {     RapidRepository<GameScore>.SynchroniseView<GameScoreView>(); } It’s worth noting that in normal operation, the view keeps itself in sync with the entities so this is only really required if you are upgrading from V1.0 to V2.0 when it gets released shortly.   Summary I really hope you like this feature, it will be great for performance and I believe supports good practice by promoting the use of View Models for specific pages. I’m hoping to produce a beta for this over the next few days, I just want to add some more tests and hopefully iron out any bugs. I would really appreciate any thoughts on this feature and would really love to know of any bugs you find. You can download the source from the following : http://rapidrepository.codeplex.com/ Kind Regards, Sean McAlinden.

    Read the article

< Previous Page | 96 97 98 99 100 101 102 103 104 105 106 107  | Next Page >