Search Results

Search found 1898 results on 76 pages for 'structures'.

Page 59/76 | < Previous Page | 55 56 57 58 59 60 61 62 63 64 65 66  | Next Page >

  • Please help me out in fetching the desired result from below given DB table structure of MySQL..

    - by OM The Eternity
    Hi All below are the table structures according to which I have to develop the desired output(given at the end) tbl_docatr docatr_id doc_id docatrtype_id docatr_float docatr_int docatr_date docatr_varchar docatr_blob 1 12 1 NULL NULL NULL testing [BLOB - NULL] 2 12 2 NULL NULL NULL Tesitng [BLOB - NULL] tbl_docatrtype docatrtype_id docatrtypegroup_id docatrtypetype_id docatrtype_name 1 1 4 Name 2 1 4 Company Name tbl_docatrtypetype docatrtypetype_id docatrtypetype_name 1 Float 2 Int 3 Date 4 String line Above are three tables from which I have to display the desired output as Name : testing Company Name : Tesitng such that at first step I have doc_id then I get docatrtype_id and then docatrtypetype_id acording to these values i have to fetch the result. Also the query must see the doactrtypetype_id from table tbl_docatrtypetype and fetch the result from tbl_docatr from respective column docatr_float, docatr_int, docatr_date, docatr_varchar, docatr_blob Please help!!!

    Read the article

  • error defining foreign key PhpMyAdmin

    - by Ngounou lassale
    I am new to PhpMyAdmin. I will like to create a foreign key for my tables. In fact i have create tableI with this structures(A as int(11) autoincrement, B as varchar) TableII ( A_2 as int(11) auto increment, B_2 as varchar, A as int(11). I have declared A as an index in tableII, now when i go to relationship view to precise A as a foreign key i always have this error Erreur lors de la création de la clé étrangère sur ID_Ville (vérifiez le type des colonnes) Erreur ALTER TABLE tb_quartier ADD FOREIGN KEY ( ID_Ville ) REFERENCES ingenieris2.tb_ville ( ID_Ville ) ON DELETE RESTRICT ; Please Help thanks!

    Read the article

  • The Importance of a Security Assessment - by Michael Terra, Oracle

    - by Darin Pendergraft
    Today's Blog was written by Michael Terra, who was the Subject Matter Expert for the recently announced Oracle Online Security Assessment. You can take the Online Assessment here: Take the Online Assessment Over the past decade, IT Security has become a recognized and respected Business discipline.  Several factors have contributed to IT Security becoming a core business and organizational enabler including, but not limited to, increased external threats and increased regulatory pressure. Security is also viewed as a key enabler for strategic corporate activities such as mergers and acquisitions.Now, the challenge for senior security professionals is to develop an ongoing dialogue within their organizations about the importance of information security and how it can impact their organization's strategic objectives/mission. The importance of conducting regular “Security Assessments” across the IT and physical infrastructure has become increasingly important. Security standards and frameworks, such as the international standard ISO 27001, are increasingly being adopted by organizations and their business partners as proof of their security posture and “Security Assessments” are a great way to ensure a continued alignment to these frameworks.Oracle offers a number of different security assessment covering a broad range of technologies. Some of these are short engagements conducted for free with our strategic customers and partners. Others are longer term paid engagements delivered by Oracle Consulting Services or one of our partners. The goal of a security assessment, (also known as a security audit or security review), is to ensure that necessary security controls are integrated into the design and implementation of a project, application or technology.  A properly completed security assessment should provide documentation outlining any security gaps that exist in an infrastructure and the associated risks for those gaps. With that knowledge, an organization can choose to either mitigate, transfer, avoid or accept the risk. One example of an Oracle offering is a Security Readiness Assessment:The Oracle Security Readiness Assessment is a practical security architecture review focused on aligning an organization’s enterprise security architecture to their business principals and strategic objectives. The service will establish a multi-phase security architecture roadmap focused on supporting new and existing business initiatives.Offering OverviewThe Security Readiness Assessment will: Define an organization’s current security posture and provide a roadmap to a desired future state architecture by mapping  security solutions to business goals Incorporate commonly accepted security architecture concepts to streamline an organization’s security vision from strategy to implementation Define the people, process and technology implications of the desired future state architecture The objective is to deliver cohesive, best practice security architectures spanning multiple domains that are unique and specific to the context of your organization. Offering DetailsThe Oracle Security Readiness Assessment is a multi-stage process with a dedicated Oracle Security team supporting your organization.  During the course of this free engagement, the team will focus on the following: Review your current business operating model and supporting IT security structures and processes Partner with your organization to establish a future state security architecture leveraging Oracle’s reference architectures, capability maps, and best practices Provide guidance and recommendations on governance practices for the rollout and adoption of your future state security architecture Create an initial business case for the adoption of the future state security architecture If you are interested in finding out more, ask your Sales Consultant or Account Manager for details.

    Read the article

  • SQL SERVER – 4 Tips for ETL Software IDE Developers

    - by pinaldave
    In a previous blog, I introduced the notion of Semantic Types. To an end-user, a seamlessly integrated semantic typing engine significantly increases the ease of use of an ETL IDE (integrated development environment, or developer studio). This led me to think about other ease-of-use issues I have encountered while building ETL applications. When I get stumped while programming, I find myself asking the variations on these questions: “How do I…?” “Now what?” “Why isn’t this working?” “Why do I have to redo the work I just did?” It seems to me that a good ETL IDE will anticipate these questions and seek to answer them before they are even asked. So here are my tips to help software vendors build developer IDEs that actually make development easier. How do I…? While developing an ETL application, have you ever asked yourself: “How do I set up the connection to my SQL Server database?”,“How do I import my table definitions from Access?”, etc. An easy answer might be “read the manual” but sometimes product manuals are not robust or easily accessible. So, integrating robust how-to instructions directly into your ETLstudio would help users get the information they need at the time they need it. Now what? IDEs in general know where you last clicked or performed an action using an input device such as a keyboard; so they should be able to reasonably predict the design context you are in and suggest the next steps accordingly. Context-sensitive suggestions based on the state of the user’s work will help users move forward in ETL application development. Why isn’t this working? Or why do I have to wait till I compile to be told about a critical design issue? If an ETL IDE is smart enough to signal to users what in their design structures is left to be completed or has been completed incorrectly, then the developer can spend much less time in the designàcompileàerror-correct loop. Just-in-time validation helps users detect and correct programming errors earlier in the ETL development life cycle. Why do I have to redo the work I just did? In ETL development, schemas, transformation rules, connectivity objects, etc., can be reused in various situations. Using mouse-clicks to build and manage libraries of reusable design objects implies that the application development effort should decrease over time and as the library acquires more objects. I met a great company at SQL Pass that is trying to address many of these usability issues. Check them out at www.expressor-software.com. What other ease-of-use suggestions do you have for ETL software vendors? Please post your valuable comments. ?Reference: Pinal Dave (http://blog.SQLAuthority.com) Filed under: Best Practices, Pinal Dave, PostADay, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, T SQL, Technology Tagged: ETL

    Read the article

  • Resources such as libraries, engines and frameworks to make Javacript-based MMORTS? [closed]

    - by hhh
    I am looking for resources outlined to make a MMORTS with Javascript as the client-side, probably just a simple canvas for the frontend. The guy in the video here mentions that JavaScript is one of the most misunderstood language -- and I do believe that. I think one can make quite cool games with it in the future. So I am now proactively looking for resources and perhaps some ideas. My first idea contained Node.js, C and NetBSD/bozohttpd (or the-4-7-chars' *ix-thing with green-logo -thing, move the q here) but I acknowledge my beginner -style approach -- this issue is broad and not only for one person to make it all-the-time-improved project! So I think perfect for community to tinker. Some games and examples possibly easy to make into MMORTS BrowserQuest here under MPL 2.0 and its content licensed under CC-BY-SA 3.0 (source here) [proprietary] LoU here and built with JS/Qooxdoo/c#/Windows-Server/ISS/etc, source. MY ANSWER BEGINS HERE TO BE MOVED BELOW, REQUIRING RE-OPENING. PLEASE, VOTE TO OPEN IT -- HELP US TO TINKER! My answer Generic Is there an MMO-related research body? Although about Android, certain things also appropriate with JS -game: Are there any 2D gaming libraries/frameworks/engines for Android? Why is it so hard to develop a MMO? Browser based MMO Architecture MMO architecture - Highly Scalable with Reporting capabilities What are the Elements of an MMO Game? Is this the right architecture for our MMORPG mobile game? Looking for architectures to develop massive multiplayer game server Information on seamless MMO server architecture Game-mechanics (search) Question sounding like about LoU: What are the different ways to balance an online multiplayer game where user spend different amounts of time online? Building an instance system What are the different ways to balance an online multiplayer game where user spend different amounts of time online? Hosting is it possible to make a MMO starting with scalable hosting? Should I keep login server apart from game server? MMO techniques, algorithms and resources for keeping bandwidth low? MMO Proxy Server Javascript and Client-based things What do I need to do a MMORTS in JavaScript with small amount of Developers? How to update the monsters in my MMO server using Node.js and Socket.IO Are there any good html 5 mmo design tutorials? Networking Loadbalancing Questions Something about TCP, routers, NAT, etc: How do I start writing an MMO game server? Who does the AI calculations in an MMO? They need someone more knowledgable to work with, a lot of cases where the same words mean different things. Data Structures What data structure should I use for a Diablo/WoW-style talent tree? Game Engine Need an engine for MMO mockup Helper sites http://www.gamedev.net/page/index.html

    Read the article

  • Introducing the Oracle MDM Blog - Why All MDM Solutions Aren't Equal

    - by ken.pulverman
    Welcome to the Oracle MDM Blog.  Dave Butler, Tony Ouk, and myself - Ken Pulverman, will be bringing you news and information from the world of MDM at Oracle.  Dave is our resident expert with more than 30 years of experience in data and information management. Tony has deep expertise in our Exadata product line which provides a strong hardware synergy with MDM.  I come from Siebel Systems where I helped found the team that built our integration product line and then our Universal Customer Master with is part of our MDM offering at Oracle. I thought I'd hit the ground running with a topic we are going to want to continue to bend your ear about.  We had a recent meeting with Ford Goodman, our head of MDM commercial sales in the US and he was very fired up about and important topic.  He's irked that all MDM solutions get painted with the same brush even though they aren't the same at all. There are companies out there trying to represent frameworks and toolkits as out of the box solutions.  They give you the pleasure (read pain) of doing things like developing your own multi-application data model, building your own web services, or creating your own APIs.  Huh?  What gets sold as flexibility in reality is a barrier to ever going live.  At Siebel Systems we obsessed over the notion of a customer.  Our data model took over 10 years to perfect as defining a customer is a very complex task indeed.  There are divisions, subsidiaries, branches, acquisitions, sites etc., etc., etc..  You'll want to do your homework, but trust me - you aren't going to want to take the time or resource to build these canonical data structures yourself.  And what about APIs?  Again, it sounds flexible.  In reality it's a lot of work. Our DNA at Oracle is to reduce the cost of information technology so we pre-integrate our technology with all of our major applications and pre-build integrations and connectors for all the major systems you work with.  This is tedious work that requires detailed knowledge of the interfaces of all the applications involved.  It is also version specific as the interface features and technology are always changing.  We have a substantial organization to manage this complexity so you don't have to.  Suffice to say, we'd like to help our customers peel back the rhetoric of companies that fly the MDM flag without a real offering that you can quickly benefit from. Please watch this space for more information on this storyline as well as news and information around Oracle MDM.

    Read the article

  • Technology Selection for a dynamic product

    - by Kuntal Shah
    We are building a product for Procurement Domain in JAVA. Following are the main technical requirements. Platform Independent Database Independent Browser Independent In functional requirements the product is very dynamic in nature. The main reason being the procurement process around the world is different from client to client. Briefly we need to have a dynamic workflow engine and a dynamic template engine. The workflow engine by which we can define any kind of workflows and the template engine allows us to define any kind of data structures and based on definition it can get the user input through workflow. We have been developing this product for almost 2 years. It has been a long time till we can get down with the dynamics of requirements. Till now we have developed a basic workflow and template engine and which is in use at one of the client. We have been using following technologies. GWT-Ext (Front End Framework) Hibernate (Database Layer) In between we have faced some issues with GWT-Ext (mainly browser compatibility) and database optimization due to sub classing in hibernate. For resolving GWT-Ext issue, which a dying community so we decided to move to SmartGWT. In SmartGWT we faced issues related to loading and now we are able to finalize that GWT 2.3 will be the way to go as the library is rich and performance is upto the mark. We are able to almost finalize GWT-Spring based front and middle layer. In hibernate, we found main issues with sub-classing due to that it was throwing astronomical queries and sometimes it would stop firing any queries for 5-10 seconds or may be around 30 seconds and then resume again. Few days back I came to one article related to ORM. I am a traditional .Net SQL developer and I have always worked with relational database. Reading through this article, I also found it relating to the issues I face. I am still not completely convinced of using hibernate and this article just supported my opinion. Following are the questions for which I am looking for an answer. Should we be going with Hibernate in case of dynamic database requirements and the load of the data will be heavy in future? How can we partition the data, how we can efficiently join the data, how we can optimize the queries? If the answer is no then how do we achieve database independence? Is our choice related to GWT and Spring proper or do we need to change that too? Should we use any other key value pair database if the data is dynamic in nature and it is very difficult to make it relational?

    Read the article

  • Design considerations on JSON schema for scalars with a consistent attachment property

    - by casperOne
    I'm trying to create a JSON schema for the results of doing statistical analysis based on disparate pieces of data. The current schema I have looks something like this: { // Basic key information. video : "http://www.youtube.com/watch?v=7uwfjpfK0jo", start : "00:00:00", end : null, // For results of analysis, to be populated: // *** This is where it gets interesting *** analysis : { game : { value: "Super Street Fighter 4: Arcade Edition Ver. 2012", confidence: 0.9725 } teams : [ { player : { value : "Desk", confidence: 0.95, } characters : [ { value : "Hakan", confidence: 0.80 } ] } ] } } The issue is the tuples that are used to store a value and the confidence related to that value (i.e. { value : "some value", confidence : 0.85 }), populated after the results of the analysis. This leads to a creep of this tuple for every value. Take a fully-fleshed out value from the characters array: { name : { value : "Hakan", confidence: 0.80 } ultra : { value: 1, confidence: 0.90 } } As the structures that represent the values become more and more detailed (and more analysis is done on them to try and determine the confidence behind that analysis), the nesting of the tuples adds great deal of noise to the overall structure, considering that the final result (when verified) will be: { name : "Hakan", ultra : 1 } (And recall that this is just a nested value) In .NET (in which I'll be using to work with this data), I'd have a little helper like this: public class UnknownValue<T> { T Value { get; set; } double? Confidence { get; set; } } Which I'd then use like so: public class Character { public UnknownValue<Character> Name { get; set; } } While the same as the JSON representation in code, it doesn't have the same creep because I don't have to redefine the tuple every time and property accessors hide the appearance of creep. Of course, this is an apples-to-oranges comparison, the above is code while the JSON is data. Is there a more formalized/cleaner/best practice way of containing the creep of these tuples in JSON, or is the approach above an accepted approach for the type of data I'm trying to store (and I'm just perceiving it the wrong way)? Note, this is being represented in JSON because this will ultimately go in a document database (something like RavenDB or elasticsearch). I'm not concerned about being able to serialize into the object above, because I can always use data transfer objects to facilitate getting data into/out of my underlying data store.

    Read the article

  • Archbeat Link-O-Rama Top 10 Facebook Faves - June 23-29, 2013

    - by Bob Rhubart
    2,947 people now follow OTN ArchBeat on Facebook. Here are the Top 10 items shared on that page for June 23-29, 2013. Podcast Show Notes: DevOps, Cloud, and Role Creep After some confusion (my bad) all three CORRECT parts of this podcast are now available. The panelists for this discussion are all Oracle ACE Directors: Ron Batra, Basheer Khan, and Cary Millsap. SOA Suite 11g Developers Cookbook Published | Antony Reynolds "The book focuses on areas that we felt we had neglected in the Developers Guide, says co-author Antony Reynolds. "There is more about Java integration and OSB, both of which we see a lot of questions about when working with customers." Using Oracle TimesTen With Oracle BI Applications (Part 2) | Peter Scott Peter Scott follows up an earlier post with a look at some of the OBIA structures and a discussion of some of the features of TimesTen. Linux-Containers — Part 1: Overview | Lenz Grimmer OTN Garage blogger Lenz Grimmer kicks off a series and expands your mind with deep detail on Linux Containers Slides from my ODTUG Kscope13 Presentation | Zeeshan Baig Oracle ACE Zeeshan Baig shares the slides from his KScope13 presentation, "Build Your Business Services Using ADF Task Flows." Fun with Enterprise Manager | Rene van Wijk Oracle ACE Rene van Wijk shares some background and some tuning and other tech tips for working with Oracle Enterprise Manager. Using VirtualBox to test drive Windows Blue | The Fat Bloke The Fat Bloke shares a tech tip for those interested in giving Windows Blue a try on Virtual Box. Podcast Show Notes: The Fusion Middleware A-Team and the Chronicles of Architecture In this three-part series Oracle Fusion Middleware A-Team members Jennifer Briscoe, Clifford Musante, Mikael Ottosson, and Pardha Reddy talk about the origins and mission of the FMW A-Team and about the great technical content you'll find on the recently launched Oracle A-Team blog. Part one is now available. 5 Best Practices - Laying the Foundation for WebCenter Projects | John Brunswick Oracle WebCenter expert John Brunswick shares best practices that "enable the creation of portal solutions with minimal resource overhead, while offering the greatest flexibility for progressive elaboration." Oracle Magazine - July/Aug 2013 The digital edition of the July/August edition of Oracle Magazine is now available. This issue includes my architect community column, "The CX Factor." which features insight from community members on "why and how CX has become a significant factor in enterprise IT." h

    Read the article

  • Big Data – Interacting with Hadoop – What is PIG? – What is PIG Latin? – Day 16 of 21

    - by Pinal Dave
    In yesterday’s blog post we learned the importance of the HIVE in Big Data Story. In this article we will understand what is PIG and PIG Latin in Big Data Story. Yahoo started working on Pig for their application deployment on Hadoop. The goal of Yahoo to manage their unstructured data. What is Pig and What is Pig Latin? Pig is a high level platform for creating MapReduce programs used with Hadoop and the language we use for this platform is called PIG Latin. The pig was designed to make Hadoop more user-friendly and approachable by power-users and nondevelopers. PIG is an interactive execution environment supporting Pig Latin language. The language Pig Latin has supported loading and processing of input data with series of transforming to produce desired results. PIG has two different execution environments 1) Local Mode – In this case all the scripts run on a single machine. 2) Hadoop – In this case all the scripts run on Hadoop Cluster. Pig Latin vs SQL Pig essentially creates set of map and reduce jobs under the hoods. Due to same users does not have to now write, compile and build solution for Big Data. The pig is very similar to SQL in many ways. The Ping Latin language provide an abstraction layer over the data. It focuses on the data and not the structure under the hood. Pig Latin is a very powerful language and it can do various operations like loading and storing data, streaming data, filtering data as well various data operations related to strings. The major difference between SQL and Pig Latin is that PIG is procedural and SQL is declarative. In simpler words, Pig Latin is very similar to SQ Lexecution plan and that makes it much easier for programmers to build various processes. Whereas SQL handles trees naturally, Pig Latin follows directed acyclic graph (DAG). DAGs is used to model several different kinds of structures in mathematics and computer science. DAG Tomorrow In tomorrow’s blog post we will discuss about very important components of the Big Data Ecosystem – Zookeeper. Reference: Pinal Dave (http://blog.sqlauthority.com) Filed under: Big Data, PostADay, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, T SQL

    Read the article

  • BI&EPM in Focus June 2013

    - by Mike.Hallett(at)Oracle-BI&EPM
    Analyst Report from Ovum: BI bites into a bigger slice of Oracle’s Red Stack Customers INC Research Ensures 24/7 Enterprise Application Availability and Supports Rapid Expansion in Asia with Managed Cloud Services – Hyperion Planning, PeopleSoft, E-Business Suite, SOA Suite PL Developments Improves Quality and Demand Planning Accuracy, Streamlines Compliance as It Moves into Manufacturing – Hyperion Planning, OBIEE, E-Business Suite Release 12.1, Agile, Demantra Kiabi Provides Store Managers with Monthly Earnings Statements in Four Business Days to Support Continued Retail Growth – Hyperion Planning, Hyperion Financial Reporting, Hyperion Smart View for Office Speedy Cash Improves Global Financial Budgeting and Forecasting to Support Continued Company Growth - Hyperion Planning, Essbase, Hyperion Smart View for Office, Hyperion Financial Management Grupo Sports World Automates and Reduces Budget Consolidation Time by 33% for 30 Fitness Centers – Hyperion Planning Jupiter Shop Channel Automates Budgeting Processes, Enhances Visibility of Project Investments to Support Strategic Decision-Making – Hyperion Planning GENBAND Saves US$1.25 Million Annually with Automated Global Trade Management, Gains Compliance Assurance – Hyperion Financial Management, E-Business Suite Aldar Properties Consolidates and Simplifies Group Planning and Reporting for Business and Finance Structures with Integrated ERP and Business Intelligence – Hyperion Planning, Essbase, Data Integrator, OBIEE, E-Business Suite, SUN Link to Complete Archive Enterprise Performance Management Hyperion EPM 11.1.2.3 Webcast Tutorials EPM Blog: Three Technologies CFOs Need to Know About The CFO as Catalyst for Change - Part 1 The CFO as Catalyst for Change - Part 2 Actions Speak Louder in Scorecards Unlocking Business Potential with Enterprise Performance Management Business Intelligence Oracle Database 12c is launched Analysis: How to Take Big Data Advantage of Oracle Database 12c by Data-informed.com Normal 0 false false false EN-GB X-NONE X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-qformat:yes; mso-style-parent:""; mso-padding-alt:0cm 5.4pt 0cm 5.4pt; mso-para-margin-top:0cm; mso-para-margin-right:0cm; mso-para-margin-bottom:10.0pt; mso-para-margin-left:0cm; line-height:115%; mso-pagination:widow-orphan; font-size:11.0pt; font-family:"Calibri","sans-serif"; mso-ascii-font-family:Calibri; mso-ascii-theme-font:minor-latin; mso-hansi-font-family:Calibri; mso-hansi-theme-font:minor-latin; mso-bidi-font-family:"Times New Roman"; mso-bidi-theme-font:minor-bidi; mso-fareast-language:EN-US;}

    Read the article

  • PTLQueue : a scalable bounded-capacity MPMC queue

    - by Dave
    Title: Fast concurrent MPMC queue -- I've used the following concurrent queue algorithm enough that it warrants a blog entry. I'll sketch out the design of a fast and scalable multiple-producer multiple-consumer (MPSC) concurrent queue called PTLQueue. The queue has bounded capacity and is implemented via a circular array. Bounded capacity can be a useful property if there's a mismatch between producer rates and consumer rates where an unbounded queue might otherwise result in excessive memory consumption by virtue of the container nodes that -- in some queue implementations -- are used to hold values. A bounded-capacity queue can provide flow control between components. Beware, however, that bounded collections can also result in resource deadlock if abused. The put() and take() operators are partial and wait for the collection to become non-full or non-empty, respectively. Put() and take() do not allocate memory, and are not vulnerable to the ABA pathologies. The PTLQueue algorithm can be implemented equally well in C/C++ and Java. Partial operators are often more convenient than total methods. In many use cases if the preconditions aren't met, there's nothing else useful the thread can do, so it may as well wait via a partial method. An exception is in the case of work-stealing queues where a thief might scan a set of queues from which it could potentially steal. Total methods return ASAP with a success-failure indication. (It's tempting to describe a queue or API as blocking or non-blocking instead of partial or total, but non-blocking is already an overloaded concurrency term. Perhaps waiting/non-waiting or patient/impatient might be better terms). It's also trivial to construct partial operators by busy-waiting via total operators, but such constructs may be less efficient than an operator explicitly and intentionally designed to wait. A PTLQueue instance contains an array of slots, where each slot has volatile Turn and MailBox fields. The array has power-of-two length allowing mod/div operations to be replaced by masking. We assume sensible padding and alignment to reduce the impact of false sharing. (On x86 I recommend 128-byte alignment and padding because of the adjacent-sector prefetch facility). Each queue also has PutCursor and TakeCursor cursor variables, each of which should be sequestered as the sole occupant of a cache line or sector. You can opt to use 64-bit integers if concerned about wrap-around aliasing in the cursor variables. Put(null) is considered illegal, but the caller or implementation can easily check for and convert null to a distinguished non-null proxy value if null happens to be a value you'd like to pass. Take() will accordingly convert the proxy value back to null. An advantage of PTLQueue is that you can use atomic fetch-and-increment for the partial methods. We initialize each slot at index I with (Turn=I, MailBox=null). Both cursors are initially 0. All shared variables are considered "volatile" and atomics such as CAS and AtomicFetchAndIncrement are presumed to have bidirectional fence semantics. Finally T is the templated type. I've sketched out a total tryTake() method below that allows the caller to poll the queue. tryPut() has an analogous construction. Zebra stripping : alternating row colors for nice-looking code listings. See also google code "prettify" : https://code.google.com/p/google-code-prettify/ Prettify is a javascript module that yields the HTML/CSS/JS equivalent of pretty-print. -- pre:nth-child(odd) { background-color:#ff0000; } pre:nth-child(even) { background-color:#0000ff; } border-left: 11px solid #ccc; margin: 1.7em 0 1.7em 0.3em; background-color:#BFB; font-size:12px; line-height:65%; " // PTLQueue : Put(v) : // producer : partial method - waits as necessary assert v != null assert Mask = 1 && (Mask & (Mask+1)) == 0 // Document invariants // doorway step // Obtain a sequence number -- ticket // As a practical concern the ticket value is temporally unique // The ticket also identifies and selects a slot auto tkt = AtomicFetchIncrement (&PutCursor, 1) slot * s = &Slots[tkt & Mask] // waiting phase : // wait for slot's generation to match the tkt value assigned to this put() invocation. // The "generation" is implicitly encoded as the upper bits in the cursor // above those used to specify the index : tkt div (Mask+1) // The generation serves as an epoch number to identify a cohort of threads // accessing disjoint slots while s-Turn != tkt : Pause assert s-MailBox == null s-MailBox = v // deposit and pass message Take() : // consumer : partial method - waits as necessary auto tkt = AtomicFetchIncrement (&TakeCursor,1) slot * s = &Slots[tkt & Mask] // 2-stage waiting : // First wait for turn for our generation // Acquire exclusive "take" access to slot's MailBox field // Then wait for the slot to become occupied while s-Turn != tkt : Pause // Concurrency in this section of code is now reduced to just 1 producer thread // vs 1 consumer thread. // For a given queue and slot, there will be most one Take() operation running // in this section. // Consumer waits for producer to arrive and make slot non-empty // Extract message; clear mailbox; advance Turn indicator // We have an obvious happens-before relation : // Put(m) happens-before corresponding Take() that returns that same "m" for T v = s-MailBox if v != null : s-MailBox = null ST-ST barrier s-Turn = tkt + Mask + 1 // unlock slot to admit next producer and consumer return v Pause tryTake() : // total method - returns ASAP with failure indication for auto tkt = TakeCursor slot * s = &Slots[tkt & Mask] if s-Turn != tkt : return null T v = s-MailBox // presumptive return value if v == null : return null // ratify tkt and v values and commit by advancing cursor if CAS (&TakeCursor, tkt, tkt+1) != tkt : continue s-MailBox = null ST-ST barrier s-Turn = tkt + Mask + 1 return v The basic idea derives from the Partitioned Ticket Lock "PTL" (US20120240126-A1) and the MultiLane Concurrent Bag (US8689237). The latter is essentially a circular ring-buffer where the elements themselves are queues or concurrent collections. You can think of the PTLQueue as a partitioned ticket lock "PTL" augmented to pass values from lock to unlock via the slots. Alternatively, you could conceptualize of PTLQueue as a degenerate MultiLane bag where each slot or "lane" consists of a simple single-word MailBox instead of a general queue. Each lane in PTLQueue also has a private Turn field which acts like the Turn (Grant) variables found in PTL. Turn enforces strict FIFO ordering and restricts concurrency on the slot mailbox field to at most one simultaneous put() and take() operation. PTL uses a single "ticket" variable and per-slot Turn (grant) fields while MultiLane has distinct PutCursor and TakeCursor cursors and abstract per-slot sub-queues. Both PTL and MultiLane advance their cursor and ticket variables with atomic fetch-and-increment. PTLQueue borrows from both PTL and MultiLane and has distinct put and take cursors and per-slot Turn fields. Instead of a per-slot queues, PTLQueue uses a simple single-word MailBox field. PutCursor and TakeCursor act like a pair of ticket locks, conferring "put" and "take" access to a given slot. PutCursor, for instance, assigns an incoming put() request to a slot and serves as a PTL "Ticket" to acquire "put" permission to that slot's MailBox field. To better explain the operation of PTLQueue we deconstruct the operation of put() and take() as follows. Put() first increments PutCursor obtaining a new unique ticket. That ticket value also identifies a slot. Put() next waits for that slot's Turn field to match that ticket value. This is tantamount to using a PTL to acquire "put" permission on the slot's MailBox field. Finally, having obtained exclusive "put" permission on the slot, put() stores the message value into the slot's MailBox. Take() similarly advances TakeCursor, identifying a slot, and then acquires and secures "take" permission on a slot by waiting for Turn. Take() then waits for the slot's MailBox to become non-empty, extracts the message, and clears MailBox. Finally, take() advances the slot's Turn field, which releases both "put" and "take" access to the slot's MailBox. Note the asymmetry : put() acquires "put" access to the slot, but take() releases that lock. At any given time, for a given slot in a PTLQueue, at most one thread has "put" access and at most one thread has "take" access. This restricts concurrency from general MPMC to 1-vs-1. We have 2 ticket locks -- one for put() and one for take() -- each with its own "ticket" variable in the form of the corresponding cursor, but they share a single "Grant" egress variable in the form of the slot's Turn variable. Advancing the PutCursor, for instance, serves two purposes. First, we obtain a unique ticket which identifies a slot. Second, incrementing the cursor is the doorway protocol step to acquire the per-slot mutual exclusion "put" lock. The cursors and operations to increment those cursors serve double-duty : slot-selection and ticket assignment for locking the slot's MailBox field. At any given time a slot MailBox field can be in one of the following states: empty with no pending operations -- neutral state; empty with one or more waiting take() operations pending -- deficit; occupied with no pending operations; occupied with one or more waiting put() operations -- surplus; empty with a pending put() or pending put() and take() operations -- transitional; or occupied with a pending take() or pending put() and take() operations -- transitional. The partial put() and take() operators can be implemented with an atomic fetch-and-increment operation, which may confer a performance advantage over a CAS-based loop. In addition we have independent PutCursor and TakeCursor cursors. Critically, a put() operation modifies PutCursor but does not access the TakeCursor and a take() operation modifies the TakeCursor cursor but does not access the PutCursor. This acts to reduce coherence traffic relative to some other queue designs. It's worth noting that slow threads or obstruction in one slot (or "lane") does not impede or obstruct operations in other slots -- this gives us some degree of obstruction isolation. PTLQueue is not lock-free, however. The implementation above is expressed with polite busy-waiting (Pause) but it's trivial to implement per-slot parking and unparking to deschedule waiting threads. It's also easy to convert the queue to a more general deque by replacing the PutCursor and TakeCursor cursors with Left/Front and Right/Back cursors that can move either direction. Specifically, to push and pop from the "left" side of the deque we would decrement and increment the Left cursor, respectively, and to push and pop from the "right" side of the deque we would increment and decrement the Right cursor, respectively. We used a variation of PTLQueue for message passing in our recent OPODIS 2013 paper. ul { list-style:none; padding-left:0; padding:0; margin:0; margin-left:0; } ul#myTagID { padding: 0px; margin: 0px; list-style:none; margin-left:0;} -- -- There's quite a bit of related literature in this area. I'll call out a few relevant references: Wilson's NYU Courant Institute UltraComputer dissertation from 1988 is classic and the canonical starting point : Operating System Data Structures for Shared-Memory MIMD Machines with Fetch-and-Add. Regarding provenance and priority, I think PTLQueue or queues effectively equivalent to PTLQueue have been independently rediscovered a number of times. See CB-Queue and BNPBV, below, for instance. But Wilson's dissertation anticipates the basic idea and seems to predate all the others. Gottlieb et al : Basic Techniques for the Efficient Coordination of Very Large Numbers of Cooperating Sequential Processors Orozco et al : CB-Queue in Toward high-throughput algorithms on many-core architectures which appeared in TACO 2012. Meneghin et al : BNPVB family in Performance evaluation of inter-thread communication mechanisms on multicore/multithreaded architecture Dmitry Vyukov : bounded MPMC queue (highly recommended) Alex Otenko : US8607249 (highly related). John Mellor-Crummey : Concurrent queues: Practical fetch-and-phi algorithms. Technical Report 229, Department of Computer Science, University of Rochester Thomasson : FIFO Distributed Bakery Algorithm (very similar to PTLQueue). Scott and Scherer : Dual Data Structures I'll propose an optimization left as an exercise for the reader. Say we wanted to reduce memory usage by eliminating inter-slot padding. Such padding is usually "dark" memory and otherwise unused and wasted. But eliminating the padding leaves us at risk of increased false sharing. Furthermore lets say it was usually the case that the PutCursor and TakeCursor were numerically close to each other. (That's true in some use cases). We might still reduce false sharing by incrementing the cursors by some value other than 1 that is not trivially small and is coprime with the number of slots. Alternatively, we might increment the cursor by one and mask as usual, resulting in a logical index. We then use that logical index value to index into a permutation table, yielding an effective index for use in the slot array. The permutation table would be constructed so that nearby logical indices would map to more distant effective indices. (Open question: what should that permutation look like? Possibly some perversion of a Gray code or De Bruijn sequence might be suitable). As an aside, say we need to busy-wait for some condition as follows : "while C == 0 : Pause". Lets say that C is usually non-zero, so we typically don't wait. But when C happens to be 0 we'll have to spin for some period, possibly brief. We can arrange for the code to be more machine-friendly with respect to the branch predictors by transforming the loop into : "if C == 0 : for { Pause; if C != 0 : break; }". Critically, we want to restructure the loop so there's one branch that controls entry and another that controls loop exit. A concern is that your compiler or JIT might be clever enough to transform this back to "while C == 0 : Pause". You can sometimes avoid this by inserting a call to a some type of very cheap "opaque" method that the compiler can't elide or reorder. On Solaris, for instance, you could use :"if C == 0 : { gethrtime(); for { Pause; if C != 0 : break; }}". It's worth noting the obvious duality between locks and queues. If you have strict FIFO lock implementation with local spinning and succession by direct handoff such as MCS or CLH,then you can usually transform that lock into a queue. Hidden commentary and annotations - invisible : * And of course there's a well-known duality between queues and locks, but I'll leave that topic for another blog post. * Compare and contrast : PTLQ vs PTL and MultiLane * Equivalent : Turn; seq; sequence; pos; position; ticket * Put = Lock; Deposit Take = identify and reserve slot; wait; extract & clear; unlock * conceptualize : Distinct PutLock and TakeLock implemented as ticket lock or PTL Distinct arrival cursors but share per-slot "Turn" variable provides exclusive role-based access to slot's mailbox field put() acquires exclusive access to a slot for purposes of "deposit" assigns slot round-robin and then acquires deposit access rights/perms to that slot take() acquires exclusive access to slot for purposes of "withdrawal" assigns slot round-robin and then acquires withdrawal access rights/perms to that slot At any given time, only one thread can have withdrawal access to a slot at any given time, only one thread can have deposit access to a slot Permissible for T1 to have deposit access and T2 to simultaneously have withdrawal access * round-robin for the purposes of; role-based; access mode; access role mailslot; mailbox; allocate/assign/identify slot rights; permission; license; access permission; * PTL/Ticket hybrid Asymmetric usage ; owner oblivious lock-unlock pairing K-exclusion add Grant cursor pass message m from lock to unlock via Slots[] array Cursor performs 2 functions : + PTL ticket + Assigns request to slot in round-robin fashion Deconstruct protocol : explication put() : allocate slot in round-robin fashion acquire PTL for "put" access store message into slot associated with PTL index take() : Acquire PTL for "take" access // doorway step seq = fetchAdd (&Grant, 1) s = &Slots[seq & Mask] // waiting phase while s-Turn != seq : pause Extract : wait for s-mailbox to be full v = s-mailbox s-mailbox = null Release PTL for both "put" and "take" access s-Turn = seq + Mask + 1 * Slot round-robin assignment and lock "doorway" protocol leverage the same cursor and FetchAdd operation on that cursor FetchAdd (&Cursor,1) + round-robin slot assignment and dispersal + PTL/ticket lock "doorway" step waiting phase is via "Turn" field in slot * PTLQueue uses 2 cursors -- put and take. Acquire "put" access to slot via PTL-like lock Acquire "take" access to slot via PTL-like lock 2 locks : put and take -- at most one thread can access slot's mailbox Both locks use same "turn" field Like multilane : 2 cursors : put and take slot is simple 1-capacity mailbox instead of queue Borrow per-slot turn/grant from PTL Provides strict FIFO Lock slot : put-vs-put take-vs-take at most one put accesses slot at any one time at most one put accesses take at any one time reduction to 1-vs-1 instead of N-vs-M concurrency Per slot locks for put/take Release put/take by advancing turn * is instrumental in ... * P-V Semaphore vs lock vs K-exclusion * See also : FastQueues-excerpt.java dice-etc/queue-mpmc-bounded-blocking-circular-xadd/ * PTLQueue is the same as PTLQB - identical * Expedient return; ASAP; prompt; immediately * Lamport's Bakery algorithm : doorway step then waiting phase Threads arriving at doorway obtain a unique ticket number Threads enter in ticket order * In the terminology of Reed and Kanodia a ticket lock corresponds to the busy-wait implementation of a semaphore using an eventcount and a sequencer It can also be thought of as an optimization of Lamport's bakery lock was designed for fault-tolerance rather than performance Instead of spinning on the release counter, processors using a bakery lock repeatedly examine the tickets of their peers --

    Read the article

  • Will my self-taught code be fine, or should I take it to the professional level?

    - by G1i1ch
    Lately I've been getting professional work, hanging out with other programmers, and making friends in the industry. The only thing is I'm 100% self-taught. It's caused my style to extremely deviate from the style of those that are properly trained. It's the techniques and organization of my code that's different. It's a mixture of several things I do. I tend to blend several programming paradigms together. Like Functional and OO. I lean to the Functional side more than OO, but I see the use of OO when something would make more sense as an abstract entity. Like a game object. Next I also go the simple route when doing something. When in contrast, it seems like sometimes the code I see from professional programmers is complicated for the sake of it! I use lots of closures. And lastly, I'm not the best commenter. I find it easier just to read through my code than reading the comment. And most cases I just end up reading the code even if there are comments. Plus I've been told that, because of how simply I write my code, it's very easy to read it. I hear professionally trained programmers go on and on about things like unit tests. Something I've never used before so I haven't even the faintest idea of what they are or how they work. Lots and lots of underscores "_", which aren't really my taste. Most of the techniques I use are straight from me, or a few books I've read. Don't know anything about MVC, I've heard a lot about it though with things like backbone.js. I think it's a way to organize an application. It just confuses me though because by now I've made my own organizational structures. It's a bit of a pain. I can't use template applications at all when learning something new like with Ubuntu's Quickly. I have trouble understanding code that I can tell is from someone trained. Complete OO programming really leaves a bad taste in my mouth, yet that seems to be what EVERYONE else is strictly using. It's left me not that confident in the look of my code, or wondering whether I'll cause sparks when joining a company or maybe contributing to open source projects. In fact I'm rather scared of the fact that people will eventually be checking out my code. Is this just something normal any programmer goes through or should I really look to change up my techniques?

    Read the article

  • Back-sliding into Unmanaged Code

    - by Laila
    It is difficult to write about Microsoft's ambivalence to .NET without mentioning clichés about dog food.  In case you've been away a long time, you'll remember that Microsoft surprised everyone with the speed and energy with which it introduced and evangelised the .NET Framework for managed code. There was good reason for this. Once it became obvious to all that it had sleepwalked into third place as a provider of development languages, behind Borland and Sun, it reacted quickly to attract the best talent in the industry to produce a windows version of the Java runtime, with Bounds-checking, Automatic Garbage collection, structures exception handling and common data types. To develop applications for this managed runtime, it produced several excellent languages, and more are being provided. The only thing Microsoft ever got wrong was to give it a stupid name. The logical step for Microsoft would be to base the entire operating system on the .NET framework, and to re-engineer its own applications. In 2002, Bill Gates, then Microsoft Chairman and Chief Software Architect said about their plans for .NET, "This is a long-term approach. These things don't happen overnight." Now, eight years later, we're still waiting for signs of the 'long-term approach'. Microsoft's vision of an entirely managed operating system has subsided since the Vista fiasco, but stays alive yet dormant as Midori, still being developed by Microsoft Research. This is an Internet-centric fork of the singularity operating system, a research project started in 2003 to build a highly-dependable operating system in which the kernel, device drivers, and applications are all written in managed code. Midori is predicated on the prevalence of connected systems, with provisions for distributed concurrency where application components exist 'in the cloud', and supports a programming model that can tolerate cancellation, intermittent connectivity and latency. It features an entirely new security model that sandboxes applications for increased security. So have Microsoft converted its existing applications to the .NET framework? It seems not. What Windows applications can run on Mono? Very few, it seems. We all thought that .NET spelt the end of DLL Hell and the need for COM interop, but it looks as if Bill Gates' idea of 'not overnight' might stretch to a decade or more. The Operating System has shown only minimal signs of migrating to .NET. Even where the use of .NET has come to dominate, when used for server applications with IIS, IIS itself is still entirely developed in unmanaged code. This is an irritation to Microsoft's greatest supporters who committed themselves fully to the NET framework, only to find parts of the Ambivalent Microsoft Empire quietly backsliding into unmanaged code and the awful C++. It is a strategic mistake that the invigorated Apple didn't make with the Mac OS X Architecture. Cheers, Laila

    Read the article

  • Space partitioning when everything is moving

    - by Roy T.
    Background Together with a friend I'm working on a 2D game that is set in space. To make it as immersive and interactive as possible we want there to be thousands of objects freely floating around, some clustered together, others adrift in empty space. Challenge To unburden the rendering and physics engine we need to implement some sort of spatial partitioning. There are two challenges we have to overcome. The first challenge is that everything is moving so reconstructing/updating the data structure has to be extremely cheap since it will have to be done every frame. The second challenge is the distribution of objects, as said before there might be clusters of objects together and vast bits of empty space and to make it even worse there is no boundary to space. Existing technologies I've looked at existing techniques like BSP-Trees, QuadTrees, kd-Trees and even R-Trees but as far as I can tell these data structures aren't a perfect fit since updating a lot of objects that have moved to other cells is relatively expensive. What I've tried I made the decision that I need a data structure that is more geared toward rapid insertion/update than on giving back the least amount of possible hits given a query. For that purpose I made the cells implicit so each object, given it's position, can calculate in which cell(s) it should be. Then I use a HashMap that maps cell-coordinates to an ArrayList (the contents of the cell). This works fairly well since there is no memory lost on 'empty' cells and its easy to calculate which cells to inspect. However creating all those ArrayLists (worst case N) is expensive and so is growing the HashMap a lot of times (although that is slightly mitigated by giving it a large initial capacity). Problem OK so this works but still isn't very fast. Now I can try to micro-optimize the JAVA code. However I'm not expecting too much of that since the profiler tells me that most time is spent in creating all those objects that I use to store the cells. I'm hoping that there are some other tricks/algorithms out there that make this a lot faster so here is what my ideal data structure looks like: The number one priority is fast updating/reconstructing of the entire data structure Its less important to finely divide the objects into equally sized bins, we can draw a few extra objects and do a few extra collision checks if that means that updating is a little bit faster Memory is not really important (PC game)

    Read the article

  • Interviews: Going Beyond the Technical Quiz

    - by Tony Davis
    All developers will be familiar with the basic format of a technical interview. After a bout of CV-trawling to gauge basic experience, strengths and weaknesses, the interview turns technical. The whiteboard takes center stage and the challenge is set to design a function or query, or solve what on the face of it might seem a disarmingly simple programming puzzle. Most developers will have experienced those few panic-stricken moments, when one’s mind goes as blank as the whiteboard, before un-popping the marker pen, and hopefully one’s mental functions, to work through the problem. It is a way to probe the candidate’s knowledge of basic programming structures and techniques and to challenge their critical thinking. However, these challenges or puzzles, often devised by some of the smartest brains in the development team, have a tendency to become unnecessarily ‘tricksy’. They often seem somewhat academic in nature. While the candidate straight out of IT school might breeze through the construction of a Markov chain, a candidate with bags of practical experience but less in the way of formal training could become nonplussed. Also, a whiteboard and a marker pen make up only a very small part of the toolkit that a programmer will use in everyday work. I remember vividly my first job interview, for a position as technical editor. It went well, but after the usual CV grilling and technical questions, I was only halfway there. Later, they sat me alongside a team of editors, in front of a computer loaded with MS Word and copy of SQL Server Query Analyzer, and my task was to edit a real chapter for a real SQL Server book that they planned to publish, including validating and testing all the code. It was a tough challenge but I came away with a sound knowledge of the sort of work I’d do, and its context. It makes perfect sense, yet my impression is that many organizations don’t do this. Indeed, it is only relatively recently that Red Gate started to move over to this model for developer interviews. Now, instead of, or perhaps in addition to, the whiteboard challenges, the candidate can expect to sit with their prospective team, in front of Visual Studio, loaded with all the useful tools in the developer’s kit (ReSharper and so on) and asked to, for example, analyze and improve a real piece of software. The same principles should apply when interviewing for a database positon. In addition to the usual questions challenging the candidate’s knowledge of such things as b-trees, object permissions, database recovery models, and so on, sit the candidate down with the other database developers or DBAs. Arm them with a copy of Management Studio, and a few other tools, then challenge them to discover the flaws in a stored procedure, and improve its performance. Or present them with a corrupt database and ask them to get the database back online, and discover the cause of the corruption.

    Read the article

  • TechEd 2010 Day Three: The Database Designer (Isn't)

    - by BuckWoody
    Yesterday at TechEd 2010 here in New Orleans I worked the front-booth, answering general SQL Server questions for the masses. I was actually a little surprised to find most of the questions I got were from folks that wanted to know more about Stream Insight and Master Data Services. In past conferences I've been asked a lot of "free consulting" questions, about problems folks have had from older products. I don't mind that a bit - in fact, I'm always happy to help in any way I can. But this time people are really interested in the new features in the product, and I like that they are thinking ahead, not just having to solve problems in production. My presentation was on "Database Design in an Hour". We had the usual fun, and SideShow Bob made an appearance - I kid you not. The guy in the back of the room looked just like Sideshow Bob, so I quickly held a "bes thair" contest, and he won. Duing the presentation, I explain the tools you can use to design databases. I also explain that the "Database Designer" tool in SQL Server Management Studio (SSMS) isn't truly a desinger - it uses non-standard notation, doesn't have a meta-data dictionary, and worst of all, it works at the physical level. In other words, whatever you do in SSMS will automatically change the field/table/relationship structures in the database. We fixed this in SSMS 2008 and higher by adding an option to block that, but the tool is not a good design function nonetheless. To be fair, no one I know of at Microsoft recommends that it is - but I was shocked to hear so many developers in the room defending it as a good tool. I think the main issue for someone who doesn't have to work with Relational Systems a great deal is that it can be difficult to figure out Foreign Keys. The syntax makes them look "backwards", so it's just easier to grab a field and place it on the table you want to point to. There are options. You can download a couple of free tools (CA has a community edition of ER-WIN, Quest has one, and Embarcadero also has one) and if you design more than one or two databases a year, it may be worth buying a true design tool. For years I used Visio, but we changed it so that it doesn't forward-engineer (create the DDL) any more, so it isn't a true design tool either. So investigate those free and not-so-free tools. You'll find they help you in your job - but stay away from the Database Designer in SSMS. Or I'll send Sideshow Bob over there to straighten you out. Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • 2D Skeletal Animation Transformations

    - by Brad Zeis
    I have been trying to build a 2D skeletal animation system for a while, and I believe that I'm fairly close to finishing. Currently, I have the following data structures: struct Bone { Bone *parent; int child_count; Bone **children; double x, y; }; struct Vertex { double x, y; int bone_count; Bone **bones; double *weights; }; struct Mesh { int vertex_count; Vertex **vertices; Vertex **tex_coords; } Bone->x and Bone->y are the coordinates of the end point of the Bone. The starting point is given by (bone->parent->x, bone->parent->y) or (0, 0). Each entity in the game has a Mesh, and Mesh->vertices is used as the bounding area for the entity. Mesh->tex_coords are texture coordinates. In the entity's update function, the position of the Bone is used to change the coordinates of the Vertices that are bound to it. Currently what I have is: void Mesh_update(Mesh *mesh) { int i, j; double sx, sy; for (i = 0; i < vertex_count; i++) { if (mesh->vertices[i]->bone_count == 0) { continue; } sx, sy = 0; for (j = 0; j < mesh->vertices[i]->bone_count; j++) { sx += (/* ??? */) * mesh->vertices[i]->weights[j]; sy += (/* ??? */) * mesh->vertices[i]->weights[j]; } mesh->vertices[i]->x = sx; mesh->vertices[i]->y = sy; } } I think I have everything I need, I just don't know how to apply the transformations to the final mesh coordinates. What tranformations do I need here? Or is my approach just completely wrong?

    Read the article

  • Getting away from a customized Magento 1.4 installation - Magento 1.6, OpenCart, or others?

    - by Phil
    I'm dealing with a Magento 1.4.0.0 Community Edition installation with various undocumented changes to the core (mostly integration with an ERP system), an outdated Sweet Tooth Points & Rewards module and some custom payment providers. It also doubles as a mediocre blogging/CMS system. It has one store each for 3 different languages, with about 40 product categories for a few hundred products. [rant] With no prior experience with any PHP e-commerce systems, I find it very difficult to work with. I attempted to install Magento 1.4.0.0 on my local WAMP dev machine, it installs fine, but the main page or search do not show any products no matter what I do in the backend admin panel. I don't know what's wrong with it, and whatever information I googled is either too old or too new from Magento 1.4. Later I'm given FTP access to the testing server, which neither my manager or I have permission to install XDebug on, as apparantly it runs on the same server as the production server (yikes). Trying to learn how Magento works is torture. I spent a week trying to add some fields into the Onepage Checkout before giving up and went to work on something else. The template system, just like the rest of Magento, is a bloated mishmash of overcomplicated directory structures, weird config xml files and EAV databases. I went into 6 different models and several content blocks in the backend just to change what the front page looks like. With little-to-none helpful and clear documentation (unlike CodeIgniter) and various breaking changes between minor point revisions which makes it hard to find useful information, Magento 1.4 is a developer killer. [/rant] The client is planning to redesign the site and has decided it might as well as move on from this unsustainable, hacky, upgrade-unfriendly, developer-unfriendly mess. Magento 1.4 is starting to show its age, with Magento 1.7 coming soon, the client is considering upgrading to Magento 1.6 or 1.7 if it has improved from 1.4. The customizations done to the current Magento 1.4 installation will have to be redone, and a new license for the Sweet Tooth Points & Rewards module will have to be bought. The client is also open to other e-commerce systems. I've looked at OpenCart and it seems to be quite developer friendly with a fairly simple structure. I found some complaints regarding its performance when the shop has thousands of categories or products, but this is not an issue with the current number of products my client has. It seems to be solid ground for easy customization to bring the rewards system and ERP integration over. What should the client upgrade to in this case?

    Read the article

  • Unit Testing TSQL

    - by Grant Fritchey
    I went through a period of time where I spent a lot of effort figuring out how to set up unit tests for TSQL. It wasn't easy. There are a few tools out there that help, but mostly it involves lots of programming. well, not as much as before. Thanks to the latest Down Tools Week at Red Gate a new utility has been built and released into the wild, SQL Test. Like a lot of the new tools coming out of Red Gate these days, this one is directly integrated into SSMS, which means you're working where you're comfortable and where you already have lots of tools at your disposal. After the install, when you launch SSMS and get connected, you're prompted to install the tSQLt example database. Go for it. It's a quick way to see how the tool works. I'd suggest using it. It' gives you a quick leg up. The concepts are pretty straight forward. There are a series of CLR commands that you use to configure a test and the test assertions. In between you're calling TSQL, either calls to your structure, queries, or stored procedures. They already have the one things that I always found wanting in database tests, a way to compare tables of results. I also like the ability to create a dummy copy of tables for the tests. It lets you control structures and behaviors so that the tests are more focused. One of the issues I always ran into with the other testing tools is that setting up the tests might require potentially destructive changes to the structure of the database (dropping FKs, etc.) which added lots of time and effort to setting up the tests, making testing more difficult, and therefor, less useful. Functionally, this is pretty similar to the Visual Studio tests and TSQLUnit tests that I used to use. The primary improvement over the Visual Studio tests is that I'm working in SSMS instead of Visual Studio. The primary improvement over TSQLUnit is the SQL Test interface it self. A lot of the functionality is the same, but having a sweet little tool to manage & run the tests from makes a huge difference. Oh, and don't worry. You can still run these tests directly from TSQL too, so automation has not gone away. I'm still thinking about how I'd use this in a dev environment where I also had source control to fret. That might be another blog post right there. I'm just getting started with SQL Test, so this is the first of several blog posts & videos. Watch this space. Try the tool.

    Read the article

  • How to TDD test that objects are being added to a collection if the collection is private?

    - by Joshua Harris
    Assume that I planned to write a class that worked something like this: public class GameCharacter { private Collection<CharacterEffect> _collection; public void Add(CharacterEffect e) { ... } public void Remove(CharacterEffect e) { ... } public void Contains(CharacterEffect e) { ... } } When added an effect does something to the character and is then added to the _collection. When it is removed the effect reverts the change to the character and is removed from the _collection. It's easy to test if the effect was applied to the character, but how do I test that the effect was added to _collection? What test could I write to start constructing this class. I could write a test where Contains would return true for a certain effect being in _collection, but I can't arrange a case where that function would return true because I haven't implemented the Add method that is needed to place things in _collection. Ok, so since Contains is dependent on having Add working, then why don't I try to create Add first. Well for my first test I need to try and figure out if the effect was added to the _collection. How would I do that? The only way to see if an effect is in _collection is with the Contains function. The only way that I could think to test this would be to use a FakeCollection that Mocks the Add, Remove, and Contains of a real collection, but I don't want _collection being affected by outside sources. I don't want to add a setEffects(Collection effects) function, because I do not want the class to have that functionality. The one thing that I am thinking could work is this: public class GameCharacter<C extends Collection> { private Collection<CharacterEffect> _collection; public GameCharacter() { _collection = new C<CharacterEffect>(); } } But, that is just silly making me declare what some private data structures type is on every declaration of the character. Is there a way for me to test this without breaking TDD principles while still allowing me to keep my collection private?

    Read the article

  • Finding Z given X & Y coordinates on terrain?

    - by mrky
    I need to know what the most efficient way of finding Z given X & Y coordinates on terrain. My terrain is set up as a grid, each grid block consisting of two triangles, which may be flipped in any direction. I want to move game objects smoothly along the floor of the terrain without "stepping." I'm currently using the following method with unexpected results: double mapClass::getZ(double x, double y) { int vertexIndex = ((floor(y))*width*2)+((floor(x))*2); vec3ray ray = {glm::vec3(x, y, 2), glm::vec3(x, y, 0)}; vec3triangle tri1 = { glmFrom(vertices[vertexIndex].v1), glmFrom(vertices[vertexIndex].v2), glmFrom(vertices[vertexIndex].v3) }; vec3triangle tri2 = { glmFrom(vertices[vertexIndex+1].v1), glmFrom(vertices[vertexIndex+1].v2), glmFrom(vertices[vertexIndex+1].v3) }; glm::vec3 intersect; if (!intersectRayTriangle(tri1, ray, intersect)) { intersectRayTriangle(tri2, ray, intersect); } return intersect.z; } intersectRayTriangle() and glmFrom() are as follows: bool intersectRayTriangle(vec3triangle tri, vec3ray ray, glm::vec3 &worldIntersect) { glm::vec3 barycentricIntersect; if (glm::intersectLineTriangle(ray.origin, ray.direction, tri.p0, tri.p1, tri.p2, barycentricIntersect)) { // Convert barycentric to world coordinates double u, v, w; u = barycentricIntersect.x; v = barycentricIntersect.y; w = 1 - (u+v); worldIntersect.x = (u * tri.p0.x + v * tri.p1.x + w * tri.p2.x); worldIntersect.y = (u * tri.p0.y + v * tri.p1.y + w * tri.p2.y); worldIntersect.z = (u * tri.p0.z + v * tri.p1.z + w * tri.p2.z); return true; } else { return false; } } glm::vec3 glmFrom(s_point3f point) { return glm::vec3(point.x, point.y, point.z); } My convenience structures are defined as: struct s_point3f { GLfloat x, y, z; }; struct s_triangle3f { s_point3f v1, v2, v3; }; struct vec3ray { glm::vec3 origin, direction; }; struct vec3triangle { glm::vec3 p0, p1, p2; }; vertices is defined as: std::vector<s_triangle3f> vertices; Basically, I'm trying to get the intersect of a ray (which is positioned at the x, and y coordinates specified facing pointing downwards toward the terrain) and one of the two triangles on the grid. getZ() rarely returns anything but 0. Other times, the numbers it generates seem to be completely off. Am I taking the wrong approach? Can anyone see a problem with my code? Any help or critique is appreciated!

    Read the article

  • How do I apply skeletal animation from a .x (Direct X) file?

    - by Byte56
    Using the .x format to export a model from Blender, I can load a mesh, armature and animation. I have no problems generating the mesh and viewing models in game. Additionally, I have animations and the armature properly loaded into appropriate data structures. My problem is properly applying the animation to the models. I have the framework for applying the models and the code for selecting animations and stepping through frames. From what I understand, the AnimationKeys inside the AnimationSet supplies the transformations to transform the bind pose to the pose in the animated frame. As small example: Animation { {Armature_001_Bone} AnimationKey { 2; //Position 121; //number of frames 0;3; 0.000000, 0.000000, 0.000000;;, 1;3; 0.000000, 0.000000, 0.005524;;, 2;3; 0.000000, 0.000000, 0.022217;;, ... } AnimationKey { 0; //Quaternion Rotation 121; 0;4; -0.707107, 0.707107, 0.000000, 0.000000;;, 1;4; -0.697332, 0.697332, 0.015710, 0.015710;;, 2;4; -0.684805, 0.684805, 0.035442, 0.035442;;, ... } AnimationKey { 1; //Scale 121; 0;3; 1.000000, 1.000000, 1.000000;;, 1;3; 1.000000, 1.000000, 1.000000;;, 2;3; 1.000000, 1.000000, 1.000000;;, ... } } So, to apply frame 2, I would take the position, rotation and scale from frame 2, create a transformation matrix (call it Transform_A) from them and apply that matrix the vertices controlled by Armature_001_Bone at their weights. So I'd stuff TransformA into my shader and transform the vertex. Something like: vertexPos = vertexPos * bones[ int(bfs_BoneIndices.x) ] * bfs_BoneWeights.x; Where bfs_BoneIndices and bfs_BoneWeights are values specific to the current vertex. When loading in the mesh vertices, I transform them by the rootTransform and the meshTransform. This ensures they're oriented and scaled correctly for viewing the bind pose. The problem is when I create that transformation matrix (using the position, rotation and scale from the animation), it doesn't properly transform the vertex. There's likely more to it than just using the animation data. I also tried applying the bone transform hierarchies, still no dice. Basically I end up with some twisted models. It should also be noted that I'm working in openGL, so any matrix transposes that might need to be applied should be considered. What data do I need and how do I combine it for applying .x animations to models?

    Read the article

  • SceneManagers as systems in entity system or as a core class used by a system?

    - by Hatoru Hansou
    It seems entity systems are really popular here. Links posted by other users convinced me of the power of such system and I decided to try it. (Well, that and my original code getting messy) In my project, I originally had a SceneManager class that maintained needed logic and structures to organize the scene (QuadTree, 2D game). Before rendering I call selectRect() and pass the x,y of the camera and the width and height of the screen and then obtain a minimized list containing only visible entities ordered from back to front. Now with Systems, originally in my first attempt my Render system required to get added all entities it should handle. This may sound like the correct approach but I realized this was not efficient. Trying to optimize It I reused the SceneManager class internally in the Renderer system, but then I realized I needed methods such as selectRect() in others systems too (AI principally) and make the SceneManager accessible globally again. Currently I converted SceneManager to a system, and ended up with the following interface (only relevant methods): /// Base system interface class System { public: virtual void tick (double delta_time) = 0; // (methods to add and remove entities) }; typedef std::vector<Entity*> EntitiesVector; /// Specialized system interface to allow query the scene class SceneManager: public System { public: virtual EntitiesVector& cull () = 0; /// Sets the entity to be used as the camera and replaces previous ones. virtual void setCamera (Entity* entity) = 0; }; class SceneRenderer // Not a system { vitual void render (EntitiesVector& entities) = 0; }; Also I could not guess how to convert renderers to systems. My game separates logic updates from screen updates, my main class have a tick() method and a render() method that may not be called the same times. In my first attempt renderers were systems but they was saved in a separated manager, updated only in render() and not in tick() like all other systems. I realized that was silly and simply created a SceneRenderer interface and give up about converting them to systems, but that may be for another question. Then... something does not feel right, isn't it? If I understood correctly a system should not depend on another or even count with another system exposing an specific interface. Each system should care only about its entities, or nodes (as optimization, so they have direct references to relevant components without having to constantly call the component() or getComponent() method of the entity).

    Read the article

  • 2D Grid Map Connectivity Check (avoiding stack overflow)

    - by SombreErmine
    I am trying to create a routine in C++ that will run before a more expensive A* algorithm that checks to see if two nodes on a 2D grid map are connected or not. What I need to know is a good way to accomplish this sequentially rather than recursively to avoid overflowing the stack. What I've Done Already I've implemented this with ease using a recursive algorithm; however, depending upon different situations it will generate a stack overflow. Upon researching this, I've come to the conclusion that it is overflowing the stack because of too many recursive function calls. I am sure that my recursion does not enter an infinite loop. I generate connected sets at the beginning of the level, and then I use those connected sets to determine connectivity on the fly later. Basically, the generating algorithm starts from left-to-right top-to-bottom. It skips wall nodes and marks them as visited. Whenever it reaches a walkable node, it recursively checks in all four cardinal directions for connected walkable nodes. Every node that gets checked is marked as visited so they aren't handled twice. After checking a node, it is added to either a walls set, a doors set, or one of multiple walkable nodes sets. Once it fills that area, it continues the original ltr ttb loop skipping already-visited nodes. I've also looked into flood-fill algorithms, but I can't make sense of the sequential algorithms and how to adapt them. Can anyone suggest a better way to accomplish this without causing a stack overflow? The only way I can think of is to do the left-to-right top-to-bottom loop generating connected sets on a row basis. Then check the previous row to see if any of the connected sets are connected and then join the sets that are. I haven't decided on the best data structures to use for that though. I also just thought about having the connected sets pre-generated outside the game, but I wouldn't know where to start with creating a tool for that. Any help is appreciated. Thanks!

    Read the article

< Previous Page | 55 56 57 58 59 60 61 62 63 64 65 66  | Next Page >