Search Results

Search found 22308 results on 893 pages for 'floating point'.

Page 88/893 | < Previous Page | 84 85 86 87 88 89 90 91 92 93 94 95  | Next Page >

  • google maps plot route between two points

    - by amarsh-anand
    I have written this innocent javascript code, which lets the user create two markers and plot the route between them. It doesnt work, instead, it gives a weird error: Uncaught TypeError: Cannot read property 'ya' of undefined Can someone suggest me whats wrong here: // called upon a click GEvent.addListener(map, "click", function(overlay,point) { if (isCreateHeadPoint) { // add the head marker headMarker = new GMarker(point,{icon:redIcon,title:'Head'}); map.addOverlay(headMarker); isCreateHeadPoint = false; } else { // add the tail marker tailMarker = new GMarker(point,{icon:greenIcon,title:'Tail'}); map.addOverlay(tailMarker); isCreateHeadPoint = true; // create a path from head to tail direction.load("from:" + headMarker.getPoint().lat()+ ", " + headMarker.getPoint().lng()+ " to:" + tailMarker.getPoint().lat() + "," + tailMarker.getPoint().lng(), { getPolyline: true, getSteps: true }); // display the path map.addOverlay(direction.getPolyline()); } });

    Read the article

  • link div to input text in firefox

    - by golf_nut
    I'm using google maps api to have a map show up on my website where visitors can also find directions. I have a starting point setup and an end point. Both are text areas. The start is empty, but I want the end point to be populated with the address from a div that I already have in the page. I have the javascript as: var address = document.getElementById("address").innerText.replace(/\n/g, " "); It works in Chrome, but Firefox shows the text in the endpoint box as 'undefined.' And Firebug is giving me - document.getElementById("address").innerText is undefined [Break On This Error] var address = document.getElementB...ress").innerText.replace(/\n/g, " "); Any help is appreciated. Thanks

    Read the article

  • How do I write an overload operator where both arguments are interface

    - by Eric Girard
    I'm using interface for most of my stuff. I can't find a way to create an overload operator + that would allow me to perform an addition on any objects implementing the IPoint interface Code interface IPoint { double X { get; set; } double Y { get; set; } } class Point : IPoint { double X { get; set; } double Y { get; set; } //How and where do I create this operator/extension ??? public static IPoint operator + (IPoint a,IPoint b) { return Add(a,b); } public static IPoint Add(IPoint a,IPoint b) { return new Point { X = a.X + b.X, Y = a.Y + b.Y }; } } //Dumb use case : public class Test { IPoint _currentLocation; public Test(IPoint initialLocation) { _currentLocation = intialLocation } public MoveOf(IPoint movement) { _currentLocation = _currentLocation + intialLocation; //Much cleaner/user-friendly than _currentLocation = Point.Add(_currentLocation,intialLocation); } }

    Read the article

  • Faster way to compare two sets of points in N-dimensional space?

    - by Amit
    List1 contains a high number (~7^10) of N-dimensional points (N <=10), List2 contains the same or fewer number of N-dimensional points (N <=10). My task is this: I want to check which point in List2 is closest (euclidean distance) to a point in List1 for every point in List1 and subsequently perform some operation on it. I have been doing it the simple- the nested loop way when I didn't have more than 50 points in List1, but with 7^10 points, this obviously takes up a lot of time. What is the fastest way to do this? Any concepts from Computational Geometry might help?

    Read the article

  • how to control textbox type to double in visual basic?

    - by fema
    Hi, I'd like to make a textbox that accepts only numbers, but not integer, but rather double. I've read here about e.Handled = Not Char.IsDigit(e.KeyChar) and it works, but again, it can be used only for integer, since it declines decimal point. Another thing I've read here is If Not Double.TryParse(TextBox2.Text, value) Then .... and it would work fine, except that it allows only decimal comma instead of point. I don't know whether it's because of my location settings (Hungary, we use commas instead of points), but I don't have any other idea how to solve my problem, and the SQL server I send my data uses decimal point. Thanks in advance.

    Read the article

  • Java Custom exception throw behaves differently between different Projects

    - by Pablo
    I am attempting to call the following in my code: public void checkParticleLightRestriction(Particle parent) throws LightException { if ( parent == null ) { throw new LightException("quantum-particle-restrict.23", this); } In one Project the exception is thrown and the effect is similar to calling "return" whereby I am returned back to the point immediately succeeding where this method was called. However in another Project I get thrown completed out of the current package and to a point way prior to the point preceeding this method. It likes instead of being kicked out of a bar I am being deported all the way out of the country. My option are the wrap the throw in a try / catch but I am wondering why this difference in behaviour beween the 2 projects ?

    Read the article

  • Bouncing a ball off a surface

    - by Sagekilla
    Hi all, I'm currently in the middle of writing a game like Breakout, and I was wondering how I could properly bounce a ball off a surface. I went with the naive way of rotating the velocity by 180 degrees, which was: [vx, vy] -> [-vy, vx] Which (unsurprisingly) didn't work so well. If I know the position and veocity of the ball, as well as the point the ball would hit (but is going to instead bounce off of) how can I bounce it off that point? I don't need any language specific code. If anyone could provide a small, mathematical formula on how to properly do this that would work fine for me. I also need this to work with integer positions and velocity (I can't use floating point anywhere). Thanks!

    Read the article

  • Which kind of method signature do you prefer and why?

    - by devoured elysium
    Ok, this is probably highly subjective but here it comes: Let's assume I'm writing a method that will take a printscreen of some region of the screen. Which method signature would you prefer and why? Bitmap DoPrintScreen(int x, int y, int width, int height); Bitmap DoPrintScreen(Rectangle rect); Bitmap DoPrintScreen(Point point, Size size); Other Why? I keep seeing myself repeatedly implementing both 1) and 2) (redirecting one of them to the other) but I end up usually just using one of them, so there really is no point in having both. I can't decide which would be better. Maybe I should use the signature that looks the most with the method I'll be calling to make the printscreen?

    Read the article

  • Detecting coincident subset of two coincident line segments

    - by Jared Updike
    This question is related to: How do I determine the intersection point of two lines in GDI+? (great explanation of algebra but no code) How do you detect where two line segments intersect? (accepted answer doesn't actually work) But note that an interesting sub-problem is completely glossed over in most solutions which just return null for the coincident case even though there are three sub-cases: coincident but do not overlap touching just points and coincident overlap/coincident line sub-segment For example we could design a C# function like this: public static PointF[] Intersection(PointF a1, PointF a2, PointF b1, PointF b2) where (a1,a2) is one line segment and (b1,b2) is another. This function would need to cover all the weird cases that most implementations or explanations gloss over. In order to account for the weirdness of coincident lines, the function could return an array of PointF's: zero result points (or null) if the lines are parallel or do not intersect (infinite lines intersect but line segments are disjoint, or lines are parallel) one result point (containing the intersection location) if they do intersect or if they are coincident at one point two result points (for the overlapping part of the line segments) if the two lines are coincident

    Read the article

  • Why does the CPU window always appear when trying to debug my project after a rebuild in Delphi 2010

    - by James
    Hi, Whenever I rebuild my application and try to step into a break-point the CPU window always appears. From what I understand the CPU window appears when DCU does not match up with the source file, however, in my case the DCU's are defintely being re-compiled and there are no old ones lying around or anything like that. The strange thing here is if I close down the application and re-open the project, place the breakpoint and run it works fine. I can even modify files and just press F9 to run the project and it works fine....it only seems to be when I rebuild the project (via IDE Project menu or the project context menu) that this CPU window never breaks into the source (even though it can find it!). I also noticed in the callstack a unit called Generics is always the top of the stack, never the unit my break point is in, this is no way related to where my break point is. Any ideas?

    Read the article

  • PropertyUtils performance

    - by mR_fr0g
    I have a problem where i need to walk through an object graph and pick out a particular property value. My original solution caches a linked list of property names that need to be applied in order to get from point A to point B in the object graph. I then use apache commons PropertyUtils to iterate through the linked list calling getProperty(Object bean, String name) until i have reached point B. My question is around how this will perform compared to perhaps cahing the Method objects for each step. What is propertyUtils doing under the bonnet? Is it doing a lot of reflection / heavy lifting?

    Read the article

  • Determining if two rays intersect

    - by Faken
    I have two rays on a 2D plane that extend to infinity but both have a starting point. They are both described by a starting point and a vector in the direction of the ray extending to infinity. I want to find out if the two rays intersect but i don't need to know where they intersect (its part of a collision detection algorithm). Everything i have looked at so far describes finding the intersection point of two lines or line segments. Anyone know a fast algorithm to solve this?

    Read the article

  • C++0x optimizing compiler quality

    - by aaa
    hello. I do some heavy numbercrunching and for me floating-point performance is very important. I like performance of Intel compiler very much and quite content with quality of assembly it produces. I am thinking at some point to try C++0x mainly for sugar parts, like auto, initializer list, etc, but also lambdas. at this point I use those features in regular C++ by the means of boost. How good of assembly code do compilers C++0x generate? specifically Intel and gcc compilers. Do they produce SSE code? is performance comparable to C++? are there any benchmarks? My Google search did not reveal much. Thank you.

    Read the article

  • How can I take zooming into account when a user touches a UIScrollView?

    - by Bill
    I have a UIImageView inside of a UIScrollView. The parent scroll view allows zooming and panning. When the user taps a point in the scroll view, I want to find the location in the raw image inside the UIImageView - i.e. I want the point after including any zooming and panning the user has done in the scroll view. Right now, I have a UIScrollView subclass called ForwardingScrollView that handles touch events and attempts to convert them into locations in the coordinate system of the child image view. I tried adding contentOffset to these points, tried multiplying them by zoomScale, and even tried doing both. I also tried calling [touch locationInView: self] and [touch locationInView: parent], but none of these methods correctly return the point that I clicked in the underlying image. What's the best way to do this? Thanks in advance.

    Read the article

  • deleting dynamically allocated object that contains vector in C++ STL

    - by kobac
    I have a class class ChartLine{ protected: vector<Point> line; // points connecting the line CString name; //line name for legend CPen pen; //color, size and style properties of the line }; where Point is a structure struct Point{ CString x; double y; }; In main() I dynamically allocate objects of type ChartLine with new operator. If I use delete afterwards, will default destructor ~ChartLine() properly dealocate (or clear) member ChartLine::line(which is vector btw) or I would have to clear that vector in ~ChartLine() manually? Thanks in advance. Cheers.

    Read the article

  • Normalized Device Coordinates to window coordinates

    - by okoman
    I just read some stuff about the theory behind 3d graphics. As I understand it, normalized device coordinates (NDC) are coordinates that describe a point in the interval from -1 to 1 on both the horizontal and vertical axis. On the other hand window coordinates describe a point somewhere between (0,0) and (width,height) of the window. So my formula to convert a point from the NDC coordinate system to the window system would be xwin = width + xndc * 0.5 * width ywin = height + ynfv * 0.5 * height The problem now is that in the OpenGL documentation for glViewport there is an other formula: xwin = ( xndc + 1 ) * width * 0.5 + x ywin = ( yndc + 1 ) * height * 0.5 + y Now I'm wondering what I am getting wrong. Especially I'm wondering what the additional "x" and "y" mean. Hope the question isn't too "not programming related", but I thought somehow it is related to graphics programming.

    Read the article

  • how to make struct member pointer in assembly?

    - by sillis
    I`m trying to create a macro which would make easier to point to a structs member. Currently i am pointing to a structs member in assembly file using the STRUCT_NAME + offset method. For example if i want to point structs third member,i would have to do it like this: STRUCT_NAME + 3. This seems stupid way to do it, and if i insert more members in the struct, i have to update all the offset values in the code. Is there a way to point using STRUCT_NAME + macro(struct_name, member_name) ? I`m using texas instruments TMS320C28x hardware. Thanks!

    Read the article

  • Mapview on tablet: How can I center the map with an offset?

    - by Waza_Be
    Hint: Here is a similar post with HTML. In the current tablet implementation of my app, I have a fullscreen MapView with some informations displayed in a RelativeLayout on a left panel, like this: (My layout is quite trivial, and I guess there is no need to post it for readability) The problem comes when I want to center the map on a specific point... If I use this code: mapController.setCenter(point); I will of course get the point in the center of the screen and not in the center of the empty area. I have really no idea where I could start to turn the offset of the left panel into map coordinates... Thanks a lot for any help or suggestion

    Read the article

  • Is there a tool that automatically saves incremental changes to files while coding?

    - by Bob.
    One of my favorite features of Google docs is the fact that it's constantly automatically saving versions of my document as I work. This means that even if I forget to save at a certain point before making a critical change there's a good chance that a save point has been created automatically. At the very least, I can return the document to a state prior to the mistaken change and continue working from that point. Is there a tool with an equivalent feature for a Ruby coder running on Mac OS (or UNIX)? For example, a tool that will do an automatic Git check-in every couple of minutes to my local repository for the files I'm working on. Maybe I'm paranoid, but this small bit of insurance could put my mind at ease during my day-to-day work.

    Read the article

  • Rails 3 many-to-many query on includes or joins

    - by Myat
    I have three models User, Activity and ActivityRecord. class User < ActiveRecord::Base # Include default devise modules. Others available are: # :token_authenticatable, :confirmable, # :lockable, :timeoutable and :omniauthable devise :database_authenticatable, :registerable, :recoverable, :rememberable, :trackable, :validatable # Setup accessible (or protected) attributes for your model attr_accessible :first_name, :last_name, :email, :gender, :password, :password_confirmation, :remember_me # attr_accessible :title, :body has_many :activities has_many :activity_records , :through=> :activities end class Activity < ActiveRecord::Base attr_accessible :point, :title belongs_to :user has_many :activity_records end class ActivityRecord < ActiveRecord::Base attr_accessible :activity_id belongs_to :activity scope :today, lambda { where("DATE(#{'activity_records'}.created_at) = '#{Date.today.to_s(:db)}'")} end I would like to query all activities for a user together with the count for their respective activity records for today. For example, after querying and converting to json format, I would like to have something like below [ { id: 23 title: "jogging", point: "5", today_activity_records_count: 1, }, { id: 12 title: "diet dinner", point: "2", today_activity_records_count: 0, }, ] Please kindly guide me how I can achieve that. Thanks

    Read the article

  • How to edit item in a listbox shown from reading a .csv file?

    - by Shuvo
    I am working in a project where my application can open a .csv file and read data from it. The .csv file contains the latitude, longitude of places. The application reads data from the file shows it in a static map and display icon on the right places. The application can open multiple file at a time and it opens with a new tab every time. But I am having trouble in couple of cases When I am trying to add a new point to the .csv file opened. I am able to write new point on the same file instead adding a new point data to the existing its replacing others and writing the new point only. I cannot use selectedIndexChange event to perform edit option on the listbox and then save the file. Any direction would be great. using System; using System.Collections.Generic; using System.ComponentModel; using System.Data; using System.Drawing; using System.Linq; using System.Text; using System.Windows.Forms; using System.IO; namespace CourseworkExample { public partial class Form1 : Form { public GPSDataPoint gpsdp; List<GPSDataPoint> data; List<PictureBox> pictures; List<TabPage> tabs; public static int pn = 0; private TabPage currentComponent; private Bitmap bmp1; string[] symbols = { "hospital", "university" }; Image[] symbolImages; ListBox lb = new ListBox(); string name = ""; string path = ""; public Form1() { InitializeComponent(); data = new List<GPSDataPoint>(); pictures = new List<PictureBox>(); tabs = new List<TabPage>(); symbolImages = new Image[symbols.Length]; for (int i = 0; i < 2; i++) { string location = "data/" + symbols[i] + ".png"; symbolImages[i] = Image.FromFile(location); } } private void openToolStripMenuItem_Click(object sender, EventArgs e) { FileDialog ofd = new OpenFileDialog(); string filter = "CSV File (*.csv)|*.csv"; ofd.Filter = filter; DialogResult dr = ofd.ShowDialog(); if (dr.Equals(DialogResult.OK)) { int i = ofd.FileName.LastIndexOf("\\"); name = ofd.FileName; path = ofd.FileName; if (i > 0) { name = ofd.FileName.Substring(i + 1); path = ofd.FileName.Substring(0, i + 1); } TextReader input = new StreamReader(ofd.FileName); string mapName = input.ReadLine(); GPSDataPoint gpsD = new GPSDataPoint(); gpsD.setBounds(input.ReadLine()); string s; while ((s = input.ReadLine()) != null) { gpsD.addWaypoint(s); } input.Close(); TabPage tabPage = new TabPage(); tabPage.Location = new System.Drawing.Point(4, 22); tabPage.Name = "tabPage" + pn; lb.Width = 300; int selectedindex = lb.SelectedIndex; lb.Items.Add(mapName); lb.Items.Add("Bounds"); lb.Items.Add(gpsD.Bounds[0] + " " + gpsD.Bounds[1] + " " + gpsD.Bounds[2] + " " + gpsD.Bounds[3]); lb.Items.Add("Waypoint"); foreach (WayPoint wp in gpsD.DataList) { lb.Items.Add(wp.Name + " " + wp.Latitude + " " + wp.Longitude + " " + wp.Ele + " " + wp.Sym); } tabPage.Controls.Add(lb); pn++; tabPage.Padding = new System.Windows.Forms.Padding(3); tabPage.Size = new System.Drawing.Size(192, 74); tabPage.TabIndex = 0; tabPage.Text = name; tabPage.UseVisualStyleBackColor = true; tabs.Add(tabPage); tabControl1.Controls.Add(tabPage); tabPage = new TabPage(); tabPage.Location = new System.Drawing.Point(4, 22); tabPage.Name = "tabPage" + pn; pn++; tabPage.Padding = new System.Windows.Forms.Padding(3); tabPage.Size = new System.Drawing.Size(192, 74); tabPage.TabIndex = 0; tabPage.Text = mapName; string location = path + mapName; tabPage.UseVisualStyleBackColor = true; tabs.Add(tabPage); PictureBox pb = new PictureBox(); pb.Name = "pictureBox" + pn; pb.Image = Image.FromFile(location); tabControl2.Controls.Add(tabPage); pb.Width = pb.Image.Width; pb.Height = pb.Image.Height; tabPage.Controls.Add(pb); currentComponent = tabPage; tabPage.Width = pb.Width; tabPage.Height = pb.Height; pn++; tabControl2.Width = pb.Width; tabControl2.Height = pb.Height; bmp1 = (Bitmap)pb.Image; int lx, ly; float realWidth = gpsD.Bounds[1] - gpsD.Bounds[3]; float imageW = pb.Image.Width; float dx = imageW * (gpsD.Bounds[1] - gpsD.getWayPoint(0).Longitude) / realWidth; float realHeight = gpsD.Bounds[0] - gpsD.Bounds[2]; float imageH = pb.Image.Height; float dy = imageH * (gpsD.Bounds[0] - gpsD.getWayPoint(0).Latitude) / realHeight; lx = (int)dx; ly = (int)dy; using (Graphics g = Graphics.FromImage(bmp1)) { Rectangle rect = new Rectangle(lx, ly, 20, 20); if (gpsD.getWayPoint(0).Sym.Equals("")) { g.DrawRectangle(new Pen(Color.Red), rect); } else { if (gpsD.getWayPoint(0).Sym.Equals("hospital")) { g.DrawImage(symbolImages[0], rect); } else { if (gpsD.getWayPoint(0).Sym.Equals("university")) { g.DrawImage(symbolImages[1], rect); } } } } pb.Image = bmp1; pb.Invalidate(); } } private void openToolStripMenuItem_Click_1(object sender, EventArgs e) { FileDialog ofd = new OpenFileDialog(); string filter = "CSV File (*.csv)|*.csv"; ofd.Filter = filter; DialogResult dr = ofd.ShowDialog(); if (dr.Equals(DialogResult.OK)) { int i = ofd.FileName.LastIndexOf("\\"); name = ofd.FileName; path = ofd.FileName; if (i > 0) { name = ofd.FileName.Substring(i + 1); path = ofd.FileName.Substring(0, i + 1); } TextReader input = new StreamReader(ofd.FileName); string mapName = input.ReadLine(); GPSDataPoint gpsD = new GPSDataPoint(); gpsD.setBounds(input.ReadLine()); string s; while ((s = input.ReadLine()) != null) { gpsD.addWaypoint(s); } input.Close(); TabPage tabPage = new TabPage(); tabPage.Location = new System.Drawing.Point(4, 22); tabPage.Name = "tabPage" + pn; ListBox lb = new ListBox(); lb.Width = 300; lb.Items.Add(mapName); lb.Items.Add("Bounds"); lb.Items.Add(gpsD.Bounds[0] + " " + gpsD.Bounds[1] + " " + gpsD.Bounds[2] + " " + gpsD.Bounds[3]); lb.Items.Add("Waypoint"); foreach (WayPoint wp in gpsD.DataList) { lb.Items.Add(wp.Name + " " + wp.Latitude + " " + wp.Longitude + " " + wp.Ele + " " + wp.Sym); } tabPage.Controls.Add(lb); pn++; tabPage.Padding = new System.Windows.Forms.Padding(3); tabPage.Size = new System.Drawing.Size(192, 74); tabPage.TabIndex = 0; tabPage.Text = name; tabPage.UseVisualStyleBackColor = true; tabs.Add(tabPage); tabControl1.Controls.Add(tabPage); tabPage = new TabPage(); tabPage.Location = new System.Drawing.Point(4, 22); tabPage.Name = "tabPage" + pn; pn++; tabPage.Padding = new System.Windows.Forms.Padding(3); tabPage.Size = new System.Drawing.Size(192, 74); tabPage.TabIndex = 0; tabPage.Text = mapName; string location = path + mapName; tabPage.UseVisualStyleBackColor = true; tabs.Add(tabPage); PictureBox pb = new PictureBox(); pb.Name = "pictureBox" + pn; pb.Image = Image.FromFile(location); tabControl2.Controls.Add(tabPage); pb.Width = pb.Image.Width; pb.Height = pb.Image.Height; tabPage.Controls.Add(pb); currentComponent = tabPage; tabPage.Width = pb.Width; tabPage.Height = pb.Height; pn++; tabControl2.Width = pb.Width; tabControl2.Height = pb.Height; bmp1 = (Bitmap)pb.Image; int lx, ly; float realWidth = gpsD.Bounds[1] - gpsD.Bounds[3]; float imageW = pb.Image.Width; float dx = imageW * (gpsD.Bounds[1] - gpsD.getWayPoint(0).Longitude) / realWidth; float realHeight = gpsD.Bounds[0] - gpsD.Bounds[2]; float imageH = pb.Image.Height; float dy = imageH * (gpsD.Bounds[0] - gpsD.getWayPoint(0).Latitude) / realHeight; lx = (int)dx; ly = (int)dy; using (Graphics g = Graphics.FromImage(bmp1)) { Rectangle rect = new Rectangle(lx, ly, 20, 20); if (gpsD.getWayPoint(0).Sym.Equals("")) { g.DrawRectangle(new Pen(Color.Red), rect); } else { if (gpsD.getWayPoint(0).Sym.Equals("hospital")) { g.DrawImage(symbolImages[0], rect); } else { if (gpsD.getWayPoint(0).Sym.Equals("university")) { g.DrawImage(symbolImages[1], rect); } } } } pb.Image = bmp1; pb.Invalidate(); MessageBox.Show(data.ToString()); } } private void exitToolStripMenuItem_Click(object sender, EventArgs e) { this.Close(); } private void addBtn_Click(object sender, EventArgs e) { string wayName = nameTxtBox.Text; float wayLat = Convert.ToSingle(latTxtBox.Text); float wayLong = Convert.ToSingle(longTxtBox.Text); float wayEle = Convert.ToSingle(elevTxtBox.Text); WayPoint wp = new WayPoint(wayName, wayLat, wayLong, wayEle); GPSDataPoint gdp = new GPSDataPoint(); data = new List<GPSDataPoint>(); gdp.Add(wp); lb.Items.Add(wp.Name + " " + wp.Latitude + " " + wp.Longitude + " " + wp.Ele + " " + wp.Sym); lb.Refresh(); StreamWriter sr = new StreamWriter(name); sr.Write(lb); sr.Close(); DialogResult result = MessageBox.Show("Save in New File?","Save", MessageBoxButtons.YesNo); if (result == DialogResult.Yes) { SaveFileDialog saveDialog = new SaveFileDialog(); saveDialog.FileName = "default.csv"; DialogResult saveResult = saveDialog.ShowDialog(); if (saveResult == DialogResult.OK) { sr = new StreamWriter(saveDialog.FileName, true); sr.WriteLine(wayName + "," + wayLat + "," + wayLong + "," + wayEle); sr.Close(); } } else { // sr = new StreamWriter(name, true); // sr.WriteLine(wayName + "," + wayLat + "," + wayLong + "," + wayEle); sr.Close(); } MessageBox.Show(name + path); } } } GPSDataPoint.cs using System; using System.Collections.Generic; using System.Linq; using System.Text; using System.IO; namespace CourseworkExample { public class GPSDataPoint { private float[] bounds; private List<WayPoint> dataList; public GPSDataPoint() { dataList = new List<WayPoint>(); } internal void setBounds(string p) { string[] b = p.Split(','); bounds = new float[b.Length]; for (int i = 0; i < b.Length; i++) { bounds[i] = Convert.ToSingle(b[i]); } } public float[] Bounds { get { return bounds; } } internal void addWaypoint(string s) { WayPoint wp = new WayPoint(s); dataList.Add(wp); } public WayPoint getWayPoint(int i) { if (i < dataList.Count) { return dataList[i]; } else return null; } public List<WayPoint> DataList { get { return dataList; } } internal void Add(WayPoint wp) { dataList.Add(wp); } } } WayPoint.cs using System; using System.Collections.Generic; using System.Linq; using System.Text; namespace CourseworkExample { public class WayPoint { private string name; private float ele; private float latitude; private float longitude; private string sym; public WayPoint(string name, float latitude, float longitude, float elevation) { this.name = name; this.latitude = latitude; this.longitude = longitude; this.ele = elevation; } public WayPoint() { name = "no name"; ele = 3.5F; latitude = 3.5F; longitude = 0.0F; sym = "no symbol"; } public WayPoint(string s) { string[] bits = s.Split(','); name = bits[0]; longitude = Convert.ToSingle(bits[2]); latitude = Convert.ToSingle(bits[1]); if (bits.Length > 4) sym = bits[4]; else sym = ""; try { ele = Convert.ToSingle(bits[3]); } catch (Exception e) { ele = 0.0f; } } public float Longitude { get { return longitude; } set { longitude = value; } } public float Latitude { get { return latitude; } set { latitude = value; } } public float Ele { get { return ele; } set { ele = value; } } public string Name { get { return name; } set { name = value; } } public string Sym { get { return sym; } set { sym = value; } } } } .csv file data birthplace.png 51.483788,-0.351906,51.460745,-0.302982 Born Here,51.473805,-0.32532,-,hospital Danced here,51,483805,-0.32532,-,hospital

    Read the article

  • Upgrading from TFS 2010 RC to TFS 2010 RTM done

    - by Martin Hinshelwood
    Today is the big day, with the Launch of Visual Studio 2010 already done in Asia, and rolling around the world towards us, we are getting ready for the RTM (Released). We have had TFS 2010 in Production for nearly 6 months and have had only minimal problems. Update 12th April 2010  – Added Scott Hanselman’s tweet about the MSDN download release time. SSW was the first company in the world outside of Microsoft to deploy Visual Studio 2010 Team Foundation Server to production, not once, but twice. I am hoping to make it 3 in a row, but with all the hype around the new version, and with it being a production release and not just a go-live, I think there will be a lot of competition. Developers: MSDN will be updated with #vs2010 downloads and details at 10am PST *today*! @shanselman - Scott Hanselman Same as before, we need to Uninstall 2010 RC and install 2010 RTM. The installer will take care of all the complexity of actually upgrading any schema changes. If you are upgrading from TFS 2008 to TFS2010 you can follow our Rules To Better TFS 2010 Migration and read my post on our successes.   We run TFS 2010 in a Hyper-V virtual environment, so we have the advantage of running a snapshot as well as taking a DB backup. Done - Snapshot the hyper-v server Microsoft does not support taking a snapshot of a running server, for very good reason, and Brian Harry wrote a post after my last upgrade with the reason why you should never snapshot a running server. Done - Uninstall Visual Studio Team Explorer 2010 RC You will need to uninstall all of the Visual Studio 2010 RC client bits that you have on the server. Done - Uninstall TFS 2010 RC Done - Install TFS 2010 RTM Done - Configure TFS 2010 RTM Pick the Upgrade option and point it at your existing “tfs_Configuration” database to load all of the existing settings Done - Upgrade the SharePoint Extensions Upgrade Build Servers (Pending) Test the server The back out plan, and you should always have one, is to restore the snapshot. Upgrading to Team Foundation Server 2010 – Done The first thing you need to do is off the TFS server and then log into the Hyper-v server and create a snapshot. Figure: Make sure you turn the server off and delete all old snapshots before you take a new one I noticed that the snapshot that was taken before the Beta 2 to RC upgrade was still there. You should really delete old snapshots before you create a new one, but in this case the SysAdmin (who is currently tucked up in bed) asked me not to. I guess he is worried about a developer messing up his server Turn your server on and wait for it to boot in anticipation of all the nice shiny RTM’ness that is coming next. The upgrade procedure for TFS2010 is to uninstal the old version and install the new one. Figure: Remove Visual Studio 2010 Team Foundation Server RC from the system.   Figure: Most of the heavy lifting is done by the Uninstaller, but make sure you have removed any of the client bits first. Specifically Visual Studio 2010 or Team Explorer 2010.  Once the uninstall is complete, this took around 5 minutes for me, you can begin the install of the RTM. Running the 64 bit OS will allow the application to use more than 2GB RAM, which while not common may be of use in heavy load situations. Figure: It is always recommended to install the 64bit version of a server application where possible. I do not think it is likely, with SharePoint 2010 and Exchange 2010  and even Windows Server 2008 R2 being 64 bit only, I do not think there will be another release of a server app that is 32bit. You then need to choose what it is you want to install. This depends on how you are running TFS and on how many servers. In our case we run TFS and the Team Foundation Build Service (controller only) on out TFS server along with Analysis services and Reporting Services. But our SharePoint server lives elsewhere. Figure: This always confuses people, but in reality it makes sense. Don’t install what you do not need. Every extra you install has an impact of performance. If you are integrating with SharePoint you will need to run this install on every Front end server in your farm and don’t forget to upgrade your Build servers and proxy servers later. Figure: Selecting only Team Foundation Server (TFS) and Team Foundation Build Services (TFBS)   It is worth noting that if you have a lot of builds kicking off, and hence a lot of get operations against your TFS server, you can use a proxy server to cache the source control on another server in between your TFS server and your build servers. Figure: Installing Microsoft .NET Framework 4 takes the most time. Figure: Now run Windows Update, and SSW Diagnostic to make sure all your bits and bobs are up to date. Note: SSW Diagnostic will check your Power Tools, Add-on’s, Check in Policies and other bits as well. Configure Team Foundation Server 2010 – Done Now you can configure the server. If you have no key you will need to pick “Install a Trial Licence”, but it is only £500, or free with a MSDN subscription. Anyway, if you pick Trial you get 90 days to get your key. Figure: You can pick trial and add your key later using the TFS Server Admin. Here is where the real choices happen. We are doing an Upgrade from a previous version, so I will pick Upgrade the same as all you folks that are using the RC or TFS 2008. Figure: The upgrade wizard takes your existing 2010 or 2008 databases and upgraded them to the release.   Once you have entered your database server name you can click “List available databases” and it will show what it can upgrade. Figure: Select your database from the list and at this point, make sure you have a valid backup. At this point you have not made ANY changes to the databases. At this point the configuration wizard will load configuration from your existing database if you have one. If you are upgrading TFS 2008 refer to Rules To Better TFS 2010 Migration. Mostly during the wizard the default values will suffice, but depending on the configuration you want you can pick different options. Figure: Set the application tier account and Authentication method to use. We use NTLM to keep things simple as we host our TFS server externally for our remote developers.  Figure: Setting your TFS server URL’s to be the remote URL’s allows the reports to be accessed without using VPN. Very handy for those remote developers. Figure: Detected the existing Warehouse no problem. Figure: Again we love green ticks. It gives us a warm fuzzy feeling. Figure: The username for connecting to Reporting services should be a domain account (if you are on a domain that is). Figure: Setup the SharePoint integration to connect to your external SharePoint server. You can take the option to connect later.   You then need to run all of your readiness checks. These check can save your life! it will check all of the settings that you have entered as well as checking all the external services are configures and running properly. There are two reasons that TFS 2010 is so easy and painless to install where previous version were not. Microsoft changes the install to two steps, Install and configuration. The second reason is that they have pulled out all of the stops in making the install run all the checks necessary to make sure that once you start the install that it will complete. if you find any errors I recommend that you report them on http://connect.microsoft.com so everyone can benefit from your misery.   Figure: Now we have everything setup the configuration wizard can do its work.  Figure: Took a while on the “Web site” stage for some point, but zipped though after that.  Figure: last wee bit. TFS Needs to do a little tinkering with the data to complete the upgrade. Figure: All upgraded. I am not worried about the yellow triangle as SharePoint was being a little silly Exception Message: TF254021: The account name or password that you specified is not valid. (type TfsAdminException) Exception Stack Trace:    at Microsoft.TeamFoundation.Management.Controls.WizardCommon.AccountSelectionControl.TestLogon(String connectionString)    at System.ComponentModel.BackgroundWorker.WorkerThreadStart(Object argument) [Info   @16:10:16.307] Benign exception caught as part of verify: Exception Message: TF255329: The following site could not be accessed: http://projects.ssw.com.au/. The server that you specified did not return the expected response. Either you have not installed the Team Foundation Server Extensions for SharePoint Products on this server, or a firewall is blocking access to the specified site or the SharePoint Central Administration site. For more information, see the Microsoft Web site (http://go.microsoft.com/fwlink/?LinkId=161206). (type TeamFoundationServerException) Exception Stack Trace:    at Microsoft.TeamFoundation.Client.SharePoint.WssUtilities.VerifyTeamFoundationSharePointExtensions(ICredentials credentials, Uri url)    at Microsoft.TeamFoundation.Admin.VerifySharePointSitesUrl.Verify() Inner Exception Details: Exception Message: TF249064: The following Web service returned an response that is not valid: http://projects.ssw.com.au/_vti_bin/TeamFoundationIntegrationService.asmx. This Web service is used for the Team Foundation Server Extensions for SharePoint Products. Either the extensions are not installed, the request resulted in HTML being returned, or there is a problem with the URL. Verify that the following URL points to a valid SharePoint Web application and that the application is available: http://projects.ssw.com.au. If the URL is correct and the Web application is operating normally, verify that a firewall is not blocking access to the Web application. (type TeamFoundationServerInvalidResponseException) Exception Data Dictionary: ResponseStatusCode = InternalServerError I’ll look at SharePoint after, probably the SharePoint box just needs a restart or a kick If there is a problem with SharePoint it will come out in testing, But I will definatly be passing this on to Microsoft.   Upgrading the SharePoint connector to TFS 2010 You will need to upgrade the Extensions for SharePoint Products and Technologies on all of your SharePoint farm front end servers. To do this uninstall  the TFS 2010 RC from it in the same way as the server, and then install just the RTM Extensions. Figure: Only install the SharePoint Extensions on your SharePoint front end servers. TFS 2010 supports both SharePoint 2007 and SharePoint 2010.   Figure: When you configure SharePoint it uploads all of the solutions and templates. Figure: Everything is uploaded Successfully. Figure: TFS even remembered the settings from the previous installation, fantastic.   Upgrading the Team Foundation Build Servers to TFS 2010 Just like on the SharePoint servers you will need to upgrade the Build Server to the RTM. Just uninstall TFS 2010 RC and then install only the Team Foundation Build Services component. Unlike on the SharePoint server you will probably have some version of Visual Studio installed. You will need to remove this as well. (Coming Soon) Connecting Visual Studio 2010 / 2008 / 2005 and Eclipse to TFS2010 If you have developers still on Visual Studio 2005 or 2008 you will need do download the respective compatibility pack: Visual Studio Team System 2005 Service Pack 1 Forward Compatibility Update for Team Foundation Server 2010 Visual Studio Team System 2008 Service Pack 1 Forward Compatibility Update for Team Foundation Server 2010 If you are using Eclipse you can download the new Team Explorer Everywhere install for connecting to TFS. Get your developers to check that you have the latest version of your applications with SSW Diagnostic which will check for Service Packs and hot fixes to Visual Studio as well.   Technorati Tags: TFS,TFS2010,TFS 2010,Upgrade

    Read the article

  • Running ASP.NET Webforms and ASP.NET MVC side by side

    - by rajbk
    One of the nice things about ASP.NET MVC and its older brother ASP.NET WebForms is that they are both built on top of the ASP.NET runtime environment. The advantage of this is that, you can still run them side by side even though MVC and WebForms are different frameworks. Another point to note is that with the release of the ASP.NET routing in .NET 3.5 SP1, we are able to create SEO friendly URLs that do not map to specific files on disk. The routing is part of the core runtime environment and therefore can be used by both WebForms and MVC. To run both frameworks side by side, we could easily create a separate folder in your MVC project for all our WebForm files and be good to go. What this post shows you instead, is how to have an MVC application with WebForm pages  that both use a common master page and common routing for SEO friendly URLs.  A sample project that shows WebForms and MVC running side by side is attached at the bottom of this post. So why would we want to run WebForms and MVC in the same project?  WebForms come with a lot of nice server controls that provide a lot of functionality. One example is the ReportViewer control. Using this control and client report definition files (RDLC), we can create rich interactive reports (with charting controls). I show you how to use the ReportViewer control in a WebForm project here :  Creating an ASP.NET report using Visual Studio 2010. We can create even more advanced reports by using SQL reporting services that can also be rendered by the ReportViewer control. Now, consider the sample MVC application I blogged about called ASP.NET MVC Paging/Sorting/Filtering using the MVCContrib Grid and Pager. Assume you were given the requirement to add a UI to the MVC application where users could interact with a report and be given the option to export the report to Excel, PDF or Word. How do you go about doing it?   This is a perfect scenario to use the ReportViewer control and RDLCs. As you saw in the post on creating the ASP.NET report, the ReportViewer control is a Web Control and is designed to be run in a WebForm project with dependencies on, amongst others, a ScriptManager control and the beloved Viewstate.  Since MVC and WebForm both run under the same runtime, the easiest thing to is to add the WebForm application files (index.aspx, rdlc, related class files) into our MVC project. You can copy the files over from the WebForm project into the MVC project. Create a new folder in our MVC application called CommonReports. Add the index.aspx and rdlc file from the Webform project   Right click on the Index.aspx file and convert it to a web application. This will add the index.aspx.designer.cs file (this step is not required if you are manually adding a WebForm aspx file into the MVC project).    Verify that all the type names for the ObjectDataSources in code behind to point to the correct ProductRepository and fix any compiler errors. Right click on Index.aspx and select “View in browser”. You should see a screen like the one below:   There are two issues with our page. It does not use our site master page and the URL is not SEO friendly. Common Master Page The easiest way to use master pages with both MVC and WebForm pages is to have a common master page that each inherits from as shown below. The reason for this is most WebForm controls require them to be inside a Form control and require ControlState or ViewState. ViewMasterPages used in MVC, on the other hand, are designed to be used with content pages that derive from ViewPage with Viewstate turned off. By having a separate master page for MVC and WebForm that inherit from the Root master page,, we can set properties that are specific to each. For example, in the Webform master, we can turn on ViewState, add a form tag etc. Another point worth noting is that if you set a WebForm page to use a MVC site master page, you may run into errors like the following: A ViewMasterPage can be used only with content pages that derive from ViewPage or ViewPage<TViewItem> or Control 'MainContent_MyButton' of type 'Button' must be placed inside a form tag with runat=server. Since the ViewMasterPage inherits from MasterPage as seen below, we make our Root.master inherit from MasterPage, MVC.master inherit from ViewMasterPage and Webform.master inherits from MasterPage. We define the attributes on the master pages like so: Root.master <%@ Master Inherits="System.Web.UI.MasterPage"  … %> MVC.master <%@ Master MasterPageFile="~/Views/Shared/Root.Master" Inherits="System.Web.Mvc.ViewMasterPage" … %> WebForm.master <%@ Master MasterPageFile="~/Views/Shared/Root.Master" Inherits="NorthwindSales.Views.Shared.Webform" %> Code behind: public partial class Webform : System.Web.UI.MasterPage {} We make changes to our reports aspx file to use the Webform.master. See the source of the master pages in the sample project for a better understanding of how they are connected. SEO friendly links We want to create SEO friendly links that point to our report. A request to /Reports/Products should render the report located in ~/CommonReports/Products.aspx. Simillarly to support future reports, a request to /Reports/Sales should render a report in ~/CommonReports/Sales.aspx. Lets start by renaming our index.aspx file to Products.aspx to be consistent with our routing criteria above. As mentioned earlier, since routing is part of the core runtime environment, we ca easily create a custom route for our reports by adding an entry in Global.asax. public static void RegisterRoutes(RouteCollection routes) { routes.IgnoreRoute("{resource}.axd/{*pathInfo}");   //Custom route for reports routes.MapPageRoute( "ReportRoute", // Route name "Reports/{reportname}", // URL "~/CommonReports/{reportname}.aspx" // File );     routes.MapRoute( "Default", // Route name "{controller}/{action}/{id}", // URL with parameters new { controller = "Home", action = "Index", id = UrlParameter.Optional } // Parameter defaults ); } With our custom route in place, a request to Reports/Employees will render the page at ~/CommonReports/Employees.aspx. We make this custom route the first entry since the routing system walks the table from top to bottom, and the first route to match wins. Note that it is highly recommended that you write unit tests for your routes to ensure that the mappings you defined are correct. Common Menu Structure The master page in our original MVC project had a menu structure like so: <ul id="menu"> <li> <%=Html.ActionLink("Home", "Index", "Home") %></li> <li> <%=Html.ActionLink("Products", "Index", "Products") %></li> <li> <%=Html.ActionLink("Help", "Help", "Home") %></li> </ul> We want this menu structure to be common to all pages/views and hence should reside in Root.master. Unfortunately the Html.ActionLink helpers will not work since Root.master inherits from MasterPage which does not have the helper methods available. The quickest way to resolve this issue is to use RouteUrl expressions. Using  RouteUrl expressions, we can programmatically generate URLs that are based on route definitions. By specifying parameter values and a route name if required, we get back a URL string that corresponds to a matching route. We move our menu structure to Root.master and change it to use RouteUrl expressions: <ul id="menu"> <li> <asp:HyperLink ID="hypHome" runat="server" NavigateUrl="<%$RouteUrl:routename=default,controller=home,action=index%>">Home</asp:HyperLink></li> <li> <asp:HyperLink ID="hypProducts" runat="server" NavigateUrl="<%$RouteUrl:routename=default,controller=products,action=index%>">Products</asp:HyperLink></li> <li> <asp:HyperLink ID="hypReport" runat="server" NavigateUrl="<%$RouteUrl:routename=ReportRoute,reportname=products%>">Product Report</asp:HyperLink></li> <li> <asp:HyperLink ID="hypHelp" runat="server" NavigateUrl="<%$RouteUrl:routename=default,controller=home,action=help%>">Help</asp:HyperLink></li> </ul> We are done adding the common navigation to our application. The application now uses a common theme, routing and navigation structure. Conclusion We have seen how to do the following through this post Add a WebForm page from a WebForm project to an existing ASP.NET MVC application Use a common master page for both WebForm and MVC pages Use routing for SEO friendly links Use a common menu structure for both WebForm and MVC. The sample project is attached below. Version: VS 2010 RTM Remember to change your connection string to point to your Northwind database NorthwindSalesMVCWebform.zip

    Read the article

  • Enable Automatic Code First Migrations On SQL Database in Azure Web Sites

    - by Steve Michelotti
    Now that Azure supports .NET Framework 4.5, you can use all the latest and greatest available features. A common scenario is to be able to use Entity Framework Code First Migrations with a SQL Database in Azure. Prior to Code First Migrations, Entity Framework provided database initializers. While convenient for demos and prototypes, database initializers weren’t useful for much beyond that because, if you delete and re-create your entire database when the schema changes, you lose all of your operational data. This is the void that Migrations are meant to fill. For example, if you add a column to your model, Migrations will alter the database to add the column rather than blowing away the entire database and re-creating it from scratch. Azure is becoming increasingly easier to use – especially with features like Azure Web Sites. Being able to use Entity Framework Migrations in Azure makes deployment easier than ever. In this blog post, I’ll walk through enabling Automatic Code First Migrations on Azure. I’ll use the Simple Membership provider for my example. First, we’ll create a new Azure Web site called “migrationstest” including creating a new SQL Database along with it:   Next we’ll go to the web site and download the publish profile:   In the meantime, we’ve created a new MVC 4 website in Visual Studio 2012 using the “Internet Application” template. This template is automatically configured to use the Simple Membership provider. We’ll do our initial Publish to Azure by right-clicking our project and selecting “Publish…”. From the “Publish Web” dialog, we’ll import the publish profile that we downloaded in the previous step:   Once the site is published, we’ll just click the “Register” link from the default site. Since the AccountController is decorated with the [InitializeSimpleMembership] attribute, the initializer will be called and the initial database is created.   We can verify this by connecting to our SQL Database on Azure with SQL Management Studio (after making sure that our local IP address is added to the list of Allowed IP Addresses in Azure): One interesting note is that these tables got created with the default Entity Framework initializer – which is to create the database if it doesn’t already exist. However, our database did already exist! This is because there is a new feature of Entity Framework 5 where Code First will add tables to an existing database as long as the target database doesn’t contain any of the tables from the model. At this point, it’s time to enable Migrations. We’ll open the Package Manger Console and execute the command: PM> Enable-Migrations -EnableAutomaticMigrations This will enable automatic migrations for our project. Because we used the "-EnableAutomaticMigrations” switch, it will create our Configuration class with a constructor that sets the AutomaticMigrationsEnabled property set to true: 1: public Configuration() 2: { 3: AutomaticMigrationsEnabled = true; 4: } We’ll now add our initial migration: PM> Add-Migration Initial This will create a migration class call “Initial” that contains the entire model. But we need to remove all of this code because our database already exists so we are just left with empty Up() and Down() methods. 1: public partial class Initial : DbMigration 2: { 3: public override void Up() 4: { 5: } 6: 7: public override void Down() 8: { 9: } 10: } If we don’t remove this code, we’ll get an exception the first time we attempt to run migrations that tells us: “There is already an object named 'UserProfile' in the database”. This blog post by Julie Lerman fully describes this scenario (i.e., enabling migrations on an existing database). Our next step is to add the Entity Framework initializer that will automatically use Migrations to update the database to the latest version. We will add these 2 lines of code to the Application_Start of the Global.asax: 1: Database.SetInitializer(new MigrateDatabaseToLatestVersion<UsersContext, Configuration>()); 2: new UsersContext().Database.Initialize(false); Note the Initialize() call will force the initializer to run if it has not been run before. At this point, we can publish again to make sure everything is still working as we are expecting. This time we’re going to specify in our publish profile that Code First Migrations should be executed:   Once we have re-published we can once again navigate to the Register page. At this point the database has not been changed but Migrations is now enabled on our SQL Database in Azure. We can now customize our model. Let’s add 2 new properties to the UserProfile class – Email and DateOfBirth: 1: [Table("UserProfile")] 2: public class UserProfile 3: { 4: [Key] 5: [DatabaseGeneratedAttribute(DatabaseGeneratedOption.Identity)] 6: public int UserId { get; set; } 7: public string UserName { get; set; } 8: public string Email { get; set; } 9: public DateTime DateOfBirth { get; set; } 10: } At this point all we need to do is simply re-publish. We’ll once again navigate to the Registration page and, because we had Automatic Migrations enabled, the database has been altered (*not* recreated) to add our 2 new columns. We can verify this by once again looking at SQL Management Studio:   Automatic Migrations provide a quick and easy way to keep your database in sync with your model without the worry of having to re-create your entire database and lose data. With Azure Web Sites you can set up automatic deployment with Git or TFS and automate the entire process to make it dead simple.

    Read the article

  • Improving Partitioned Table Join Performance

    - by Paul White
    The query optimizer does not always choose an optimal strategy when joining partitioned tables. This post looks at an example, showing how a manual rewrite of the query can almost double performance, while reducing the memory grant to almost nothing. Test Data The two tables in this example use a common partitioning partition scheme. The partition function uses 41 equal-size partitions: CREATE PARTITION FUNCTION PFT (integer) AS RANGE RIGHT FOR VALUES ( 125000, 250000, 375000, 500000, 625000, 750000, 875000, 1000000, 1125000, 1250000, 1375000, 1500000, 1625000, 1750000, 1875000, 2000000, 2125000, 2250000, 2375000, 2500000, 2625000, 2750000, 2875000, 3000000, 3125000, 3250000, 3375000, 3500000, 3625000, 3750000, 3875000, 4000000, 4125000, 4250000, 4375000, 4500000, 4625000, 4750000, 4875000, 5000000 ); GO CREATE PARTITION SCHEME PST AS PARTITION PFT ALL TO ([PRIMARY]); There two tables are: CREATE TABLE dbo.T1 ( TID integer NOT NULL IDENTITY(0,1), Column1 integer NOT NULL, Padding binary(100) NOT NULL DEFAULT 0x,   CONSTRAINT PK_T1 PRIMARY KEY CLUSTERED (TID) ON PST (TID) );   CREATE TABLE dbo.T2 ( TID integer NOT NULL, Column1 integer NOT NULL, Padding binary(100) NOT NULL DEFAULT 0x,   CONSTRAINT PK_T2 PRIMARY KEY CLUSTERED (TID, Column1) ON PST (TID) ); The next script loads 5 million rows into T1 with a pseudo-random value between 1 and 5 for Column1. The table is partitioned on the IDENTITY column TID: INSERT dbo.T1 WITH (TABLOCKX) (Column1) SELECT (ABS(CHECKSUM(NEWID())) % 5) + 1 FROM dbo.Numbers AS N WHERE n BETWEEN 1 AND 5000000; In case you don’t already have an auxiliary table of numbers lying around, here’s a script to create one with 10 million rows: CREATE TABLE dbo.Numbers (n bigint PRIMARY KEY);   WITH L0 AS(SELECT 1 AS c UNION ALL SELECT 1), L1 AS(SELECT 1 AS c FROM L0 AS A CROSS JOIN L0 AS B), L2 AS(SELECT 1 AS c FROM L1 AS A CROSS JOIN L1 AS B), L3 AS(SELECT 1 AS c FROM L2 AS A CROSS JOIN L2 AS B), L4 AS(SELECT 1 AS c FROM L3 AS A CROSS JOIN L3 AS B), L5 AS(SELECT 1 AS c FROM L4 AS A CROSS JOIN L4 AS B), Nums AS(SELECT ROW_NUMBER() OVER (ORDER BY (SELECT NULL)) AS n FROM L5) INSERT dbo.Numbers WITH (TABLOCKX) SELECT TOP (10000000) n FROM Nums ORDER BY n OPTION (MAXDOP 1); Table T1 contains data like this: Next we load data into table T2. The relationship between the two tables is that table 2 contains ‘n’ rows for each row in table 1, where ‘n’ is determined by the value in Column1 of table T1. There is nothing particularly special about the data or distribution, by the way. INSERT dbo.T2 WITH (TABLOCKX) (TID, Column1) SELECT T.TID, N.n FROM dbo.T1 AS T JOIN dbo.Numbers AS N ON N.n >= 1 AND N.n <= T.Column1; Table T2 ends up containing about 15 million rows: The primary key for table T2 is a combination of TID and Column1. The data is partitioned according to the value in column TID alone. Partition Distribution The following query shows the number of rows in each partition of table T1: SELECT PartitionID = CA1.P, NumRows = COUNT_BIG(*) FROM dbo.T1 AS T CROSS APPLY (VALUES ($PARTITION.PFT(TID))) AS CA1 (P) GROUP BY CA1.P ORDER BY CA1.P; There are 40 partitions containing 125,000 rows (40 * 125k = 5m rows). The rightmost partition remains empty. The next query shows the distribution for table 2: SELECT PartitionID = CA1.P, NumRows = COUNT_BIG(*) FROM dbo.T2 AS T CROSS APPLY (VALUES ($PARTITION.PFT(TID))) AS CA1 (P) GROUP BY CA1.P ORDER BY CA1.P; There are roughly 375,000 rows in each partition (the rightmost partition is also empty): Ok, that’s the test data done. Test Query and Execution Plan The task is to count the rows resulting from joining tables 1 and 2 on the TID column: SET STATISTICS IO ON; DECLARE @s datetime2 = SYSUTCDATETIME();   SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID;   SELECT DATEDIFF(Millisecond, @s, SYSUTCDATETIME()); SET STATISTICS IO OFF; The optimizer chooses a plan using parallel hash join, and partial aggregation: The Plan Explorer plan tree view shows accurate cardinality estimates and an even distribution of rows across threads (click to enlarge the image): With a warm data cache, the STATISTICS IO output shows that no physical I/O was needed, and all 41 partitions were touched: Running the query without actual execution plan or STATISTICS IO information for maximum performance, the query returns in around 2600ms. Execution Plan Analysis The first step toward improving on the execution plan produced by the query optimizer is to understand how it works, at least in outline. The two parallel Clustered Index Scans use multiple threads to read rows from tables T1 and T2. Parallel scan uses a demand-based scheme where threads are given page(s) to scan from the table as needed. This arrangement has certain important advantages, but does result in an unpredictable distribution of rows amongst threads. The point is that multiple threads cooperate to scan the whole table, but it is impossible to predict which rows end up on which threads. For correct results from the parallel hash join, the execution plan has to ensure that rows from T1 and T2 that might join are processed on the same thread. For example, if a row from T1 with join key value ‘1234’ is placed in thread 5’s hash table, the execution plan must guarantee that any rows from T2 that also have join key value ‘1234’ probe thread 5’s hash table for matches. The way this guarantee is enforced in this parallel hash join plan is by repartitioning rows to threads after each parallel scan. The two repartitioning exchanges route rows to threads using a hash function over the hash join keys. The two repartitioning exchanges use the same hash function so rows from T1 and T2 with the same join key must end up on the same hash join thread. Expensive Exchanges This business of repartitioning rows between threads can be very expensive, especially if a large number of rows is involved. The execution plan selected by the optimizer moves 5 million rows through one repartitioning exchange and around 15 million across the other. As a first step toward removing these exchanges, consider the execution plan selected by the optimizer if we join just one partition from each table, disallowing parallelism: SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = 1 AND $PARTITION.PFT(T2.TID) = 1 OPTION (MAXDOP 1); The optimizer has chosen a (one-to-many) merge join instead of a hash join. The single-partition query completes in around 100ms. If everything scaled linearly, we would expect that extending this strategy to all 40 populated partitions would result in an execution time around 4000ms. Using parallelism could reduce that further, perhaps to be competitive with the parallel hash join chosen by the optimizer. This raises a question. If the most efficient way to join one partition from each of the tables is to use a merge join, why does the optimizer not choose a merge join for the full query? Forcing a Merge Join Let’s force the optimizer to use a merge join on the test query using a hint: SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID OPTION (MERGE JOIN); This is the execution plan selected by the optimizer: This plan results in the same number of logical reads reported previously, but instead of 2600ms the query takes 5000ms. The natural explanation for this drop in performance is that the merge join plan is only using a single thread, whereas the parallel hash join plan could use multiple threads. Parallel Merge Join We can get a parallel merge join plan using the same query hint as before, and adding trace flag 8649: SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID OPTION (MERGE JOIN, QUERYTRACEON 8649); The execution plan is: This looks promising. It uses a similar strategy to distribute work across threads as seen for the parallel hash join. In practice though, performance is disappointing. On a typical run, the parallel merge plan runs for around 8400ms; slower than the single-threaded merge join plan (5000ms) and much worse than the 2600ms for the parallel hash join. We seem to be going backwards! The logical reads for the parallel merge are still exactly the same as before, with no physical IOs. The cardinality estimates and thread distribution are also still very good (click to enlarge): A big clue to the reason for the poor performance is shown in the wait statistics (captured by Plan Explorer Pro): CXPACKET waits require careful interpretation, and are most often benign, but in this case excessive waiting occurs at the repartitioning exchanges. Unlike the parallel hash join, the repartitioning exchanges in this plan are order-preserving ‘merging’ exchanges (because merge join requires ordered inputs): Parallelism works best when threads can just grab any available unit of work and get on with processing it. Preserving order introduces inter-thread dependencies that can easily lead to significant waits occurring. In extreme cases, these dependencies can result in an intra-query deadlock, though the details of that will have to wait for another time to explore in detail. The potential for waits and deadlocks leads the query optimizer to cost parallel merge join relatively highly, especially as the degree of parallelism (DOP) increases. This high costing resulted in the optimizer choosing a serial merge join rather than parallel in this case. The test results certainly confirm its reasoning. Collocated Joins In SQL Server 2008 and later, the optimizer has another available strategy when joining tables that share a common partition scheme. This strategy is a collocated join, also known as as a per-partition join. It can be applied in both serial and parallel execution plans, though it is limited to 2-way joins in the current optimizer. Whether the optimizer chooses a collocated join or not depends on cost estimation. The primary benefits of a collocated join are that it eliminates an exchange and requires less memory, as we will see next. Costing and Plan Selection The query optimizer did consider a collocated join for our original query, but it was rejected on cost grounds. The parallel hash join with repartitioning exchanges appeared to be a cheaper option. There is no query hint to force a collocated join, so we have to mess with the costing framework to produce one for our test query. Pretending that IOs cost 50 times more than usual is enough to convince the optimizer to use collocated join with our test query: -- Pretend IOs are 50x cost temporarily DBCC SETIOWEIGHT(50);   -- Co-located hash join SELECT COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID OPTION (RECOMPILE);   -- Reset IO costing DBCC SETIOWEIGHT(1); Collocated Join Plan The estimated execution plan for the collocated join is: The Constant Scan contains one row for each partition of the shared partitioning scheme, from 1 to 41. The hash repartitioning exchanges seen previously are replaced by a single Distribute Streams exchange using Demand partitioning. Demand partitioning means that the next partition id is given to the next parallel thread that asks for one. My test machine has eight logical processors, and all are available for SQL Server to use. As a result, there are eight threads in the single parallel branch in this plan, each processing one partition from each table at a time. Once a thread finishes processing a partition, it grabs a new partition number from the Distribute Streams exchange…and so on until all partitions have been processed. It is important to understand that the parallel scans in this plan are different from the parallel hash join plan. Although the scans have the same parallelism icon, tables T1 and T2 are not being co-operatively scanned by multiple threads in the same way. Each thread reads a single partition of T1 and performs a hash match join with the same partition from table T2. The properties of the two Clustered Index Scans show a Seek Predicate (unusual for a scan!) limiting the rows to a single partition: The crucial point is that the join between T1 and T2 is on TID, and TID is the partitioning column for both tables. A thread that processes partition ‘n’ is guaranteed to see all rows that can possibly join on TID for that partition. In addition, no other thread will see rows from that partition, so this removes the need for repartitioning exchanges. CPU and Memory Efficiency Improvements The collocated join has removed two expensive repartitioning exchanges and added a single exchange processing 41 rows (one for each partition id). Remember, the parallel hash join plan exchanges had to process 5 million and 15 million rows. The amount of processor time spent on exchanges will be much lower in the collocated join plan. In addition, the collocated join plan has a maximum of 8 threads processing single partitions at any one time. The 41 partitions will all be processed eventually, but a new partition is not started until a thread asks for it. Threads can reuse hash table memory for the new partition. The parallel hash join plan also had 8 hash tables, but with all 5,000,000 build rows loaded at the same time. The collocated plan needs memory for only 8 * 125,000 = 1,000,000 rows at any one time. Collocated Hash Join Performance The collated join plan has disappointing performance in this case. The query runs for around 25,300ms despite the same IO statistics as usual. This is much the worst result so far, so what went wrong? It turns out that cardinality estimation for the single partition scans of table T1 is slightly low. The properties of the Clustered Index Scan of T1 (graphic immediately above) show the estimation was for 121,951 rows. This is a small shortfall compared with the 125,000 rows actually encountered, but it was enough to cause the hash join to spill to physical tempdb: A level 1 spill doesn’t sound too bad, until you realize that the spill to tempdb probably occurs for each of the 41 partitions. As a side note, the cardinality estimation error is a little surprising because the system tables accurately show there are 125,000 rows in every partition of T1. Unfortunately, the optimizer uses regular column and index statistics to derive cardinality estimates here rather than system table information (e.g. sys.partitions). Collocated Merge Join We will never know how well the collocated parallel hash join plan might have worked without the cardinality estimation error (and the resulting 41 spills to tempdb) but we do know: Merge join does not require a memory grant; and Merge join was the optimizer’s preferred join option for a single partition join Putting this all together, what we would really like to see is the same collocated join strategy, but using merge join instead of hash join. Unfortunately, the current query optimizer cannot produce a collocated merge join; it only knows how to do collocated hash join. So where does this leave us? CROSS APPLY sys.partitions We can try to write our own collocated join query. We can use sys.partitions to find the partition numbers, and CROSS APPLY to get a count per partition, with a final step to sum the partial counts. The following query implements this idea: SELECT row_count = SUM(Subtotals.cnt) FROM ( -- Partition numbers SELECT p.partition_number FROM sys.partitions AS p WHERE p.[object_id] = OBJECT_ID(N'T1', N'U') AND p.index_id = 1 ) AS P CROSS APPLY ( -- Count per collocated join SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals; The estimated plan is: The cardinality estimates aren’t all that good here, especially the estimate for the scan of the system table underlying the sys.partitions view. Nevertheless, the plan shape is heading toward where we would like to be. Each partition number from the system table results in a per-partition scan of T1 and T2, a one-to-many Merge Join, and a Stream Aggregate to compute the partial counts. The final Stream Aggregate just sums the partial counts. Execution time for this query is around 3,500ms, with the same IO statistics as always. This compares favourably with 5,000ms for the serial plan produced by the optimizer with the OPTION (MERGE JOIN) hint. This is another case of the sum of the parts being less than the whole – summing 41 partial counts from 41 single-partition merge joins is faster than a single merge join and count over all partitions. Even so, this single-threaded collocated merge join is not as quick as the original parallel hash join plan, which executed in 2,600ms. On the positive side, our collocated merge join uses only one logical processor and requires no memory grant. The parallel hash join plan used 16 threads and reserved 569 MB of memory:   Using a Temporary Table Our collocated merge join plan should benefit from parallelism. The reason parallelism is not being used is that the query references a system table. We can work around that by writing the partition numbers to a temporary table (or table variable): SET STATISTICS IO ON; DECLARE @s datetime2 = SYSUTCDATETIME();   CREATE TABLE #P ( partition_number integer PRIMARY KEY);   INSERT #P (partition_number) SELECT p.partition_number FROM sys.partitions AS p WHERE p.[object_id] = OBJECT_ID(N'T1', N'U') AND p.index_id = 1;   SELECT row_count = SUM(Subtotals.cnt) FROM #P AS p CROSS APPLY ( SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals;   DROP TABLE #P;   SELECT DATEDIFF(Millisecond, @s, SYSUTCDATETIME()); SET STATISTICS IO OFF; Using the temporary table adds a few logical reads, but the overall execution time is still around 3500ms, indistinguishable from the same query without the temporary table. The problem is that the query optimizer still doesn’t choose a parallel plan for this query, though the removal of the system table reference means that it could if it chose to: In fact the optimizer did enter the parallel plan phase of query optimization (running search 1 for a second time): Unfortunately, the parallel plan found seemed to be more expensive than the serial plan. This is a crazy result, caused by the optimizer’s cost model not reducing operator CPU costs on the inner side of a nested loops join. Don’t get me started on that, we’ll be here all night. In this plan, everything expensive happens on the inner side of a nested loops join. Without a CPU cost reduction to compensate for the added cost of exchange operators, candidate parallel plans always look more expensive to the optimizer than the equivalent serial plan. Parallel Collocated Merge Join We can produce the desired parallel plan using trace flag 8649 again: SELECT row_count = SUM(Subtotals.cnt) FROM #P AS p CROSS APPLY ( SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals OPTION (QUERYTRACEON 8649); The actual execution plan is: One difference between this plan and the collocated hash join plan is that a Repartition Streams exchange operator is used instead of Distribute Streams. The effect is similar, though not quite identical. The Repartition uses round-robin partitioning, meaning the next partition id is pushed to the next thread in sequence. The Distribute Streams exchange seen earlier used Demand partitioning, meaning the next partition id is pulled across the exchange by the next thread that is ready for more work. There are subtle performance implications for each partitioning option, but going into that would again take us too far off the main point of this post. Performance The important thing is the performance of this parallel collocated merge join – just 1350ms on a typical run. The list below shows all the alternatives from this post (all timings include creation, population, and deletion of the temporary table where appropriate) from quickest to slowest: Collocated parallel merge join: 1350ms Parallel hash join: 2600ms Collocated serial merge join: 3500ms Serial merge join: 5000ms Parallel merge join: 8400ms Collated parallel hash join: 25,300ms (hash spill per partition) The parallel collocated merge join requires no memory grant (aside from a paltry 1.2MB used for exchange buffers). This plan uses 16 threads at DOP 8; but 8 of those are (rather pointlessly) allocated to the parallel scan of the temporary table. These are minor concerns, but it turns out there is a way to address them if it bothers you. Parallel Collocated Merge Join with Demand Partitioning This final tweak replaces the temporary table with a hard-coded list of partition ids (dynamic SQL could be used to generate this query from sys.partitions): SELECT row_count = SUM(Subtotals.cnt) FROM ( VALUES (1),(2),(3),(4),(5),(6),(7),(8),(9),(10), (11),(12),(13),(14),(15),(16),(17),(18),(19),(20), (21),(22),(23),(24),(25),(26),(27),(28),(29),(30), (31),(32),(33),(34),(35),(36),(37),(38),(39),(40),(41) ) AS P (partition_number) CROSS APPLY ( SELECT cnt = COUNT_BIG(*) FROM dbo.T1 AS T1 JOIN dbo.T2 AS T2 ON T2.TID = T1.TID WHERE $PARTITION.PFT(T1.TID) = p.partition_number AND $PARTITION.PFT(T2.TID) = p.partition_number ) AS SubTotals OPTION (QUERYTRACEON 8649); The actual execution plan is: The parallel collocated hash join plan is reproduced below for comparison: The manual rewrite has another advantage that has not been mentioned so far: the partial counts (per partition) can be computed earlier than the partial counts (per thread) in the optimizer’s collocated join plan. The earlier aggregation is performed by the extra Stream Aggregate under the nested loops join. The performance of the parallel collocated merge join is unchanged at around 1350ms. Final Words It is a shame that the current query optimizer does not consider a collocated merge join (Connect item closed as Won’t Fix). The example used in this post showed an improvement in execution time from 2600ms to 1350ms using a modestly-sized data set and limited parallelism. In addition, the memory requirement for the query was almost completely eliminated  – down from 569MB to 1.2MB. The problem with the parallel hash join selected by the optimizer is that it attempts to process the full data set all at once (albeit using eight threads). It requires a large memory grant to hold all 5 million rows from table T1 across the eight hash tables, and does not take advantage of the divide-and-conquer opportunity offered by the common partitioning. The great thing about the collocated join strategies is that each parallel thread works on a single partition from both tables, reading rows, performing the join, and computing a per-partition subtotal, before moving on to a new partition. From a thread’s point of view… If you have trouble visualizing what is happening from just looking at the parallel collocated merge join execution plan, let’s look at it again, but from the point of view of just one thread operating between the two Parallelism (exchange) operators. Our thread picks up a single partition id from the Distribute Streams exchange, and starts a merge join using ordered rows from partition 1 of table T1 and partition 1 of table T2. By definition, this is all happening on a single thread. As rows join, they are added to a (per-partition) count in the Stream Aggregate immediately above the Merge Join. Eventually, either T1 (partition 1) or T2 (partition 1) runs out of rows and the merge join stops. The per-partition count from the aggregate passes on through the Nested Loops join to another Stream Aggregate, which is maintaining a per-thread subtotal. Our same thread now picks up a new partition id from the exchange (say it gets id 9 this time). The count in the per-partition aggregate is reset to zero, and the processing of partition 9 of both tables proceeds just as it did for partition 1, and on the same thread. Each thread picks up a single partition id and processes all the data for that partition, completely independently from other threads working on other partitions. One thread might eventually process partitions (1, 9, 17, 25, 33, 41) while another is concurrently processing partitions (2, 10, 18, 26, 34) and so on for the other six threads at DOP 8. The point is that all 8 threads can execute independently and concurrently, continuing to process new partitions until the wider job (of which the thread has no knowledge!) is done. This divide-and-conquer technique can be much more efficient than simply splitting the entire workload across eight threads all at once. Related Reading Understanding and Using Parallelism in SQL Server Parallel Execution Plans Suck © 2013 Paul White – All Rights Reserved Twitter: @SQL_Kiwi

    Read the article

< Previous Page | 84 85 86 87 88 89 90 91 92 93 94 95  | Next Page >