Search Results

Search found 545 results on 22 pages for 'overlap'.

Page 20/22 | < Previous Page | 16 17 18 19 20 21 22  | Next Page >

  • Subterranean IL: Compiling C# exception handlers

    - by Simon Cooper
    An exception handler in C# combines the IL catch and finally exception handling clauses into a single try statement: try { Console.WriteLine("Try block") // ... } catch (IOException) { Console.WriteLine("IOException catch") // ... } catch (Exception e) { Console.WriteLine("Exception catch") // ... } finally { Console.WriteLine("Finally block") // ... } How does this get compiled into IL? Initial implementation If you remember from my earlier post, finally clauses must be specified with their own .try clause. So, for the initial implementation, we take the try/catch/finally, and simply split it up into two .try clauses (I have to use label syntax for this): StartTry: ldstr "Try block" call void [mscorlib]System.Console::WriteLine(string) // ... leave.s End EndTry: StartIOECatch: ldstr "IOException catch" call void [mscorlib]System.Console::WriteLine(string) // ... leave.s End EndIOECatch: StartECatch: ldstr "Exception catch" call void [mscorlib]System.Console::WriteLine(string) // ... leave.s End EndECatch: StartFinally: ldstr "Finally block" call void [mscorlib]System.Console::WriteLine(string) // ... endfinally EndFinally: End: // ... .try StartTry to EndTry catch [mscorlib]System.IO.IOException handler StartIOECatch to EndIOECatch catch [mscorlib]System.Exception handler StartECatch to EndECatch .try StartTry to EndTry finally handler StartFinally to EndFinally However, the resulting program isn't verifiable, and doesn't run: [IL]: Error: Shared try has finally or fault handler. Nested try blocks What's with the verification error? Well, it's a condition of IL verification that all exception handling regions (try, catch, filter, finally, fault) of a single .try clause have to be completely contained within any outer exception region, and they can't overlap with any other exception handling clause. In other words, IL exception handling clauses must to be representable in the scoped syntax, and in this example, we're overlapping catch and finally clauses. Not only is this example not verifiable, it isn't semantically correct. The finally handler is specified round the .try. What happens if you were able to run this code, and an exception was thrown? Program execution enters top of try block, and exception is thrown within it CLR searches for an exception handler, finds catch Because control flow is leaving .try, finally block is run The catch block is run leave.s End inside the catch handler branches to End label. We're actually running the finally before the catch! What we do about it What we actually need to do is put the catch clauses inside the finally clause, as this will ensure the finally gets executed at the correct time (this time using scoped syntax): .try { .try { ldstr "Try block" call void [mscorlib]System.Console::WriteLine(string) // ... leave.s End } catch [mscorlib]System.IO.IOException { ldstr "IOException catch" call void [mscorlib]System.Console::WriteLine(string) // ... leave.s End } catch [mscorlib]System.Exception { ldstr "Exception catch" call void [mscorlib]System.Console::WriteLine(string) // ... leave.s End } } finally { ldstr "Finally block" call void [mscorlib]System.Console::WriteLine(string) // ... endfinally } End: ret Returning from methods There is a further semantic mismatch that the C# compiler has to deal with; in C#, you are allowed to return from within an exception handling block: public int HandleMethod() { try { // ... return 0; } catch (Exception) { // ... return -1; } } However, you can't ret inside an exception handling block in IL. So the C# compiler does a leave.s to a ret outside the exception handling area, loading/storing any return value to a local variable along the way (as leave.s clears the stack): .method public instance int32 HandleMethod() { .locals init ( int32 retVal ) .try { // ... ldc.i4.0 stloc.0 leave.s End } catch [mscorlib]System.Exception { // ... ldc.i4.m1 stloc.0 leave.s End } End: ldloc.0 ret } Conclusion As you can see, the C# compiler has quite a few hoops to jump through to translate C# code into semantically-correct IL, and hides the numerous conditions on IL exception handling blocks from the C# programmer. Next up: catch-all blocks, and how the runtime deals with non-Exception exceptions.

    Read the article

  • Collision 2D Quads

    - by Vico Pelaez
    I want to detect collision between two 2D squares, one square is static and the other one moves according to keyboard arrows. I have implemented some code, however nothing happens when they overlap each other and what I tried to achieve in the code was to detect an overlapping between them. I think I am either not understanding the concept really well or that because one of the squares is moving this is not working. Please I would really appreciate your help. Thank you! float x1=0.05 ,Y1=0.05; float x2=0.05 ,Y2=0.05; float posX1 =0.5, posY1 = 0.5; float movX2 = 0.0 , movY2 = 0.0; struct box{ int width=0.1; int heigth=0.1; }; void init(){ glClearColor(0.0, 0.0, 0.0, 0.0); glColor3f(1.0, 1.0, 1.0); } void quad1(){ glTranslatef(posX1, posY1, 0.0); glBegin(GL_POLYGON); glColor3f(0.5, 1.0, 0.5); glVertex2f(-x1, -Y1); glVertex2f(-x1, Y1); glVertex2f(x1,Y1); glVertex2f(x1,-Y1); glEnd(); } void quad2(){ glMatrixMode(GL_PROJECTION); glLoadIdentity(); glPushMatrix(); glTranslatef(movX2, movY2, 0.0); glBegin(GL_POLYGON); glColor3f(1.5, 1.0, 0.5); glVertex2f(-x2, -Y2); glVertex2f(-x2, Y2); glVertex2f(x2,Y2); glVertex2f(x2,-Y2); glEnd(); glPopMatrix(); } void reset(){ //Reset position of square??? movX2 = 0.0; movY2 = 0.0; collisionB = false; } bool collision(box A, box B){ int leftA, leftB; int rightA, rightB; int topA, topB; int bottomA, bottomB; //Calculate the sides of box A leftA = x1; rightA = x1 + A.width; topA = Y1; bottomA = Y1 + A.heigth; //Calculate the sides of box B leftB = x2; rightB = x2 + B.width; topB = Y1; bottomB = Y1+ B.heigth ; if( bottomA <= topB ) return false; if( topA >= bottomB ) return false; if( rightA <= leftB ) return false; if( leftA >= rightB ) return false; return true; } float move_unit = 0.1; void keyboardown(int key, int x, int y) { switch (key){ case GLUT_KEY_UP: movY2 += move_unit; break; case GLUT_KEY_RIGHT: movX2 += move_unit; break; case GLUT_KEY_LEFT: movX2 -= move_unit; break; case GLUT_KEY_DOWN: movY2 -= move_unit; break; default: break; } glutPostRedisplay(); } void display(){ glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT); glMatrixMode(GL_PROJECTION); glLoadIdentity(); cuad1(); if (!collision) { cuad2(); } else{ reset(); } glFlush(); } int main(int argc, char** argv){ glutInit(&argc, argv); glutInitDisplayMode(GLUT_SINGLE | GLUT_RGB); glutInitWindowSize(500,500); glutInitWindowPosition(0, 0); glutCreateWindow("Collision Practice"); glutSpecialFunc(keyboardown); glutDisplayFunc(display); init(); glutMainLoop(); }

    Read the article

  • Is this over-abstraction? (And is there a name for it?)

    - by mwhite
    I work on a large Django application that uses CouchDB as a database and couchdbkit for mapping CouchDB documents to objects in Python, similar to Django's default ORM. It has dozens of model classes and a hundred or two CouchDB views. The application allows users to register a "domain", which gives them a unique URL containing the domain name that gives them access to a project whose data has no overlap with the data of other domains. Each document that is part of a domain has its domain property set to that domain's name. As far as relationships between the documents go, all domains are effectively mutually exclusive subsets of the data, except for a few edge cases (some users can be members of more than one domain, and there are some administrative reports that include all domains, etc.). The code is full of explicit references to the domain name, and I'm wondering if it would be worth the added complexity to abstract this out. I'd also like to know if there's a name for the sort of bound property approach I'm taking here. Basically, I have something like this in mind: Before in models.py class User(Document): domain = StringProperty() class Group(Document): domain = StringProperty() name = StringProperty() user_ids = StringListProperty() # method that returns related document set def users(self): return [User.get(id) for id in self.user_ids] # method that queries a couch view optimized for a specific lookup @classmethod def by_name(cls, domain, name): # the view method is provided by couchdbkit and handles # wrapping json CouchDB results as Python objects, and # can take various parameters modifying behavior return cls.view('groups/by_name', key=[domain, name]) # method that creates a related document def get_new_user(self): user = User(domain=self.domain) user.save() self.user_ids.append(user._id) return user in views.py: from models import User, Group # there are tons of views like this, (request, domain, ...) def create_new_user_in_group(request, domain, group_name): group = Group.by_name(domain, group_name)[0] user = User(domain=domain) user.save() group.user_ids.append(user._id) group.save() in group/by_name/map.js: function (doc) { if (doc.doc_type == "Group") { emit([doc.domain, doc.name], null); } } After models.py class DomainDocument(Document): domain = StringProperty() @classmethod def domain_view(cls, *args, **kwargs): kwargs['key'] = [cls.domain.default] + kwargs['key'] return super(DomainDocument, cls).view(*args, **kwargs) @classmethod def get(cls, *args, **kwargs, validate_domain=True): ret = super(DomainDocument, cls).get(*args, **kwargs) if validate_domain and ret.domain != cls.domain.default: raise Exception() return ret def models(self): # a mapping of all models in the application. accessing one returns the equivalent of class BoundUser(User): domain = StringProperty(default=self.domain) class User(DomainDocument): pass class Group(DomainDocument): name = StringProperty() user_ids = StringListProperty() def users(self): return [self.models.User.get(id) for id in self.user_ids] @classmethod def by_name(cls, name): return cls.domain_view('groups/by_name', key=[name]) def get_new_user(self): user = self.models.User() user.save() views.py @domain_view # decorator that sets request.models to the same sort of object that is returned by DomainDocument.models and removes the domain argument from the URL router def create_new_user_in_group(request, group_name): group = request.models.Group.by_name(group_name) user = request.models.User() user.save() group.user_ids.append(user._id) group.save() (Might be better to leave the abstraction leaky here in order to avoid having to deal with a couchapp-style //! include of a wrapper for emit that prepends doc.domain to the key or some other similar solution.) function (doc) { if (doc.doc_type == "Group") { emit([doc.name], null); } } Pros and Cons So what are the pros and cons of this? Pros: DRYer prevents you from creating related documents but forgetting to set the domain. prevents you from accidentally writing a django view - couch view execution path that leads to a security breach doesn't prevent you from accessing underlying self.domain and normal Document.view() method potentially gets rid of the need for a lot of sanity checks verifying whether two documents whose domains we expect to be equal are. Cons: adds some complexity hides what's really happening requires no model modules to have classes with the same name, or you would need to add sub-attributes to self.models for modules. However, requiring project-wide unique class names for models should actually be fine because they correspond to the doc_type property couchdbkit uses to decide which class to instantiate them as, which should be unique. removes explicit dependency documentation (from group.models import Group)

    Read the article

  • I need help with 2D collision response (of stacking rotating polygons, with friction and gravity, for a game)

    - by Register Sole
    Hi I am looking for suggestions on how to write a collision response for game programming purpose (so not a scientific simulation). I am dealing with 2D polygons that are rotating, and I want them to be able to stack. I also want friction and gravity. I have a detection mechanism that returns the separating axis, how long the polygons are overlapping, and up to 2 points of contact. For the response, I am currently using an impulse-based response, which main idea is: find the separating axis, length of overlap, and the point of contact (if there are two, pick a random point between to simulate averaged force. i believe there are better ways than this) separate the object (modifying their positions, taking into account of their masses. i do not separate them completely though, to keep track that they are colliding to reduce jitter) calculate normal force based on the coefficient of restitution as if there is no friction. calculate friction, as if there is no normal force. I also assume that the direction of the friction is the same throughout the collision. apply the two forces (which result in a rather inaccurate result, since each force is calculated as if the other is not present. for non-rotating bodies though, this method is exact) I am aware that this method requires the coefficient of friction to be sufficiently small due to the assumption that the direction of friction stays the same in a collision. Also, the result is visually satisfying if gravity is not present. However, when there is gravity, objects on ground jitter and drift (even with zero coefficient of restitution)! It also happens for stacking objects. Larger coefficient of restitution and gravity increase the jittering. I hope you can help me with this. Some things i would like to know more about is how to handle collision with two point of contacts (how to end up having an object sitting still on the ground?), how to reduce, and prevent if possible, jitter and drift (do people use the most accurate method possible, or is there a trick to overcome this?), and how to handle multiple objects collision (for example, in the case of stacking objects, how do I check collisions between all of them and keep them all stable at every frame so they don't jitter?). A total reformulation of my algorithm is also welcomed, as long as it works. Another thing to note is that I am not making a Physics game, so I only need a visually satisfying response (though a realistic response is preferable, if it is not performance-heavy). But surely jittering and drifting objects on flat ground are not at all acceptable. In addition, I am a Physics student, so feel free to talk about impulse and whatever needed. Finally, I'm sorry for the long post. I tried to be as concise as I can. Thank you for reading it! EDIT It seems what I didn't manage to come up all this time is to separate resting contact as a class of its own and how to solve them. Currently reading the paper suggested by Jedediah. More suggestions on the topic are welcome :) CASE CLOSED After reading various papers referenced in the paper, successfully implemented simultaneous impulse method (referring to the original paper by Erin Catto, [Catt05]). Thanks maaaan!! The paper is wonderful. The current system is visibly much better than the previous. Still haven't separated resting contact as a class of its own though, which brings me to my next question. Love you all! Haha (sorry, I'm just so happy thanks to you).

    Read the article

  • How to calculate CPU % based on raw CPU ticks in SNMP

    - by bjeanes
    According to http://net-snmp.sourceforge.net/docs/mibs/ucdavis.html#scalar_notcurrent ssCpuUser, ssCpuSystem, ssCpuIdle, etc are deprecated in favor of the raw variants (ssCpuRawUser, etc). The former values (which don't cover things like nice, wait, kernel, interrupt, etc) returned a percentage value: The percentage of CPU time spent processing user-level code, calculated over the last minute. This object has been deprecated in favour of 'ssCpuRawUser(50)', which can be used to calculate the same metric, but over any desired time period. The raw values return the "raw" number of ticks the CPU spent: The number of 'ticks' (typically 1/100s) spent processing user-level code. On a multi-processor system, the 'ssCpuRaw*' counters are cumulative over all CPUs, so their sum will typically be N*100 (for N processors). My question is: how do you turn the number of ticks into percentage? That is, how do you know how many ticks per second (it's typically — which implies not always — 1/100s, which either means 1 every 100 seconds or that a tick represents 1/100th of a second). I imagine you also need to know how many CPUs there are or you need to fetch all the CPU values to add them all together. I can't seem to find a MIB that gives you an integer value for # of CPUs which makes the former route awkward. The latter route seems unreliable because some of the numbers overlap (sometimes). For example, ssCpuRawWait has the following warning: This object will not be implemented on hosts where the underlying operating system does not measure this particular CPU metric. This time may also be included within the 'ssCpuRawSystem(52)' counter. Some help would be appreciated. Everywhere seems to just say that % is deprecated because it can be derived, but I haven't found anywhere that shows the official standard way to perform this derivation. The second component is that these "ticks" seem to be cumulative instead of over some time period. How do I sample values over some time period? The ultimate information I want is: % of user, system, idle, nice (and ideally steal, though there doesn't seem to be a standard MIB for this) "currently" (over the last 1-60s would probably be sufficient, with a preference for smaller time spans).

    Read the article

  • Apache VirtualHost Blockhole (Eats All Requests on All Ports on an IP)

    - by Synetech inc.
    I’m exhausted. I just spent the last two hours chasing a goose that I have been after on-and-off for the past year. Here is the goal, put as succinctly as possible. Step 1: HOSTS File: 127.0.0.5 NastyAdServer.com 127.0.0.5 xssServer.com 127.0.0.5 SQLInjector.com 127.0.0.5 PornAds.com 127.0.0.5 OtherBadSites.com … Step 2: Apache httpd.conf <VirtualHost 127.0.0.5:80> ServerName adkiller DocumentRoot adkiller RewriteEngine On RewriteRule (\.(gif|jpg|png|jpeg)$) /p.png [L] RewriteRule (.*) /ad.htm [L] </VirtualHost> So basically what happens is that the HOSTS file redirects designated domains to the localhost, but to a specific loopback IP address. Apache listens for any requests on this address and serves either a transparent pixel graphic, or else an empty HTML file. Thus, any page or graphic on any of the bad sites is replaced with nothing (in other words an ad/malware/porn/etc. blocker). This works great as is (and has been for me for years now). The problem is that these bad things are no longer limited to just HTTP traffic. For example: <script src="http://NastyAdServer.com:99"> or <iframe src="https://PornAds.com/ad.html"> or a Trojan using ftp://spammaster.com/[email protected];[email protected];[email protected] or an app “phoning home” with private info in a crafted ICMP packet by pinging CardStealer.ru:99 Handling HTTPS is a relatively minor bump. I can create a separate VirtualHost just like the one above, replacing port 80 with 443, and adding in SSL directives. This leaves the other ports to be dealt with. I tried using * for the port, but then I get overlap errors. I tried redirecting all request to the HTTPS server and visa-versa but neither worked; either the SSL requests wouldn’t redirect correctly or else the HTTP requests gave the You’re speaking plain HTTP to an SSL-enabled server port… error. Further, I cannot figure out a way to test if other ports are being successfully redirected (I could try using a browser, but what about FTP, ICMP, etc.?) I realize that I could just use a port-blocker (eg ProtoWall, PeerBlock, etc.), but there’s two issues with that. First, I am blocking domains with this method, not IP addresses, so to use a port-blocker, I would have to get each and every domain’s IP, and update theme frequently. Second, using this method, I can have Apache keep logs of all the ad/malware/spam/etc. requests for future analysis (my current AdKiller logs are already 466MB right now). I appreciate any help in successfully setting up an Apache VirtualHost blackhole. Thanks.

    Read the article

  • Control layout using graphviz twopi

    - by vy32
    I am trying to draw a graph showing search prefixes using twopi. I have a simple input file and am getting this output: (full image) Here is the input file: digraph search { // ordering=out; // color=blue; // rank=same; // overlap=scale; rankdir=LR; root=root; ranksep=1.25; overlap=true; "root"; a [color=none,fontsize=12]; b [color=none,fontsize=12]; c [color=none,fontsize=12]; d [color=none,fontsize=12]; e [color=none,fontsize=12]; f [color=none,fontsize=12]; #g [color=none,fontsize=12]; h [color=none,fontsize=12]; i [color=none,fontsize=12]; j [color=none,fontsize=12]; k [color=none,fontsize=12]; l [color=none,fontsize=12]; m [color=none,fontsize=12]; n [color=none,fontsize=12]; o [color=none,fontsize=12]; p [color=none,fontsize=12]; q [color=none,fontsize=12]; r [color=none,fontsize=12]; s [color=none,fontsize=12]; t [color=none,fontsize=12]; u [color=none,fontsize=12]; v [color=none,fontsize=12]; w [color=none,fontsize=12]; x [color=none,fontsize=12]; y [color=none,fontsize=12]; #ga [color=none,fontsize=12]; gb [color=none,fontsize=12]; gc [color=none,fontsize=12]; gd [color=none,fontsize=12]; ge [color=none,fontsize=12]; gf [color=none,fontsize=12]; gg [color=none,fontsize=12]; gh [color=none,fontsize=12]; gi [color=none,fontsize=12]; gj [color=none,fontsize=12]; gk [color=none,fontsize=12]; gl [color=none,fontsize=12]; gm [color=none,fontsize=12]; gn [color=none,fontsize=12]; go [color=none,fontsize=12]; gp [color=none,fontsize=12]; gq [color=none,fontsize=12]; gr [color=none,fontsize=12]; gs [color=none,fontsize=12]; gt [color=none,fontsize=12]; gu [color=none,fontsize=12]; gv [color=none,fontsize=12]; gw [color=none,fontsize=12]; gx [color=none,fontsize=12]; gy [color=none,fontsize=12]; gaa [color=none,fontsize=12]; gab [color=none,fontsize=12]; gac [color=none,fontsize=12]; gad [color=none,fontsize=12]; gae [color=none,fontsize=12]; gaf [color=none,fontsize=12]; gag [color=none,fontsize=12]; gah [color=none,fontsize=12]; gai [color=none,fontsize=12]; gaj [color=none,fontsize=12]; gak [color=none,fontsize=12]; gal [color=none,fontsize=12]; gam [color=none,fontsize=12]; gan [color=none,fontsize=12]; gao [color=none,fontsize=12]; gap [color=none,fontsize=12]; gaq [color=none,fontsize=12]; #gaz [color=none,fontsize=12]; gas [color=none,fontsize=12]; gat [color=none,fontsize=12]; gau [color=none,fontsize=12]; gav [color=none,fontsize=12]; gaw [color=none,fontsize=12]; gax [color=none,fontsize=12]; gay [color=none,fontsize=12]; gaza [color=none,fontsize=12]; gazb [color=none,fontsize=12]; gazc [color=none,fontsize=12]; gazd [color=none,fontsize=12]; gaze [color=none,fontsize=12]; #gazf [color=none,fontsize=12]; gazg [color=none,fontsize=12]; gazh [color=none,fontsize=12]; gazi [color=none,fontsize=12]; gazj [color=none,fontsize=12]; gazk [color=none,fontsize=12]; gazl [color=none,fontsize=12]; gazm [color=none,fontsize=12]; gazn [color=none,fontsize=12]; gazo [color=none,fontsize=12]; gazp [color=none,fontsize=12]; gazq [color=none,fontsize=12]; gazr [color=none,fontsize=12]; gazs [color=none,fontsize=12]; gazt [color=none,fontsize=12]; gazu [color=none,fontsize=12]; gazv [color=none,fontsize=12]; gazw [color=none,fontsize=12]; gazx [color=none,fontsize=12]; gazy [color=none,fontsize=12]; root -> a [minlen=2]; root -> b [minlen=2]; root -> c [minlen=2]; root -> d [minlen=2]; root -> e [minlen=2]; root -> f [minlen=2]; root -> g [minlen=2]; root -> h [minlen=2]; root -> i [minlen=2]; root -> j [minlen=2]; root -> k [minlen=2]; root -> l [minlen=2]; root -> m [minlen=2]; root -> n [minlen=2]; root -> o [minlen=2]; root -> p [minlen=2]; root -> q [minlen=2]; root -> r [minlen=2]; root -> s [minlen=20]; root -> t [minlen=2]; root -> u [minlen=2]; root -> v [minlen=2]; root -> w [minlen=2]; root -> x [minlen=2]; root -> y [minlen=2]; root -> 0 [minlen=2]; root -> 1 [minlen=2]; root -> 2 [minlen=2]; root -> 3 [minlen=2]; root -> 4 [minlen=2]; root -> 5 [minlen=2]; root -> 6 [minlen=2]; root -> 7 [minlen=2]; root -> 8 [minlen=2]; root -> 9 [minlen=2]; root -> "." [minlen=2]; g -> ga ; g -> gb ; g -> gc ; g -> gd ; g -> ge ; g -> gf ; g -> gg ; g -> gh ; g -> gi ; g -> gj ; g -> gk ; g -> gl ; g -> gm ; g -> gn ; g -> go ; g -> gp ; g -> gq ; g -> gr ; g -> gs ; g -> gt ; g -> gu ; g -> gv ; g -> gw ; g -> gx ; g -> gy ; ga -> gaa ; ga -> gab ; ga -> gac ; ga -> gad ; ga -> gae ; ga -> gaf ; ga -> gag ; ga -> gah ; ga -> gai ; ga -> gaj ; ga -> gak ; ga -> gal ; ga -> gam ; ga -> gan ; ga -> gao ; ga -> gap ; ga -> gaq ; ga -> gaz ; ga -> gas ; ga -> gat ; ga -> gau ; ga -> gav ; ga -> gaw ; ga -> gax ; ga -> gay ; gaz -> gaza ; gaz -> gazb ; gaz -> gazc ; gaz -> gazd ; gaz -> gaze ; gaz -> gazf ; gaz -> gazg ; gaz -> gazh ; gaz -> gazi ; gaz -> gazj ; gaz -> gazk ; gaz -> gazl ; gaz -> gazm ; gaz -> gazn ; gaz -> gazo ; gaz -> gazp ; gaz -> gazq ; gaz -> gazr ; gaz -> gazs ; gaz -> gazt ; gaz -> gazu ; gaz -> gazv ; gaz -> gazw ; gaz -> gazx ; gaz -> gazy ; gazo -> "Blue Tuesday" ; "Blue Tuesday" [ fontsize=10]; // Layout engines: circo dot fdp neato nop nop1 nop2 osage patchwork sfdp twopi } This output is generated with: twopi -os1.png -Tpng s1.dot I'm posting here because the printout is pretty dreadful. All of the nodes hung of "gaz" are overlapping; I've tried specifying nodesep and it is simply ignored. I would like to see the lines from root to the single letters further apart, but again, I can't control that. This seems to be a bug in twopi. The documentation says it should clearly follow these directives, but it doesn't seem to. My questions: Is there any way to make twopi behave? Failing that, is there a better layout engine to use? Thanks.

    Read the article

  • Big Data – Role of Cloud Computing in Big Data – Day 11 of 21

    - by Pinal Dave
    In yesterday’s blog post we learned the importance of the NewSQL. In this article we will understand the role of Cloud in Big Data Story What is Cloud? Cloud is the biggest buzzword around from last few years. Everyone knows about the Cloud and it is extremely well defined online. In this article we will discuss cloud in the context of the Big Data. Cloud computing is a method of providing a shared computing resources to the application which requires dynamic resources. These resources include applications, computing, storage, networking, development and various deployment platforms. The fundamentals of the cloud computing are that it shares pretty much share all the resources and deliver to end users as a service.  Examples of the Cloud Computing and Big Data are Google and Amazon.com. Both have fantastic Big Data offering with the help of the cloud. We will discuss this later in this blog post. There are two different Cloud Deployment Models: 1) The Public Cloud and 2) The Private Cloud Public Cloud Public Cloud is the cloud infrastructure build by commercial providers (Amazon, Rackspace etc.) creates a highly scalable data center that hides the complex infrastructure from the consumer and provides various services. Private Cloud Private Cloud is the cloud infrastructure build by a single organization where they are managing highly scalable data center internally. Here is the quick comparison between Public Cloud and Private Cloud from Wikipedia:   Public Cloud Private Cloud Initial cost Typically zero Typically high Running cost Unpredictable Unpredictable Customization Impossible Possible Privacy No (Host has access to the data Yes Single sign-on Impossible Possible Scaling up Easy while within defined limits Laborious but no limits Hybrid Cloud Hybrid Cloud is the cloud infrastructure build with the composition of two or more clouds like public and private cloud. Hybrid cloud gives best of the both the world as it combines multiple cloud deployment models together. Cloud and Big Data – Common Characteristics There are many characteristics of the Cloud Architecture and Cloud Computing which are also essentially important for Big Data as well. They highly overlap and at many places it just makes sense to use the power of both the architecture and build a highly scalable framework. Here is the list of all the characteristics of cloud computing important in Big Data Scalability Elasticity Ad-hoc Resource Pooling Low Cost to Setup Infastructure Pay on Use or Pay as you Go Highly Available Leading Big Data Cloud Providers There are many players in Big Data Cloud but we will list a few of the known players in this list. Amazon Amazon is arguably the most popular Infrastructure as a Service (IaaS) provider. The history of how Amazon started in this business is very interesting. They started out with a massive infrastructure to support their own business. Gradually they figured out that their own resources are underutilized most of the time. They decided to get the maximum out of the resources they have and hence  they launched their Amazon Elastic Compute Cloud (Amazon EC2) service in 2006. Their products have evolved a lot recently and now it is one of their primary business besides their retail selling. Amazon also offers Big Data services understand Amazon Web Services. Here is the list of the included services: Amazon Elastic MapReduce – It processes very high volumes of data Amazon DynammoDB – It is fully managed NoSQL (Not Only SQL) database service Amazon Simple Storage Services (S3) – A web-scale service designed to store and accommodate any amount of data Amazon High Performance Computing – It provides low-tenancy tuned high performance computing cluster Amazon RedShift – It is petabyte scale data warehousing service Google Though Google is known for Search Engine, we all know that it is much more than that. Google Compute Engine – It offers secure, flexible computing from energy efficient data centers Google Big Query – It allows SQL-like queries to run against large datasets Google Prediction API – It is a cloud based machine learning tool Other Players Besides Amazon and Google we also have other players in the Big Data market as well. Microsoft is also attempting Big Data with the Cloud with Microsoft Azure. Additionally Rackspace and NASA together have initiated OpenStack. The goal of Openstack is to provide a massively scaled, multitenant cloud that can run on any hardware. Thing to Watch The cloud based solutions provides a great integration with the Big Data’s story as well it is very economical to implement as well. However, there are few things one should be very careful when deploying Big Data on cloud solutions. Here is a list of a few things to watch: Data Integrity Initial Cost Recurring Cost Performance Data Access Security Location Compliance Every company have different approaches to Big Data and have different rules and regulations. Based on various factors, one can implement their own custom Big Data solution on a cloud. Tomorrow In tomorrow’s blog post we will discuss about various Operational Databases supporting Big Data. 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

  • SPARC T4-4 Delivers World Record First Result on PeopleSoft Combined Benchmark

    - by Brian
    Oracle's SPARC T4-4 servers running Oracle's PeopleSoft HCM 9.1 combined online and batch benchmark achieved World Record 18,000 concurrent users while executing a PeopleSoft Payroll batch job of 500,000 employees in 43.32 minutes and maintaining online users response time at < 2 seconds. This world record is the first to run online and batch workloads concurrently. This result was obtained with a SPARC T4-4 server running Oracle Database 11g Release 2, a SPARC T4-4 server running PeopleSoft HCM 9.1 application server and a SPARC T4-2 server running Oracle WebLogic Server in the web tier. The SPARC T4-4 server running the application tier used Oracle Solaris Zones which provide a flexible, scalable and manageable virtualization environment. The average CPU utilization on the SPARC T4-2 server in the web tier was 17%, on the SPARC T4-4 server in the application tier it was 59%, and on the SPARC T4-4 server in the database tier was 35% (online and batch) leaving significant headroom for additional processing across the three tiers. The SPARC T4-4 server used for the database tier hosted Oracle Database 11g Release 2 using Oracle Automatic Storage Management (ASM) for database files management with I/O performance equivalent to raw devices. This is the first three tier mixed workload (online and batch) PeopleSoft benchmark also processing PeopleSoft payroll batch workload. Performance Landscape PeopleSoft HR Self-Service and Payroll Benchmark Systems Users Ave Response Search (sec) Ave Response Save (sec) Batch Time (min) Streams SPARC T4-2 (web) SPARC T4-4 (app) SPARC T4-2 (db) 18,000 0.944 0.503 43.32 64 Configuration Summary Application Configuration: 1 x SPARC T4-4 server with 4 x SPARC T4 processors, 3.0 GHz 512 GB memory 5 x 300 GB SAS internal disks 1 x 100 GB and 2 x 300 GB internal SSDs 2 x 10 Gbe HBA Oracle Solaris 11 11/11 PeopleTools 8.52 PeopleSoft HCM 9.1 Oracle Tuxedo, Version 10.3.0.0, 64-bit, Patch Level 031 Java Platform, Standard Edition Development Kit 6 Update 32 Database Configuration: 1 x SPARC T4-4 server with 4 x SPARC T4 processors, 3.0 GHz 256 GB memory 3 x 300 GB SAS internal disks Oracle Solaris 11 11/11 Oracle Database 11g Release 2 Web Tier Configuration: 1 x SPARC T4-2 server with 2 x SPARC T4 processors, 2.85 GHz 256 GB memory 2 x 300 GB SAS internal disks 1 x 100 GB internal SSD Oracle Solaris 11 11/11 PeopleTools 8.52 Oracle WebLogic Server 10.3.4 Java Platform, Standard Edition Development Kit 6 Update 32 Storage Configuration: 1 x Sun Server X2-4 as a COMSTAR head for data 4 x Intel Xeon X7550, 2.0 GHz 128 GB memory 1 x Sun Storage F5100 Flash Array (80 flash modules) 1 x Sun Storage F5100 Flash Array (40 flash modules) 1 x Sun Fire X4275 as a COMSTAR head for redo logs 12 x 2 TB SAS disks with Niwot Raid controller Benchmark Description This benchmark combines PeopleSoft HCM 9.1 HR Self Service online and PeopleSoft Payroll batch workloads to run on a unified database deployed on Oracle Database 11g Release 2. The PeopleSoft HRSS benchmark kit is a Oracle standard benchmark kit run by all platform vendors to measure the performance. It's an OLTP benchmark where DB SQLs are moderately complex. The results are certified by Oracle and a white paper is published. PeopleSoft HR SS defines a business transaction as a series of HTML pages that guide a user through a particular scenario. Users are defined as corporate Employees, Managers and HR administrators. The benchmark consist of 14 scenarios which emulate users performing typical HCM transactions such as viewing paycheck, promoting and hiring employees, updating employee profile and other typical HCM application transactions. All these transactions are well-defined in the PeopleSoft HR Self-Service 9.1 benchmark kit. This benchmark metric is the weighted average response search/save time for all the transactions. The PeopleSoft 9.1 Payroll (North America) benchmark demonstrates system performance for a range of processing volumes in a specific configuration. This workload represents large batch runs typical of a ERP environment during a mass update. The benchmark measures five application business process run times for a database representing large organization. They are Paysheet Creation, Payroll Calculation, Payroll Confirmation, Print Advice forms, and Create Direct Deposit File. The benchmark metric is the cumulative elapsed time taken to complete the Paysheet Creation, Payroll Calculation and Payroll Confirmation business application processes. The benchmark metrics are taken for each respective benchmark while running simultaneously on the same database back-end. Specifically, the payroll batch processes are started when the online workload reaches steady state (the maximum number of online users) and overlap with online transactions for the duration of the steady state. Key Points and Best Practices Two Oracle PeopleSoft Domain sets with 200 application servers each on a SPARC T4-4 server were hosted in 2 separate Oracle Solaris Zones to demonstrate consolidation of multiple application servers, ease of administration and performance tuning. Each Oracle Solaris Zone was bound to a separate processor set, each containing 15 cores (total 120 threads). The default set (1 core from first and third processor socket, total 16 threads) was used for network and disk interrupt handling. This was done to improve performance by reducing memory access latency by using the physical memory closest to the processors and offload I/O interrupt handling to default set threads, freeing up cpu resources for Application Servers threads and balancing application workload across 240 threads. See Also Oracle PeopleSoft Benchmark White Papers oracle.com SPARC T4-2 Server oracle.com OTN SPARC T4-4 Server oracle.com OTN PeopleSoft Enterprise Human Capital Management oracle.com OTN PeopleSoft Enterprise Human Capital Management (Payroll) oracle.com OTN Oracle Solaris oracle.com OTN Oracle Database 11g Release 2 Enterprise Edition oracle.com OTN Disclosure Statement Oracle's PeopleSoft HR and Payroll combined benchmark, www.oracle.com/us/solutions/benchmark/apps-benchmark/peoplesoft-167486.html, results 09/30/2012.

    Read the article

  • Android Bitmap : collision Detecting [on hold]

    - by user2505374
    I am writing an Android game right now and I would need some help in the collision of the wall on screen. When I drag the ball in the top and right it able to collide in wall but when I drag it faster it was able to overlap in the wall. public boolean onTouchEvent(MotionEvent event) { int x = (int) event.getX(); int y = (int) event.getY(); switch (event.getAction()) { // if the player moves case MotionEvent.ACTION_MOVE: { if (playerTouchRect.contains(x, y)) { boolean left = false; boolean right = false; boolean up = false; boolean down = false; boolean canMove = false; boolean foundFinish = false; if (x != pLastXPos) { if (x < pLastXPos) { left = true; } else { right = true; } pLastXPos = x; } if (y != pLastYPos) { if (y < pLastYPos) { up = true; } else { down = true; } pLastYPos = y; } plCellRect = getRectFromPos(x, y); newplRect.set(playerRect); newplRect.left = x - (int) (playerRect.width() / 2); newplRect.right = x + (int) (playerRect.width() / 2); newplRect.top = y - (int) (playerRect.height() / 2); newplRect.bottom = y + (int) (playerRect.height() / 2); int currentRow = 0; int currentCol = 0; currentRow = getRowFromYPos(newplRect.top); currentCol = getColFromXPos(newplRect.right); if(!canMove){ canMove = mapManager.getCurrentTile().pMaze[currentRow][currentCol] == Cell.wall; canMove =true; } finishTest = mapManager.getCurrentTile().pMaze[currentRow][currentCol]; foundA = finishTest == Cell.valueOf(letterNotGet + ""); canMove = mapManager.getCurrentTile().pMaze[currentRow][currentCol] != Cell.wall; canMove = (finishTest == Cell.floor || finishTest == Cell.pl) && canMove; if (canMove) { invalidate(); setTitle(); } if (foundA) { mapManager.getCurrentTile().pMaze[currentRow][currentCol] = Cell.floor; // finishTest letterGotten.add(letterNotGet); playCurrentLetter(); /*sounds.play(sExplosion, 1.0f, 1.0f, 0, 0, 1.5f);*/ foundS = letterNotGet == 's'; letterNotGet++; }if(foundS){ AlertDialog.Builder builder = new AlertDialog.Builder(mainActivity); builder.setTitle(mainActivity.getText(R.string.finished_title)); LayoutInflater inflater = mainActivity.getLayoutInflater(); View view = inflater.inflate(R.layout.finish, null); builder.setView(view); View closeButton =view.findViewById(R.id.closeGame); closeButton.setOnClickListener(new OnClickListener() { @Override public void onClick(View clicked) { if(clicked.getId() == R.id.closeGame) { mainActivity.finish(); } } }); AlertDialog finishDialog = builder.create(); finishDialog.show(); } else { Log.d(TAG, "INFO: updated player position"); playerRect.set(newplRect); setTouchZone(); updatePlayerCell(); } } // end of (CASE) if playerTouch break; } // end of (SWITCH) Case motion }//end of Switch return true; }//end of TouchEvent private void finish() { // TODO Auto-generated method stub } public int getColFromXPos(int xPos) { val = xPos / (pvWidth / mapManager.getCurrentTile().pCols); if (val == mapManager.getCurrentTile().pCols) { val = mapManager.getCurrentTile().pCols - 1; } return val; } /** * Given a y pixel position, return the row of the cell it is in This is * used when determining the type of adjacent Cells. * * @param yPos * y position in pixels * @return The cell this position is in */ public int getRowFromYPos(int yPos) { val = yPos / (pvHeight / mapManager.getCurrentTile().pRows); if (val == mapManager.getCurrentTile().pRows) { val = mapManager.getCurrentTile().pRows - 1; } return val; } /** * When preserving the position we need to know which cell the player is in, * so calculate it from the centre on its Rect */ public void updatePlayerCell() { plCell.x = (playerRect.left + (playerRect.width() / 2)) / (pvWidth / mapManager.getCurrentTile().pCols); plCell.y = (playerRect.top + (playerRect.height() / 2)) / (pvHeight / mapManager.getCurrentTile().pRows); if (mapManager.getCurrentTile().pMaze[plCell.y][plCell.x] == Cell.floor) { for (int row = 0; row < mapManager.getCurrentTile().pRows; row++) { for (int col = 0; col < mapManager.getCurrentTile().pCols; col++) { if (mapManager.getCurrentTile().pMaze[row][col] == Cell.pl) { mapManager.getCurrentTile().pMaze[row][col] = Cell.floor; break; } } } mapManager.getCurrentTile().pMaze[plCell.y][plCell.x] = Cell.pl; } } public Rect getRectFromPos(int x, int y) { calcCell.left = ((x / cellWidth) + 0) * cellWidth; calcCell.right = calcCell.left + cellWidth; calcCell.top = ((y / cellHeight) + 0) * cellHeight; calcCell.bottom = calcCell.top + cellHeight; Log.d(TAG, "Rect: " + calcCell + " Player: " + playerRect); return calcCell; } public void setPlayerRect(Rect newplRect) { playerRect.set(newplRect); } private void setTouchZone() { playerTouchRect.set( playerRect.left - playerRect.width() / TOUCH_ZONE, playerRect.top - playerRect.height() / TOUCH_ZONE, playerRect.right + playerRect.width() / TOUCH_ZONE, playerRect.bottom + playerRect.height() / TOUCH_ZONE); } public Rect getPlayerRect() { return playerRect; } public Point getPlayerCell() { return plCell; } public void setPlayerCell(Point cell) { plCell = cell; }

    Read the article

  • Cocos2d: Moving background on update: offsett issue

    - by mm24
    working with Objective C, iOS and Cocos2d I am developing a vertical scrolling shooter game for iPhone (retina display models with 640 width x 960 height pixel resolution). My basic algorithm works as following: I create two instances of an image that has exactly 640 width x 960 height pixel of resolution, which we will call imageA and imageB I then set the two imags with exactly 480.0f of offset from each other, as the screenSize of a CCScene is set by default to 480.0f. At each update method call I move the two images by the same value. I make sure that their offsett stays to 480.0f However when running the game I see a 1 pixel height line between the two images. This literally bugs me and would like to adjust this. What am I doing wrong? This is a zoom in on the background when the "offsett line" is visible. The white line you can see divides the two background images and is not meant to exist as both images are completely black :): If I change the yPositionOfSecondElement value to 479.0f until the first loop the two images overlap correctly, but as soon as the loop starts the two images starts having an offsett of -1.0f. Here is the initialization code: -(void) init { //... screenHeight = 480.0f; yPositionOfSecondElement= screenHeight;//I tried subtracting an offsett of -1 but eventually the image would go wrong again yPositionOfFirstElement = 0.0f; loopedBackgroundImageInstanceA = [BackgroundLoopedImage loopImageForLevel:levelName]; loopedBackgroundImageInstanceA.anchorPoint = CGPointMake(0.5f, 0.0f); loopedBackgroundImageInstanceA.position = CGPointMake(160.0f, yPositionOfFirstElement); [node addChild:loopedBackgroundImageInstanceA z:zLevelBackground]; //loopedBackgroundImageInstanceA.color= ccRED; loopedBackgroundImageInstanceB = [BackgroundLoopedImage loopImageForLevel:levelName]; loopedBackgroundImageInstanceB.anchorPoint = CGPointMake(0.5f, 0.0f); loopedBackgroundImageInstanceB.position = CGPointMake(160.0f, yPositionOfSecondElement); [node addChild:loopedBackgroundImageInstanceB z:zLevelBackground]; //.... } And here is the move code called at each update: -(void) moveBackgroundSprites:(BackgroundLoopedImage*)imageA :(BackgroundLoopedImage*)imageB :(ccTime)delta { isEligibleToMove=false; //This is done to avoid rounding errors float yStep = delta * [GameController sharedGameController].currentBackgroundSpeed; NSString* formattedNumber = [NSString stringWithFormat:@"%.02f", yStep]; yStep = atof([formattedNumber UTF8String]); //First should adjust position of images [self adjustPosition:imageA :imageB]; //The can get the actual image position CGPoint posA = imageA.position; CGPoint posB = imageB.position; //Here could verify if the checksum is equal to the required difference (should be 479.0f) if (![self verifyCheckSum:posA :posB]) { CCLOG(@"does not comply A"); } //At this stage can compute the hypotetical new position CGPoint newPosA = CGPointMake(posA.x, posA.y - yStep); CGPoint newPosB = CGPointMake(posB.x, posB.y - yStep); // Reposition stripes when they're out of bounds if (newPosA.y <= -yPositionOfSecondElement) { newPosA.y = yPositionOfSecondElement; [imageA shuffle]; if (timeElapsed>=endTime && hasReachedEndLevel==FALSE) { hasReachedEndLevel=TRUE; shouldMoveImageEnd=TRUE; } } else if (newPosB.y <= -yPositionOfSecondElement) { newPosB.y = yPositionOfSecondElement; [imageB shuffle]; if (timeElapsed>=endTime && hasReachedEndLevel==FALSE) { hasReachedEndLevel=TRUE; shouldMoveImageEnd=TRUE; } } //Here should verify that the check sum is equal to 479.0f if (![self verifyCheckSum:posA :posB]) { CCLOG(@"does not comply B"); } imageA.position = newPosA; imageB.position = newPosB; //Here could verify that the check sum is equal to 479.0f if (![self verifyCheckSum:posA :posB]) { CCLOG(@"does not comply C"); } isEligibleToMove=true; } -(BOOL) verifyCheckSum:(CGPoint)posA :(CGPoint)posB { BOOL comply = false; float sum = 0.0f; if (posA.y > posB.y) { sum = posA.y - posB.y; } else if (posB.y > posA.y){ sum = posB.y - posA.y; } else{ return false; } if (sum!=yPositionOfSecondElement) { comply= false; } else{ comply=true; } return comply; } And here is what happens on the update: if(shouldMoveImageA && shouldMoveImageB) { if (isEligibleToMove) { [self moveBackgroundSprites:loopedBackgroundImageInstanceA :loopedBackgroundImageInstanceB :delta]; } Forget about shouldMoveImageA and shouldMoveImageB, this is just for when the background reaches the end of level, this works.

    Read the article

  • What to leave when you're leaving

    - by BuckWoody
    There's already a post on this topic - sort of. I read this entry, where the author did a good job on a few steps, but I found that a few other tips might be useful, so if you want to check that one out and then this post, you might be able to put together your own plan for when you leave your job.  I once took over the system administrator (of which the Oracle and SQL Server servers were a part) at a mid-sized firm. The outgoing administrator had about a two- week-long scheduled overlap with me, but was angry at the company and told me "hey, I know this is going to be hard on you, but I want them to know how important I was. I'm not telling you where anything is or what the passwords are. Good luck!" He then quit that day. It took me about three days to find all of the servers and crack the passwords. Yes, the company tried to take legal action against the guy and all that, but he moved back to his home country and so largely got away with it. Obviously, this isn't the way to leave a job. Many of us have changed jobs in the past, and most of us try to be very professional about the transition to a new team, regardless of the feelings about a particular company. I've been treated badly at a firm, but that is no reason to leave a mess for someone else. So here's what you should put into place at a minimum before you go. Most of this is common sense - which of course isn't very common these days - and another good rule is just to ask yourself "what would I want to know"? The article I referenced at the top of this post focuses on a lot of documentation of the systems. I think that's fine, but in actuality, I really don't need that. Even with this kind of documentation, I still perform a full audit on the systems, so in the end I create my own system documentation. There are actually only four big items I need to know to get started with the systems: 1. Where is everything/everybody?The first thing I need to know is where all of the systems are. I mean not only the street address, but the closet or room, the rack number, the IU number in the rack, the SAN luns, all that. A picture here is worth a thousand words, which is why I really like Visio. It combines nice graphics, full text and all that. But use whatever you have to tell someone the physical locations of the boxes. Also, tell them the physical location of the folks in charge of those boxes (in case you aren't) or who share that responsibility. And by "where" in this case, I mean names and phones.  2. What do they do?For both the servers and the people, tell them what they do. If it's a database server, detail what each database does and what application goes to that, and who "owns" that application. In my mind, this is one of hte most important things a Data Professional needs to know. In the case of the other administrtors or co-owners, document each person's responsibilities.   3. What are the credentials?Logging on/in and gaining access to the buildings are things that the new Data Professional will need to do to successfully complete their job. This means service accounts, certificates, all of that. The first thing they should do, of course, is change the passwords on all that, but the first thing they need is the ability to do that!  4. What is out of the ordinary?This is the most tricky, and perhaps the next most important thing to know. Did you have to use a "special" driver for that video card on server X? Is the person that co-owns an application with you mentally unstable (like me) or have special needs, like "don't talk to Buck before he's had coffee. Nothing will make any sense"? Do you have service pack requirements for a specific setup? Write all that down. Anything that took you a day or longer to make work is probably a candidate here. This is my short list - anything you care to add? Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • WPF - Overlapping Custom Tabs in a TabControl and ZIndex

    - by Rachel
    Problem I have a custom tab control using Chrome-shaped tabs that binds to a ViewModel. Because of the shape, the edges overlap a bit. I have a function that sets the tabItem's ZIndex on TabControl_SelectionChanged which works fine for selecting tabs, and dragging/dropping tabs, however when I Add or Close a tab via a Relay Command I am getting unusual results. Does anyone have any ideas? Default View http://i193.photobucket.com/albums/z197/Lady53461/tabs_default.jpg Removing Tabs http:/i193.photobucket.com/albums/z197/Lady53461/tabs_removing.jpg Adding 2 or more Tabs in a row http:/i193.photobucket.com/albums/z197/Lady53461/tabs_adding.jpg Code to set ZIndex private void PrimaryTabControl_SelectionChanged(object sender, SelectionChangedEventArgs e) { if (e.Source is TabControl) { TabControl tabControl = sender as TabControl; ItemContainerGenerator icg = tabControl.ItemContainerGenerator; if (icg.Status == System.Windows.Controls.Primitives.GeneratorStatus.ContainersGenerated) { foreach (object o in tabControl.Items) { UIElement tabItem = icg.ContainerFromItem(o) as UIElement; Panel.SetZIndex(tabItem, (o == tabControl.SelectedItem ? 100 : 90 - tabControl.Items.IndexOf(o))); } } } } By using breakpoints I can see that it is correctly setting the ZIndex to what I want it to, however the layout is not displaying the changes. I know some of the changes are in effect because if none of them were working then the tab edges would be reversed (the right tabs would be drawn on top of the left ones). Clicking a tab will correctly set the zindex of all tabs (including the one that should be drawn on top) and dragging/dropping them to rearrange them also renders correctly (which removes and reinserts the tab item). The only difference I can think of is I am using the MVVM design pattern and the buttons that Add/Close tabs are relay commands. Does anyone have any idea why this is happening and how I can fix it?? p.s. I did try setting a ZIndex in my ViewModel and binding to it, however the same thing happens when adding/removing tabs via the relay command. EDIT: Being a new user I couldn't post images and could only post 1 link. Images just show a picture of what the tags render as after each scenario. Adding more then 1 at a time will not reset the zindex of other recently-added tabs so they go behind the tab on the Right, and closing tabs does not correctly render the ZIndex of the SelectedTab that replaces it and it shows up behind the tab on its right.

    Read the article

  • How do I work around an HTML Table rendering bug in IE 7?

    - by osmaniac
    I have a table. Some cells span multiple columns. Some cells span multiple rows and columns. But one row (which spans all columns but the rightmost one) creates an artifact. Part of the text in the cell is erroneously repeated left justified on a new row just below the table. I'm baffled. I tried rendering with and without "table-layout: fixed;". Same result. When I originally composed the design using just HTML and CSS, it looked great. But then I worked it into a page and had to add more columns to my master table the right to hold buttons. These buttons are in three groups, each having their own div to control floating and rewrapping when the window gets narrower. One div has another table inside it that groups a single row of buttons. Thus I have table inside div inside td inside outer table. I would prefer a simpler design, but how? This is what I want to have: ................................................................................... . . . . Four rows of data . Three groups of buttons that can reflow . . With several columns . if window gets narrower . . meticulously layed out, . . . That should not resize . . . when window gets narrower . . ................................................................................... . One more row of data spanning the whole screen which stays below the buttons . ................................................................................... What I was doing was putting the three divs with the buttons in the upper right in a single cell that spanned four rows. What other opportunities does CSS offer? The buttons are not allowed to overlap the data on the left or go past the data line below. The original design had the divs with the buttons NOT in a table with the data, but when the window gets narrow, some of the buttons flow such that they go underneath the data, which looks bad. I would post actual HTML, except it is generated by ASP, huge, with lots of CSS styling, and the feature that lets me view the final HTML is not working at the moment. (Built in security in the application.)

    Read the article

  • Get non-overlapping dates ranges for prices history data

    - by Anonymouse
    Hello, Let's assume that I have the following table: CREATE TABLE [dbo].[PricesHist]( [Product] varchar NOT NULL, [Price] [float] NOT NULL, [StartDate] [datetime] NOT NULL, [EndDate] [datetime] NOT NULL ) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D2C00000000 AS DateTime), CAST(0x00009D2C00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D2D00000000 AS DateTime), CAST(0x00009D2D00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 2.5, CAST(0x00009D2E00000000 AS DateTime), CAST(0x00009D2E00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3000000000 AS DateTime), CAST(0x00009D3000000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3100000000 AS DateTime), CAST(0x00009D3100000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3400000000 AS DateTime), CAST(0x00009D3400000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 2.5, CAST(0x00009D3500000000 AS DateTime), CAST(0x00009D3500000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3600000000 AS DateTime), CAST(0x00009D3600000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3700000000 AS DateTime), CAST(0x00009D3700000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3800000000 AS DateTime), CAST(0x00009D3800000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3A00000000 AS DateTime), CAST(0x00009D3A00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3B00000000 AS DateTime), CAST(0x00009D3B00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 2.5, CAST(0x00009D3C00000000 AS DateTime), CAST(0x00009D3C00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3D00000000 AS DateTime), CAST(0x00009D3D00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3E00000000 AS DateTime), CAST(0x00009D3E00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D3F00000000 AS DateTime), CAST(0x00009D3F00000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D4100000000 AS DateTime), CAST(0x00009D4100000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D4200000000 AS DateTime), CAST(0x00009D4200000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 2.5, CAST(0x00009D4300000000 AS DateTime), CAST(0x00009D4300000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D4400000000 AS DateTime), CAST(0x00009D4400000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D4500000000 AS DateTime), CAST(0x00009D4500000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D4600000000 AS DateTime), CAST(0x00009D4600000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 4.9, CAST(0x00009D4800000000 AS DateTime), CAST(0x00009D4800000000 AS DateTime)) INSERT [dbo].[PricesHist] ([Product], [Price], [StartDate], [EndDate]) VALUES (N'Apples', 2.5, CAST(0x00009D4A00000000 AS DateTime), CAST(0x00009D4A00000000 AS DateTime)) As you can see, there are two prices on that month for Apples. 4.90 and 2.50. In order to tidy this table up, I need to get this information as a date range rather than a row per day as it currently is. I can obviously do this with Min and Max aggregates easily but the ranges overlap and other business code expect non-overlapping ranges. I also tried to achieve this with self joins and row_number(), but without much success... Here is what I'm trying to achieve as the output: Product | StartDate | EndDate | Price ------------------------------------------- Apples | 01 Mar 2010 | 02 Mar 2010 | 4.90 Apples | 03 Mar 2010 | 03 Mar 2010 | 2.50 Apples | 05 Mar 2010 | 09 Mar 2010 | 4.90 Apples | 10 Mar 2010 | 10 Mar 2010 | 2.50 Apples | 11 Mar 2010 | 16 Mar 2010 | 4.90 Apples | 17 Mar 2010 | 17 Mar 2010 | 2.50 Apples | 18 Mar 2010 | 23 Mar 2010 | 4.90 Apples | 24 Mar 2010 | 24 Mar 2010 | 2.50 Apples | 25 Mar 2010 | 30 Mar 2010 | 4.90 Apples | 31 Mar 2010 | 31 Mar 2010 | 2.50 What would please be the best approach to get this done? Thanks a lot in advance,

    Read the article

  • How do I prevent the concurrent execution of a javascript function?

    - by RyanV
    I am making a ticker similar to the "From the AP" one at The Huffington Post, using jQuery. The ticker rotates through a ul, either by user command (clicking an arrow) or by an auto-scroll. Each list-item is display:none by default. It is revealed by the addition of a "showHeadline" class which is display:list-item. HTML for the UL Looks like this: <ul class="news" id="news"> <li class="tickerTitle showHeadline">Test Entry</li> <li class="tickerTitle">Test Entry2</li> <li class="tickerTitle">Test Entry3</li> </ul> When the user clicks the right arrow, or the auto-scroll setTimeout goes off, it runs a tickForward() function: function tickForward(){ var $active = $('#news li.showHeadline'); var $next = $active.next(); if($next.length==0) $next = $('#news li:first'); $active.stop(true, true); $active.fadeOut('slow', function() {$active.removeClass('showHeadline');}); setTimeout(function(){$next.fadeIn('slow', function(){$next.addClass('showHeadline');})}, 1000); if(isPaused == true){ } else{ startScroll() } }; This is heavily inspired by Jon Raasch's A Simple jQuery Slideshow. Basically, find what's visible, what should be visible next, make the visible thing fade and remove the class that marks it as visible, then fade in the next thing and add the class that makes it visible. Now, everything is hunky-dory if the auto-scroll is running, kicking off tickForward() once every three seconds. But if the user clicks the arrow button repeatedly, it creates two negative conditions: Rather than advance quickly through the list for just the number of clicks made, it continues scrolling at a faster-than-normal rate indefinitely. It can produce a situation where two (or more) list items are given the .showHeadline class, so there's overlap on the list. I can see these happening (especially #2) because the tickForward() function can run concurrently with itself, producing different sets of $active and $next. So I think my question is: What would be the best way to prevent concurrent execution of the tickForward() method? Some things I have tried or considered: Setting a Flag: When tickForward() runs, it sets an isRunning flag to true, and sets it back to false right before it ends. The logic for the event handler is set to only call tickForward() if isRunning is false. I tried a simple implementation of this, and isRunning never appeared to be changed. The jQuery queue(): I think it would be useful to queue up the tickForward() commands, so if you clicked it five times quickly, it would still run as commanded but wouldn't run concurrently. However, in my cursory reading on the subject, it appears that a queue has to be attached to the object its queue applies to, and since my tickForward() method affects multiple lis, I don't know where I'd attach it.

    Read the article

  • Conceal packet loss in PCM stream

    - by ZeroDefect
    I am looking to use 'Packet Loss Concealment' to conceal lost PCM frames in an audio stream. Unfortunately, I cannot find a library that is accessible without all the licensing restrictions and code bloat (...up for some suggestions though). I have located some GPL code written by Steve Underwood for the Asterisk project which implements PLC. There are several limitations; although, as Steve suggests in his code, his algorithm can be applied to different streams with a bit of work. Currently, the code works with 8kHz 16-bit signed mono streams. Variations of the code can be found through a simple search of Google Code Search. My hope is that I can adapt the code to work with other streams. Initially, the goal is to adjust the algorithm for 8+ kHz, 16-bit signed, multichannel audio (all in a C++ environment). Eventually, I'm looking to make the code available under the GPL license in hopes that it could be of benefit to others... Attached is the code below with my efforts. The code includes a main function that will "drop" a number of frames with a given probability. Unfortunately, the code does not quite work as expected. I'm receiving EXC_BAD_ACCESS when running in gdb, but I don't get a trace from gdb when using 'bt' command. Clearly, I'm trampimg on memory some where but not sure exactly where. When I comment out the *amdf_pitch* function, the code runs without crashing... int main (int argc, char *argv[]) { std::ifstream fin("C:\\cc32kHz.pcm"); if(!fin.is_open()) { std::cout << "Failed to open input file" << std::endl; return 1; } std::ofstream fout_repaired("C:\\cc32kHz_repaired.pcm"); if(!fout_repaired.is_open()) { std::cout << "Failed to open output repaired file" << std::endl; return 1; } std::ofstream fout_lossy("C:\\cc32kHz_lossy.pcm"); if(!fout_lossy.is_open()) { std::cout << "Failed to open output repaired file" << std::endl; return 1; } audio::PcmConcealer Concealer; Concealer.Init(1, 16, 32000); //Generate random numbers; srand( time(NULL) ); int value = 0; int probability = 5; while(!fin.eof()) { char arr[2]; fin.read(arr, 2); //Generate's random number; value = rand() % 100 + 1; if(value <= probability) { char blank[2] = {0x00, 0x00}; fout_lossy.write(blank, 2); //Fill in data; Concealer.Fill((int16_t *)blank, 1); fout_repaired.write(blank, 2); } else { //Write data to file; fout_repaired.write(arr, 2); fout_lossy.write(arr, 2); Concealer.Receive((int16_t *)arr, 1); } } fin.close(); fout_repaired.close(); fout_lossy.close(); return 0; } PcmConcealer.hpp /* * Code adapted from Steve Underwood of the Asterisk Project. This code inherits * the same licensing restrictions as the Asterisk Project. */ #ifndef __PCMCONCEALER_HPP__ #define __PCMCONCEALER_HPP__ /** 1. What does it do? The packet loss concealment module provides a suitable synthetic fill-in signal, to minimise the audible effect of lost packets in VoIP applications. It is not tied to any particular codec, and could be used with almost any codec which does not specify its own procedure for packet loss concealment. Where a codec specific concealment procedure exists, the algorithm is usually built around knowledge of the characteristics of the particular codec. It will, therefore, generally give better results for that particular codec than this generic concealer will. 2. How does it work? While good packets are being received, the plc_rx() routine keeps a record of the trailing section of the known speech signal. If a packet is missed, plc_fillin() is called to produce a synthetic replacement for the real speech signal. The average mean difference function (AMDF) is applied to the last known good signal, to determine its effective pitch. Based on this, the last pitch period of signal is saved. Essentially, this cycle of speech will be repeated over and over until the real speech resumes. However, several refinements are needed to obtain smooth pleasant sounding results. - The two ends of the stored cycle of speech will not always fit together smoothly. This can cause roughness, or even clicks, at the joins between cycles. To soften this, the 1/4 pitch period of real speech preceeding the cycle to be repeated is blended with the last 1/4 pitch period of the cycle to be repeated, using an overlap-add (OLA) technique (i.e. in total, the last 5/4 pitch periods of real speech are used). - The start of the synthetic speech will not always fit together smoothly with the tail of real speech passed on before the erasure was identified. Ideally, we would like to modify the last 1/4 pitch period of the real speech, to blend it into the synthetic speech. However, it is too late for that. We could have delayed the real speech a little, but that would require more buffer manipulation, and hurt the efficiency of the no-lost-packets case (which we hope is the dominant case). Instead we use a degenerate form of OLA to modify the start of the synthetic data. The last 1/4 pitch period of real speech is time reversed, and OLA is used to blend it with the first 1/4 pitch period of synthetic speech. The result seems quite acceptable. - As we progress into the erasure, the chances of the synthetic signal being anything like correct steadily fall. Therefore, the volume of the synthesized signal is made to decay linearly, such that after 50ms of missing audio it is reduced to silence. - When real speech resumes, an extra 1/4 pitch period of sythetic speech is blended with the start of the real speech. If the erasure is small, this smoothes the transition. If the erasure is long, and the synthetic signal has faded to zero, the blending softens the start up of the real signal, avoiding a kind of "click" or "pop" effect that might occur with a sudden onset. 3. How do I use it? Before audio is processed, call plc_init() to create an instance of the packet loss concealer. For each received audio packet that is acceptable (i.e. not including those being dropped for being too late) call plc_rx() to record the content of the packet. Note this may modify the packet a little after a period of packet loss, to blend real synthetic data smoothly. When a real packet is not available in time, call plc_fillin() to create a sythetic substitute. That's it! */ /*! Minimum allowed pitch (66 Hz) */ #define PLC_PITCH_MIN(SAMPLE_RATE) ((double)(SAMPLE_RATE) / 66.6) /*! Maximum allowed pitch (200 Hz) */ #define PLC_PITCH_MAX(SAMPLE_RATE) ((SAMPLE_RATE) / 200) /*! Maximum pitch OLA window */ //#define PLC_PITCH_OVERLAP_MAX(SAMPLE_RATE) ((PLC_PITCH_MIN(SAMPLE_RATE)) >> 2) /*! The length over which the AMDF function looks for similarity (20 ms) */ #define CORRELATION_SPAN(SAMPLE_RATE) ((20 * (SAMPLE_RATE)) / 1000) /*! History buffer length. The buffer must also be at leat 1.25 times PLC_PITCH_MIN, but that is much smaller than the buffer needs to be for the pitch assessment. */ //#define PLC_HISTORY_LEN(SAMPLE_RATE) ((CORRELATION_SPAN(SAMPLE_RATE)) + (PLC_PITCH_MIN(SAMPLE_RATE))) namespace audio { typedef struct { /*! Consecutive erased samples */ int missing_samples; /*! Current offset into pitch period */ int pitch_offset; /*! Pitch estimate */ int pitch; /*! Buffer for a cycle of speech */ float *pitchbuf;//[PLC_PITCH_MIN]; /*! History buffer */ short *history;//[PLC_HISTORY_LEN]; /*! Current pointer into the history buffer */ int buf_ptr; } plc_state_t; class PcmConcealer { public: PcmConcealer(); ~PcmConcealer(); void Init(int channels, int bit_depth, int sample_rate); //Process a block of received audio samples. int Receive(short amp[], int frames); //Fill-in a block of missing audio samples. int Fill(short amp[], int frames); void Destroy(); private: int amdf_pitch(int min_pitch, int max_pitch, short amp[], int channel_index, int frames); void save_history(plc_state_t *s, short *buf, int channel_index, int frames); void normalise_history(plc_state_t *s); /** Holds the states of each of the channels **/ std::vector< plc_state_t * > ChannelStates; int plc_pitch_min; int plc_pitch_max; int plc_pitch_overlap_max; int correlation_span; int plc_history_len; int channel_count; int sample_rate; bool Initialized; }; } #endif PcmConcealer.cpp /* * Code adapted from Steve Underwood of the Asterisk Project. This code inherits * the same licensing restrictions as the Asterisk Project. */ #include "audio/PcmConcealer.hpp" /* We do a straight line fade to zero volume in 50ms when we are filling in for missing data. */ #define ATTENUATION_INCREMENT 0.0025 /* Attenuation per sample */ #if !defined(INT16_MAX) #define INT16_MAX (32767) #define INT16_MIN (-32767-1) #endif #ifdef WIN32 inline double rint(double x) { return floor(x + 0.5); } #endif inline short fsaturate(double damp) { if (damp > 32767.0) return INT16_MAX; if (damp < -32768.0) return INT16_MIN; return (short)rint(damp); } namespace audio { PcmConcealer::PcmConcealer() : Initialized(false) { } PcmConcealer::~PcmConcealer() { Destroy(); } void PcmConcealer::Init(int channels, int bit_depth, int sample_rate) { if(Initialized) return; if(channels <= 0 || bit_depth != 16) return; Initialized = true; channel_count = channels; this->sample_rate = sample_rate; ////////////// double min = PLC_PITCH_MIN(sample_rate); int imin = (int)min; double max = PLC_PITCH_MAX(sample_rate); int imax = (int)max; plc_pitch_min = imin; plc_pitch_max = imax; plc_pitch_overlap_max = (plc_pitch_min >> 2); correlation_span = CORRELATION_SPAN(sample_rate); plc_history_len = correlation_span + plc_pitch_min; ////////////// for(int i = 0; i < channel_count; i ++) { plc_state_t *t = new plc_state_t; memset(t, 0, sizeof(plc_state_t)); t->pitchbuf = new float[plc_pitch_min]; t->history = new short[plc_history_len]; ChannelStates.push_back(t); } } void PcmConcealer::Destroy() { if(!Initialized) return; while(ChannelStates.size()) { plc_state_t *s = ChannelStates.at(0); if(s) { if(s->history) delete s->history; if(s->pitchbuf) delete s->pitchbuf; memset(s, 0, sizeof(plc_state_t)); delete s; } ChannelStates.erase(ChannelStates.begin()); } ChannelStates.clear(); Initialized = false; } //Process a block of received audio samples. int PcmConcealer::Receive(short amp[], int frames) { if(!Initialized) return 0; int j = 0; for(int k = 0; k < ChannelStates.size(); k++) { int i; int overlap_len; int pitch_overlap; float old_step; float new_step; float old_weight; float new_weight; float gain; plc_state_t *s = ChannelStates.at(k); if (s->missing_samples) { /* Although we have a real signal, we need to smooth it to fit well with the synthetic signal we used for the previous block */ /* The start of the real data is overlapped with the next 1/4 cycle of the synthetic data. */ pitch_overlap = s->pitch >> 2; if (pitch_overlap > frames) pitch_overlap = frames; gain = 1.0 - s->missing_samples * ATTENUATION_INCREMENT; if (gain < 0.0) gain = 0.0; new_step = 1.0/pitch_overlap; old_step = new_step*gain; new_weight = new_step; old_weight = (1.0 - new_step)*gain; for (i = 0; i < pitch_overlap; i++) { int index = (i * channel_count) + j; amp[index] = fsaturate(old_weight * s->pitchbuf[s->pitch_offset] + new_weight * amp[index]); if (++s->pitch_offset >= s->pitch) s->pitch_offset = 0; new_weight += new_step; old_weight -= old_step; if (old_weight < 0.0) old_weight = 0.0; } s->missing_samples = 0; } save_history(s, amp, j, frames); j++; } return frames; } //Fill-in a block of missing audio samples. int PcmConcealer::Fill(short amp[], int frames) { if(!Initialized) return 0; int j =0; for(int k = 0; k < ChannelStates.size(); k++) { short *tmp = new short[plc_pitch_overlap_max]; int i; int pitch_overlap; float old_step; float new_step; float old_weight; float new_weight; float gain; short *orig_amp; int orig_len; orig_amp = amp; orig_len = frames; plc_state_t *s = ChannelStates.at(k); if (s->missing_samples == 0) { // As the gap in real speech starts we need to assess the last known pitch, //and prepare the synthetic data we will use for fill-in normalise_history(s); s->pitch = amdf_pitch(plc_pitch_min, plc_pitch_max, s->history + plc_history_len - correlation_span - plc_pitch_min, j, correlation_span); // We overlap a 1/4 wavelength pitch_overlap = s->pitch >> 2; // Cook up a single cycle of pitch, using a single of the real signal with 1/4 //cycle OLA'ed to make the ends join up nicely // The first 3/4 of the cycle is a simple copy for (i = 0; i < s->pitch - pitch_overlap; i++) s->pitchbuf[i] = s->history[plc_history_len - s->pitch + i]; // The last 1/4 of the cycle is overlapped with the end of the previous cycle new_step = 1.0/pitch_overlap; new_weight = new_step; for ( ; i < s->pitch; i++) { s->pitchbuf[i] = s->history[plc_history_len - s->pitch + i]*(1.0 - new_weight) + s->history[plc_history_len - 2*s->pitch + i]*new_weight; new_weight += new_step; } // We should now be ready to fill in the gap with repeated, decaying cycles // of what is in pitchbuf // We need to OLA the first 1/4 wavelength of the synthetic data, to smooth // it into the previous real data. To avoid the need to introduce a delay // in the stream, reverse the last 1/4 wavelength, and OLA with that. gain = 1.0; new_step = 1.0/pitch_overlap; old_step = new_step; new_weight = new_step; old_weight = 1.0 - new_step; for (i = 0; i < pitch_overlap; i++) { int index = (i * channel_count) + j; amp[index] = fsaturate(old_weight * s->history[plc_history_len - 1 - i] + new_weight * s->pitchbuf[i]); new_weight += new_step; old_weight -= old_step; if (old_weight < 0.0) old_weight = 0.0; } s->pitch_offset = i; } else { gain = 1.0 - s->missing_samples*ATTENUATION_INCREMENT; i = 0; } for ( ; gain > 0.0 && i < frames; i++) { int index = (i * channel_count) + j; amp[index] = s->pitchbuf[s->pitch_offset]*gain; gain -= ATTENUATION_INCREMENT; if (++s->pitch_offset >= s->pitch) s->pitch_offset = 0; } for ( ; i < frames; i++) { int index = (i * channel_count) + j; amp[i] = 0; } s->missing_samples += orig_len; save_history(s, amp, j, frames); delete [] tmp; j++; } return frames; } void PcmConcealer::save_history(plc_state_t *s, short *buf, int channel_index, int frames) { if (frames >= plc_history_len) { /* Just keep the last part of the new data, starting at the beginning of the buffer */ //memcpy(s->history, buf + len - plc_history_len, sizeof(short)*plc_history_len); int frames_to_copy = plc_history_len; for(int i = 0; i < frames_to_copy; i ++) { int index = (channel_count * (i + frames - plc_history_len)) + channel_index; s->history[i] = buf[index]; } s->buf_ptr = 0; return; } if (s->buf_ptr + frames > plc_history_len) { /* Wraps around - must break into two sections */ //memcpy(s->history + s->buf_ptr, buf, sizeof(short)*(plc_history_len - s->buf_ptr)); short *hist_ptr = s->history + s->buf_ptr; int frames_to_copy = plc_history_len - s->buf_ptr; for(int i = 0; i < frames_to_copy; i ++) { int index = (channel_count * i) + channel_index; hist_ptr[i] = buf[index]; } frames -= (plc_history_len - s->buf_ptr); //memcpy(s->history, buf + (plc_history_len - s->buf_ptr), sizeof(short)*len); frames_to_copy = frames; for(int i = 0; i < frames_to_copy; i ++) { int index = (channel_count * (i + (plc_history_len - s->buf_ptr))) + channel_index; s->history[i] = buf[index]; } s->buf_ptr = frames; return; } /* Can use just one section */ //memcpy(s->history + s->buf_ptr, buf, sizeof(short)*len); short *hist_ptr = s->history + s->buf_ptr; int frames_to_copy = frames; for(int i = 0; i < frames_to_copy; i ++) { int index = (channel_count * i) + channel_index; hist_ptr[i] = buf[index]; } s->buf_ptr += frames; } void PcmConcealer::normalise_history(plc_state_t *s) { short *tmp = new short[plc_history_len]; if (s->buf_ptr == 0) return; memcpy(tmp, s->history, sizeof(short)*s->buf_ptr); memcpy(s->history, s->history + s->buf_ptr, sizeof(short)*(plc_history_len - s->buf_ptr)); memcpy(s->history + plc_history_len - s->buf_ptr, tmp, sizeof(short)*s->buf_ptr); s->buf_ptr = 0; delete [] tmp; } int PcmConcealer::amdf_pitch(int min_pitch, int max_pitch, short amp[], int channel_index, int frames) { int i; int j; int acc; int min_acc; int pitch; pitch = min_pitch; min_acc = INT_MAX; for (i = max_pitch; i <= min_pitch; i++) { acc = 0; for (j = 0; j < frames; j++) { int index1 = (channel_count * (i+j)) + channel_index; int index2 = (channel_count * j) + channel_index; //std::cout << "Index 1: " << index1 << ", Index 2: " << index2 << std::endl; acc += abs(amp[index1] - amp[index2]); } if (acc < min_acc) { min_acc = acc; pitch = i; } } std::cout << "Pitch: " << pitch << std::endl; return pitch; } } P.S. - I must confess that digital audio is not my forte...

    Read the article

  • PagerView overlapping PagerTabStrip / PagerTitleStrip

    - by user1256169
    I've been trying for about a week to get my PagerView not to overlap the TitleStrip. I've tried absolutely everything I can think of, and one StackOverflow question that looked like the same question, had an answer that wasn't applicable. It appears that both the PagerTitleStrip and the TextView start at 0,0 (left,top) Any help would be appreciated. Note that I can't use any XML (inc Layout.xml) so it's all done programatically. Here's a full working example of my problem: package com.example.projname; import android.app.Activity; import android.os.Bundle; import android.support.v4.view.PagerAdapter; import android.support.v4.view.PagerTabStrip; import android.support.v4.view.ViewPager; import android.view.Gravity; import android.view.View; import android.view.ViewGroup; import android.view.ViewGroup.LayoutParams; import android.widget.TextView; public class MainActivity extends Activity { @Override public void onCreate(Bundle savedInstanceState) { super.onCreate(savedInstanceState); MyPagerAdapter myPagerAdapter = new MyPagerAdapter(); PagerTabStrip myPagerTabStrip = new PagerTabStrip(this); myPagerTabStrip.setGravity(Gravity.TOP); ViewPager viewPager = new ViewPager(this); viewPager.addView(myPagerTabStrip, LayoutParams.MATCH_PARENT, LayoutParams.WRAP_CONTENT); viewPager.setAdapter(myPagerAdapter); setContentView(viewPager); } class MyPagerAdapter extends PagerAdapter { public final String[] Titles = { "Title One", "Title Two", "Title Three", "Title Four", "Title Five" }; @Override public int getCount() { return Titles.length; } @Override public boolean isViewFromObject(View view, Object object) { return (view == object); } @Override public CharSequence getPageTitle(int position) { return Titles[position]; } @Override public Object instantiateItem(ViewGroup container, int position) { TextView textView = new TextView(getApplicationContext()); String myString = new String("Page " + (position + 1) + "\r\n"); textView.setText(myString + myString + myString + myString + myString + myString); container.addView(textView); return textView; } @Override public void destroyItem(ViewGroup container, int position, Object object) { container.removeView((View) object); } } } Edit: Adding a screenshot of the problem:

    Read the article

  • Evaluating points in time by months, but without referencing years in Rails

    - by MikeH
    FYI, There is some overlap in the initial description of this question with a question I asked yesterday, but the question is different. My app has users who have seasonal products. When a user selects a product, we allow him to also select the product's season. We accomplish this by letting him select a start date and an end date for each product. We're using date_select to generate two sets of drop-downs: one for the start date and one for the end date. Including years doesn't make sense for our model. So we're using the option: discard_year => true When you use discard_year => true, Rails sets a year in the database, it just doesn't appear in the views. Rails sets all the years to either 0001 or 0002 in our app. Yes, we could make it 2009 and 2010 or any other pair. But the point is that we want the months and days to function independent of a particular year. If we used 2009 and 2010, then those dates would be wrong next year because we don't expect these records to be updated every year. My problem is that we need to dynamically evaluate the availability of products based on their relationship to the current month. For example, assume it's March 15. Regardless of the year, I need a method that can tell me that a product available from October to January is not available right now. If we were using actual years, this would be pretty easy. For example, in the products model, I can do this: def is_available? (season_start.past? && season_end.future?) end I can also evaluate a start_date and an end_date against current_date However, in setup I've described above where we have arbitrary years that only make sense relative to each other, these methods don't work. For example, is_available? would return false for all my products because their end date is in the year 0001 or 0002. What I need is a method just like the ones I used as examples above, except that they evaluate against current_month instead of current_date, and past? and future months instead of years. I have no idea how to do this or whether Rails has any built in functionality that could help. I've gone through all the date and time methods/helpers in the API docs, but I'm not seeing anything equivalent to what I'm describing. Thanks.

    Read the article

  • iphone calls both if and else at the same time?

    - by David Schiefer
    Hi, I need to determine what file type a file is and then perform a certain action for it. this seems to work fine for some types, however all media types such as videos and sound files get mixed up. I determine the file type by doing this: BOOL matchedMP3 = ([[rowValue pathExtension] isEqualToString:@"mp3"]); if (matchedMP3 == YES) { NSLog(@"Matched MP3"); } I do this for various file types and just define an "else" for all the others. Here's the problem though. The iPhone calls them both. Here's what the log reveals: 2010-05-11 18:51:12.421 Test [5113:207] Matched MP3 2010-05-11 18:51:12.449 Test [5113:207] Matched ELSE I've never seen anything like this before. This is my "matchedMP3" function: BOOL matchedMP3 = ([[rowValue pathExtension] isEqualToString:@"mp3"]); if (matchedMP3 == YES) { NSLog(@"Matched MP3"); NSArray *paths = NSSearchPathForDirectoriesInDomains(NSDocumentDirectory, NSUserDomainMask, YES); NSString *documentsDirectory = [paths objectAtIndex:0]; NSFileManager *manager = [NSFileManager defaultManager]; self.directoryContent = [manager directoryContentsAtPath:documentsDirectory]; NSString *errorMessage = [documentsDirectory stringByAppendingString:@"/"]; NSString *urlAddress = [errorMessage stringByAppendingString:rowValue]; MPMoviePlayerController *moviePlayer = [[MPMoviePlayerController alloc] initWithContentURL:[NSURL fileURLWithPath:urlAddress]]; moviePlayer.movieControlMode = MPMovieControlModeDefault; moviePlayer.backgroundColor = [UIColor blackColor]; [moviePlayer play]; } and here's the else statement: else { NSLog(@"Matched ELSE"); [[NSUserDefaults standardUserDefaults] setObject:rowValue forKey:@"rowValue"]; NSString*rowValue = [[NSUserDefaults standardUserDefaults] objectForKey:@"rowValue"]; NSArray *paths = NSSearchPathForDirectoriesInDomains(NSDocumentDirectory, NSUserDomainMask, YES); NSString *documentsDirectory = [paths objectAtIndex:0]; NSFileManager *manager = [NSFileManager defaultManager]; self.directoryContent = [manager directoryContentsAtPath:documentsDirectory]; NSString *errorMessage = [documentsDirectory stringByAppendingString:@"/"]; NSString *urlAddress = [errorMessage stringByAppendingString:rowValue]; webViewHeader.prompt = rowValue; [documentViewer setDelegate:self]; NSString *encodedString = [urlAddress stringByAddingPercentEscapesUsingEncoding:NSUTF8StringEncoding]; //Create a URL object. NSURL *url = [NSURL URLWithString:encodedString]; //URL Requst Object NSURLRequest *requestObj = [NSURLRequest requestWithURL:url]; //Load the request in the UIWebView. [documentViewer loadRequest:requestObj]; [navigationController pushViewController:webView animated:YES]; } I can't see a reason why it wouldn't work. What happens is that both the webview and the MediaPlayer toggle their own player, so they overlap and play their sound/video a few secs apart from each other. Any help would be appreciated & thank for you taking the time to read through my code.

    Read the article

  • stats::reorder vs Hmisc::reorder

    - by learnr
    I am trying to get around the strange overlap of stats::reorder vs Hmisc::reorder. Without Hmisc loaded I get the result I want, i.e. an unordered factor: > with(InsectSprays, reorder(spray, count, median)) [1] A A A A A A A A A A A A B B B B B B B B B B B B C C C C C C C C C C C C D D [39] D D D D D D D D D D E E E E E E E E E E E E F F F F F F F F F F F F attr(,"scores") A B C D E F 14.0 16.5 1.5 5.0 3.0 15.0 Levels: C E D A F B Now after loading Hmisc the result is an ordered factor: > library(Hmisc) Loading required package: survival Loading required package: splines Attaching package: 'Hmisc' The following object(s) are masked from 'package:survival': untangle.specials The following object(s) are masked from 'package:base': format.pval, round.POSIXt, trunc.POSIXt, units > with(InsectSprays, reorder(spray, count, median)) [1] A A A A A A A A A A A A B B B B B B B B B B B B C C C C C C C C C C C C D D [39] D D D D D D D D D D E E E E E E E E E E E E F F F F F F F F F F F F Levels: C < E < D < A < F < B In calling stats::reorder directly, I now for some reason get an ordered factor. > with(InsectSprays, stats::reorder(spray, count, median)) [1] A A A A A A A A A A A A B B B B B B B B B B B B C C C C C C C C C C C C D D [39] D D D D D D D D D D E E E E E E E E E E E E F F F F F F F F F F F F Levels: C < E < D < A < F < B Specifying, that I would need an unordered factor results in an error suggesting that stats::reorder is not used? > with(InsectSprays, stats::reorder(spray, count, median, order = FALSE)) Error in FUN(X[[1L]], ...) : unused argument(s) (order = FALSE) So the question really is how do I get an unordered factor with Hmisc loaded?

    Read the article

  • Complex relationship between tables in NHibernate

    - by Ilya Kogan
    Hi all, I'm writing a Fluent NHibernate mapping for a legacy Oracle database. The challenge is that the tables have composite primary keys. If I were at total freedom, I would redesign the relationships and auto-generate primary keys, but other applications must write to the same database and read from it, so I cannot do it. These are the two tables I'll focus on: Example data Trips table: 1, 10:00, 11:00 ... 1, 12:00, 15:00 ... 1, 16:00, 19:00 ... 2, 12:00, 13:00 ... 3, 9:00, 18:00 ... Faults table: 1, 13:00 ... 1, 23:00 ... 2, 12:30 ... In this case, vehicle 1 made three trips and has two faults. The first fault happened during the second trip, and the second fault happened while the vehicle was resting. Vehicle 2 had one trip, during which a fault happened. Constraints Trips of the same vehicle never overlap. So the tables have an optional one-to-many relationship, because every fault either happens during a trip or it doesn't. If I wanted to join them in SQL, I would write: select ... from Faults left outer join Trips on Faults.VehicleId = Trips.VehicleId and Faults.FaultTime between Trips.TripStartTime and Trips.TripEndTime and then I'd get a dataset where every fault appears exactly once (one-to-many as I said). Note that there is no Vehicles table, and I don't need one. But I did create a view that contains all VehicleIds from both tables, so I can use it as a junction table. What am I actually looking for? The tables are huge because they cover years of data, and every time I only need to fetch a range of a few hours. So I need a mapping and a criteria that will run something like the following SQL underneath: select ... from Faults left outer join Trips on Faults.VehicleId = Trips.VehicleId and Faults.FaultTime between Trips.TripStartTime and Trips.TripEndTime where Faults.FaultTime between :p0 and :p1 Do you have any ideas how to achieve it? Note 1: Currently the application shouldn't write to the database, so persistence is not a must, although if the mapping supports persistence, it may help at some point in the future. Note 2: I know it's a tough one, so if you give me a great answer, you will be properly rewarded :) Thank you for reading this long question, and now I only hope for the best :)

    Read the article

  • Opinion on "loop invariants", and are these frequently used in the industry?

    - by Michael Aaron Safyan
    I was thinking back to my freshman year at college (five years ago) when I took an exam to place-out of intro-level computer science. There was a question about loop invariants, and I was wondering if loop invariants are really necessary in this case or if the question was simply a bad example... the question was to write an iterative definition for a factorial function, and then to prove that the function was correct. The code that I provided for the factorial function was as follows: public static int factorial(int x) { if ( x < 0 ){ throw new IllegalArgumentException("Parameter must be = 0"); }else if ( x == 0 ){ return 1; }else{ int result = 1; for ( int i = 1; i <= x; i++ ){ result*=i; } return result; } } My own proof of correctness was a proof by cases, and in each I asserted that it was correct by definition (x! is undefined for negative values, 0! is 1, and x! is 1*2*3...*x for a positive value of x). The professor wanted me to prove the loop using a loop invariant; however, my argument was that it was correct "by definition", because the definition of "x!" for a positive integer x is "the product of the integers from 1... x", and the for-loop in the else clause is simply a literal translation of this definition. Is a loop invariant really needed as a proof of correctness in this case? How complicated must a loop be before a loop invariant (and proper initialization and termination conditions) become necessary for a proof of correctness? Additionally, I was wondering... how often are such formal proofs used in the industry? I have found that about half of my courses are very theoretical and proof-heavy and about half are very implementation and coding-heavy, without any formal or theoretical material. How much do these overlap in practice? If you do use proofs in the industry, when do you apply them (always, only if it's complicated, rarely, never)?

    Read the article

  • Is there a scheduling algorithm that optimizes for "maker's schedules"?

    - by John Feminella
    You may be familiar with Paul Graham's essay, "Maker's Schedule, Manager's Schedule". The crux of the essay is that for creative and technical professionals, meetings are anathema to productivity, because they tend to lead to "schedule fragmentation", breaking up free time into chunks that are too small to acquire the focus needed to solve difficult problems. In my firm we've seen significant benefits by minimizing the amount of disruption caused, but the brute-force algorithm we use to decide schedules is not sophisticated enough to handle scheduling large groups of people well. (*) What I'm looking for is if there's are any well-known algorithms which minimize this productivity disruption, among a group of N makers and managers. In our model, There are N people. Each person pi is either a maker (Mk) or a manager (Mg). Each person has a schedule si. Everyone's schedule is H hours long. A schedule consists of a series of non-overlapping intervals si = [h1, ..., hj]. An interval is either free or busy. Two adjacent free intervals are equivalent to a single free interval that spans both. A maker's productivity is maximized when the number of free intervals is minimized. A manager's productivity is maximized when the total length of free intervals is maximized. Notice that if there are no meetings, both the makers and the managers experience optimum productivity. If meetings must be scheduled, then makers prefer that meetings happen back-to-back, while managers don't care where the meeting goes. Note that because all disruptions are treated as equally harmful to makers, there's no difference between a meeting that lasts 1 second and a meeting that lasts 3 hours if it segments the available free time. The problem is to decide how to schedule M different meetings involving arbitrary numbers of the N people, where each person in a given meeting must place a busy interval into their schedule such that it doesn't overlap with any other busy interval. For each meeting Mt the start time for the busy interval must be the same for all parties. Does an algorithm exist to solve this problem or one similar to it? My first thought was that this looks really similar to defragmentation (minimize number of distinct chunks), and there are a lot of algorithms about that. But defragmentation doesn't have much to do with scheduling. Thoughts? (*) Practically speaking this is not really a problem, because it's rare that we have meetings with more than ~5 people at once, so the space of possibilities is small.

    Read the article

  • Flatten date range memberships retaining only the highest priority membership (TRANSACT-SQL)

    - by shadowranger
    Problem statement: A table contains an item_id, a category_id and a date range (begin_date and end_date). No item may be in more than one category on any given date (in general; during daily rebuilding it can be in an invalid state temporarily). By default, all items are added (and re-added if removed) to a category (derived from outside data) automatically on a daily basis, and their membership in that category matches the lifespan of the item (items have their own begin and end date, and usually spend their entire lives in the same category, which is why this matches). For items in category X, it is occasionally desirable to override the default category by adding them to category Y. Membership in category Y could entirely replace membership in category X (that is, the begin and end dates for membership in category Y would match the begin and end dates of the item itself), or it could override it for an arbitrary period of time (at the beginning, middle or end the item's lifespan, possibly overriding for short periods at multiple times). Membership in category Y is not renewed automatically and additions to that category is done by manual data entry. Every day, when category X is rebuilt, we get an overlap, where any item in category Y will now be in category X as well (which is forbidden, as noted previously). Goal: After each repopulation of category X (which is done in a rather complicated and fragile manner, and ideally would be left alone), I'm trying to find an efficient means of writing a stored procedure that: Identifies the overlaps Changes existing entries, adds new ones where necessary (such as in the case where an item starts in category X, switches to category Y, then eventually switches back to category X before ending), or removes entries (when an item is in category Y for its entire life) such that every item remains in category Y (and only Y) where specified, while category X membership is maintained when not overridden by category Y. Does not affect memberships of categories A, B, C, Z, etc., which do not have override categories and are built separately, by completely different rules. Note: It can be assumed that X membership covers the entire lifespan of the item before this procedure is called, so it is unnecessary to query any data outside this table. Bonus credit: If for some reason there are two adjacent or overlapping memberships in for the same item in category Y, stitching them together into a single entry is appreciated, but not necessary. Example: item_id category_id begin_date end_date 1 X 20080101 20090628 1 Y 20090101 20090131 1 Y 20090601 20090628 2 X 20080201 20080731 2 Y 20080201 20080731 Should become: item_id category_id begin_date end_date 1 X 20080101 20081231 1 Y 20090101 20090131 1 X 20090201 20090531 1 Y 20090601 20090628 2 Y 20080201 20080731 If it matters, this needs to work on SQL Server 2005 and SQL Server 2008

    Read the article

< Previous Page | 16 17 18 19 20 21 22  | Next Page >