Search Results

Search found 3486 results on 140 pages for 'chris b behrens'.

Page 94/140 | < Previous Page | 90 91 92 93 94 95 96 97 98 99 100 101  | Next Page >

  • Disable button after submit

    - by Chris Philpotts
    I'm trying to disable a button when a user submits a payment form and the code to post the form is causing a double post in firefox. This problem does not occur when the code is removed, and does not occur in any browser other than firefox. Any idea how to prevent the double post here? System.Text.StringBuilder sb = new StringBuilder(); sb.Append("if (typeof(Page_ClientValidate) == 'function') { "); sb.Append("if (Page_ClientValidate() == false) { return false; }} "); sb.Append("this.value = 'Please wait...';"); sb.Append("this.disabled = true;"); sb.Append(Page.GetPostBackEventReference(btnSubmit )); sb.Append(";"); btnSubmit.Attributes.Add("onclick", sb.ToString()); it's the sb.Append(Page.GetPostBackEventReference(btnSubmit )) line that's causing the issue Thanks EDIT: Here's the c# of the button: <asp:Button ID="cmdSubmit" runat="server" Text="Submit" /> here's the html This code posts twice (and disables the submit button and verifies input): <input type="submit" name="ctl00$MainContent$cmdSubmit" value="Submit" onclick="if (typeof(Page_ClientValidate) == 'function') { if (Page_ClientValidate() == false) { return false; }} this.value = 'Please wait...';this.disabled = true;document.getElementById('ctl00_MainContent_cmdBack').disabled = true;__doPostBack('ctl00$MainContent$cmdSubmit','');" id="ctl00_MainContent_cmdSubmit" /> This code posts twice (but doesn’t disable the submit button): <input type="submit" name="ctl00$MainContent$cmdSubmit" value="Submit" onclick="__doPostBack('ctl00$MainContent$cmdSubmit','');" id="ctl00_MainContent_cmdSubmit" /> This code posts once (but doesn’t verify the user input and doesn’t disable the submit button): <input type="submit" name="ctl00$MainContent$cmdSubmit" value="Submit" id="ctl00_MainContent_cmdSubmit" /> This code posts once (but doesn’t disable submit button): <input type="submit" name="ctl00$MainContent$cmdSubmit" value="Submit" onclick="javascript:WebForm_DoPostBackWithOptions(new WebForm_PostBackOptions(&quot;ctl00$MainContent$cmdSubmit&quot;, &quot;&quot;, true, &quot;&quot;, &quot;&quot;, false, false))" id="ctl00_MainContent_cmdSubmit" /> This code doesn’t post at all: <input type="submit" name="ctl00$MainContent$cmdSubmit" value="Submit" onclick="this.disabled = true;WebForm_DoPostBackWithOptions(new WebForm_PostBackOptions(&quot;ctl00$MainContent$cmdSubmit&quot;, &quot;&quot;, true, &quot;&quot;, &quot;&quot;, false, false))" id="ctl00_MainContent_cmdSubmit" /> Obviously it’s the disabling of the submit button that’s posing the problem. Do you have any ideas how we can disable the submit to avoid multiple clicking?

    Read the article

  • What does xp_qv do in SQL Server?

    - by Chris
    Last night one of our SQL servers developed some major problems and after a colleague stopped, started, and all the usual things it started checking and rebuilding databases and is now running an extended stored procedure called "xp_qv". The internet seems to be very short of information on what this procedure does or anythign like that so I was hoping somebody here might be able to help. I should add that I assume it is meant to be running so the question isn't "Can I stop it" or anything like that, its just curiosity in what it is doing in the hope that it will help determine how long before things are usable again...

    Read the article

  • Choosing a type for search results in C#

    - by Chris M
    I have a result set that will never exceed 500; the results that come back from the web-service are assigned to a search results object. The data from the webservice is about 2mb; the bit I want to use is about a third of each record, so this allows me to cache and quickly manipulate it. I want to be able to sort and filter the results with the least amount of overhead and as fast as possible so I used the VCSKICKS timing class to measure their performance Average Total (10,000) Type Create Sort Create Sort HashSet 0.1579 0.0003 1579 3 IList 0.0633 0.0002 633 2 IQueryable 0.0072 0.0432 72 432 Measured in Seconds using http://www.vcskicks.com/algorithm-performance.php I created the hashset through a for loop over the web-service response (adding to the hashset). The List & IQueryable were created using LINQ. Question I can understand why HashSet takes longer to create (the foreach loop vs linq); but why would IQueryable take longer to sort than the other two; and finally is there a better way to assign the HashSet. Thanks Actual Program public class Program { private static AuthenticationHeader _authHeader; private static OPSoapClient _opSession; private static AccommodationSearchResponse _searchResults; private static HashSet<SearchResults> _myHash; private static IList<SearchResults> _myList; private static IQueryable<SearchResults> _myIQuery; static void Main(string[] args) { #region Setup WebService _authHeader = new AuthenticationHeader { UserName = "xx", Password = "xx" }; _opSession = new OPSoapClient(); #region Setup Search Results _searchResults = _opgSession.SearchCR(_authHeader, "ENG", "GBP", "GBR"); #endregion Setup Search Results #endregion Setup WebService // HASHSET SpeedTester hashTest = new SpeedTester(TestHashSet); hashTest.RunTest(); Console.WriteLine("- Hash Test \nAverage Running Time: {0}; Total Time: {1}", hashTest.AverageRunningTime, hashTest.TotalRunningTime); SpeedTester hashSortTest = new SpeedTester(TestSortingHashSet); hashSortTest.RunTest(); Console.WriteLine("- Hash Sort Test \nAverage Running Time: {0}; Total Time: {1}", hashSortTest.AverageRunningTime, hashSortTest.TotalRunningTime); // ILIST SpeedTester listTest = new SpeedTester(TestList); listTest.RunTest(); Console.WriteLine("- List Test \nAverage Running Time: {0}; Total Time: {1}", listTest.AverageRunningTime, listTest.TotalRunningTime); SpeedTester listSortTest = new SpeedTester(TestSortingList); listSortTest.RunTest(); Console.WriteLine("- List Sort Test \nAverage Running Time: {0}; Total Time: {1}", listSortTest.AverageRunningTime, listSortTest.TotalRunningTime); // IQUERIABLE SpeedTester iqueryTest = new SpeedTester(TestIQueriable); iqueryTest.RunTest(); Console.WriteLine("- iquery Test \nAverage Running Time: {0}; Total Time: {1}", iqueryTest.AverageRunningTime, iqueryTest.TotalRunningTime); SpeedTester iquerySortTest = new SpeedTester(TestSortableIQueriable); iquerySortTest.RunTest(); Console.WriteLine("- iquery Sort Test \nAverage Running Time: {0}; Total Time: {1}", iquerySortTest.AverageRunningTime, iquerySortTest.TotalRunningTime); } static void TestHashSet() { var test = _searchResults.Items; _myHash = new HashSet<SearchResults>(); foreach(var x in test) { _myHash.Add(new SearchResults { Ref = x.Ref, Price = x.StandardPrice }); } } static void TestSortingHashSet() { var sorted = _myHash.OrderBy(s => s.Price); } static void TestList() { var test = _searchResults.Items; _myList = (from x in test select new SearchResults { Ref = x.Ref, Price = x.StandardPrice }).ToList(); } static void TestSortingList() { var sorted = _myList.OrderBy(s => s.Price); } static void TestIQueriable() { var test = _searchResults.Items; _myIQuery = (from x in test select new SearchResults { Ref = x.Ref, Price = x.StandardPrice }).AsQueryable(); } static void TestSortableIQueriable() { var sorted = _myIQuery.OrderBy(s => s.Price); } }

    Read the article

  • How do you use Canadian currency?

    - by chris
    When I pass an option of :currency = "CAD" to the setup_purchase and purchase methods the transaction still goes through in US funds. My paypal account does have CAD as the default currency. What am I missing?

    Read the article

  • What is the worst class/variable/function name you have ever encountered

    - by Chris Noe
    Naming things well is arguably Job 1 for professional programmers. Yet we have all suffered from some bad naming choices from time to time. So just to vent a little, what are some doozies that you may have run across? Just to get things started: One of our original developers wasn't sure what to call a secondary key - on what turned out to be a primary table for this app - so he called it: DL2WhateverTheHellThatIs. Unfortunately this system generates entity mappings from the XML, and attributes defined there result in classes, methods, and constants that are referenced through-out the app. To this day it is very hard to find a source file that does not reference this, er, thing! A few actual examples: DL2WhateverTheHellThatIsBean cos = (DL2WhateverTheHellThatIsBean)itr.next(); String code = getDL2WhateverTheHellThatIs().getCode(); From from = new From("DL2WhateverTheHellThatIs"); String filter = "_dL2WhateverTheHellThatIs._code"; (Very difficult to refactor)

    Read the article

  • SSRS 2008 Need to lookup customer name with largest order

    - by Chris
    Hi, I'm creating an SSRS report which contains a table of orders, grouped by day. Now I can easily get the max order value for the day and put it in the group header by using the SSRS MAX() function. However, I also want to get the corresponding customer name who placed this order, and place this in the group header too. We can assume my result set simply contains date, name and order value. Is there any way to do this in SSRS 2008? Thanks

    Read the article

  • Do controls need to be defined in a web app or will .NET do it for you

    - by Chris
    I always thought that when you dropped a control onto an .aspx page that a declaration of that control ended up being generated for you (either in a designer file, or within your code behind). All of the apps I have worked on have worked this way. A coworker of mine was installing resharper and it was showing that all her code behind pages would not build. Turned out that resharper could not find a definition for any control that she has dropped onto her markup. She has no designer files, and no declarations in markup. Do they get automatically built when putting together the partial classes? Is there an option at that page/project level to instruct .NET to do this for you? Is this a difference between web app and web site? This is a a .NET 3.5 site, C#, and it is running in a production environment. Thanks in advance

    Read the article

  • How do I start using Linux for web development?

    - by Chris Maple
    OK, so that's maybe not the best title, but I don't know exactly what I want to do, so, please, hear me out. I've used Windows pretty much all my life although I played with Linux on several occasions. At work everyone does web development with php on Windows using the same IDE and stuff. I would like to experience the powerful Linux command line as well as test my web apps locally running the Linux version of php, however I prefer to keep using the IDE because I feel that version control, FTP, code completion etc. are helpful to my productivity. Eventually I would also like to dive into Ruby and Python and I hear that they're not really suited for Windows users. Is running a Linux distro in a virtual machine my best bet? Should I try something like Wubi (Ubuntu running inside Windows)? Or are there any other options out there?

    Read the article

  • Can Perforce and SourceSafe co-exist in Visual Studio?

    - by Chris
    Visual Studio 2008, to be more specific. We're testing out moving to Perforce for source control, so I'd like to install the P4SCC plugin to monkey around with. However, I'd also like to continue using SourceSafe's IDE capabilities for projects that haven't been moved over yet. Can the two co-exist peacefully, or is it one or the other for a specific install of VS?

    Read the article

  • e4x filter on more then on childeren?

    - by Chris
    My XML Looks like this: <?xml version="1.0" encoding="utf-8" ?> <projects> <project id="1" thumb="media/images/thumb.jpg" > <categories> <id>1</id> <id>2</id> </categories> <director>Director name</director> <name><![CDATA[IPhone commercial]]></name> <url><![CDATA[http://www.iphone.com]]></url> <description><![CDATA[Description about the project]]></description> <thumb><![CDATA[/upload/images/thumb.jpg]]></thumb> </project> </projects> But I cannot figure out how to filter projects based on a category id? Does anybody know how to do ? :)

    Read the article

  • Best way to detect similar email addresses?

    - by Chris
    I have a list of ~20,000 email addresses, some of which I know to be fraudulent attempts to get around a "1 per e-mail" limit. ([email protected], [email protected], [email protected], etc...). I want to find similar email addresses for evaluation. Currently I'm using a levenshtein algorithm to check each e-mail against the others in the list and report any with an edit distance of less than 2. However, this is painstakingly slow. Is there a more efficient approach? The test code I'm using now is: using System; using System.Collections.Generic; using System.Linq; using System.Text; using System.IO; using System.Threading; namespace LevenshteinAnalyzer { class Program { const string INPUT_FILE = @"C:\Input.txt"; const string OUTPUT_FILE = @"C:\Output.txt"; static void Main(string[] args) { var inputWords = File.ReadAllLines(INPUT_FILE); var outputWords = new SortedSet<string>(); for (var i = 0; i < inputWords.Length; i++) { if (i % 100 == 0) Console.WriteLine("Processing record #" + i); var word1 = inputWords[i].ToLower(); for (var n = i + 1; n < inputWords.Length; n++) { if (i == n) continue; var word2 = inputWords[n].ToLower(); if (word1 == word2) continue; if (outputWords.Contains(word1)) continue; if (outputWords.Contains(word2)) continue; var distance = LevenshteinAlgorithm.Compute(word1, word2); if (distance <= 2) { outputWords.Add(word1); outputWords.Add(word2); } } } File.WriteAllLines(OUTPUT_FILE, outputWords.ToArray()); Console.WriteLine("Found {0} words", outputWords.Count); } } }

    Read the article

  • Need an app config file opinion here.....

    - by Chris
    I know that app config is used to provide environmental type of values for process variables. And it is very much key/value oriented. But I have a process (windows service) that uses a skeleton xml template file to produce a fuller version of that file after processesing. What about the idea of including that 'skeleton xml' in the app config file? I am trying to aviod having a standalone xml skeleton file, since it is 'configuration' related, in a way.

    Read the article

  • Trigger ad-hoc activity within a workflow

    - by Chris Taylor
    I am looking to use WF 4 to replace an existing workflow solution we have. One feature that is currently used in the existing workflow engine is the ability to cancel a current activity and loopback to a FlowSwitch type activity. So given the following crude workflow where we start at 'O' and base in the input data the workflow follows the path to 'A2' which is currently blocking on s bookmark waiting for input. ---------A1--\ | \ /\ \ O------- ---->--(A2)-------| ^ \/ / | | | / | | ---------A3--/ | | | |----------------------| However in the meantime some out of band data comes in that means we should cancel 'A2' and return to the FlowSwitch to re-evaluate based on the new data. The question is what is the best way to handle the out of band data that arrived? My initial guess is to have a Parallel activity with one branch waiting for out of band data and the other branch containing the workflow sequence described above. If data came in on the brach waiting for the out of band data, how would I cancel the current activity in the workflow and force it to return to the FlowSwitch. Or of course is there a better way to handle this. I have not actually done any work with the WF4 stuff for WF3 for that matter so I might be missing something obvious here.

    Read the article

  • JavaScript Encapsulation / JQuery

    - by Chris
    I am trying to figure out how to keep my page variables in my application from being defined globally. I've come up with a few methods but am wondering if there is a general standard approach people use. I've got my plugin design pattern down using this approach: http://www.virgentech.com/blog/2009/10/building-object-oriented-jquery-plugin.html. But I'm just not sure how to handle my page level encapsulation.

    Read the article

  • Copy NSAttributedString to pasteboard

    - by Chris
    Brand new to Cocoa and I'm trying to figure out how to copy an NSAttributedString to the pasteboard. I've looked in the docs and not sure if I'm supposed to use a NSPasteboardItem or not. Here's what I have to copy a regular NSString: NSPasteboard *pb = [NSPasteboard generalPasteboard]; NSArray *types = [NSArray arrayWithObjects:NSStringPboardType, nil]; [pb declareTypes:types owner:self]; [pb setString:@"asdfasdf" forType:NSStringPboardType]; How do I set a NSAttributedString? Thanks

    Read the article

  • How to have abstract and overriding constants in C# ?

    - by Chris
    Hi all, My code below won't compile. What am i doing wrong? I'm basically trying to have a public constant that is overridden in the base class. public abstract class MyBaseClass { public abstract const string bank = "???"; } public class SomeBankClass : MyBaseClass { public override const string bank = "Some Bank"; } Thanks as always for being so helpful!

    Read the article

  • How could I apply a genetic algorithm to a simple game that follows rollercoaster tracks?

    - by Chris
    I have free-rein over what I do on a final assignment for school, with respect to modifying a simple direct-x game that currently just has the camera follow some rollercoaster rails. I've developed an interest in genetic algorithms and would like to take this opportunity to apply one and learn something about them. However, I can't think of any way I could possibly apply one in this case. What are some options available to me?

    Read the article

  • Code to show UIPickerview under clicked UITextField

    - by Chris F
    I thought I'd share a code snippet where I show a UIPickerView when you click a UITextField. The code uses a UIPickerView, but there's no reason to use a different view controller, like a UITableViewController that uses a table instead of a picker. Just create a single-view project with a nib, and add a UITextField to the view and make you connections in IB. // .h file #import @interface MyPickerViewViewController : UIViewController <UIPickerViewDelegate, UIPickerViewDataSource, UITextFieldDelegate> - (IBAction)dismissPickerView:(id)sender; @end // .m file #import "MyPickerViewViewController.h" @interface MyPickerViewViewController () { UIPickerView *_pv; NSArray *_array; IBOutlet __weak UITextField *_tf; BOOL _pickerViewShown; } @end @implementation MyPickerViewViewController - (void)viewDidLoad { [super viewDidLoad]; _pickerViewShown = NO; _array = [NSArray arrayWithObjects:@"One", @"Two", @"Three", @"Four", nil]; _pv = [[UIPickerView alloc] initWithFrame:CGRectZero]; _pv.showsSelectionIndicator = YES; _pv.dataSource = self; _pv.delegate = self; _tf.delegate = self; _tf.inputView = _pv; } - (IBAction)dismissPickerView:(id)sender { [_pv removeFromSuperview]; [_tf.inputView removeFromSuperview]; [_tf resignFirstResponder]; _pickerViewShown = NO; } - (void)didReceiveMemoryWarning { [super didReceiveMemoryWarning]; // Dispose of any resources that can be recreated. } - (BOOL)textFieldShouldBeginEditing:(UITextField *)textField { if (!_pickerViewShown) { [self setRectForPickerViewRelativeToTextField:textField]; [self.view addSubview:_tf.inputView]; _pickerViewShown = YES; } else { [self dismissPickerView:self]; } return NO; } - (void)setRectForPickerViewRelativeToTextField:(UITextField*)textField { CGFloat xPos = textField.frame.origin.x; CGFloat yPos = textField.frame.origin.y; CGFloat width = textField.frame.size.width; CGFloat height = textField.frame.size.height; CGFloat pvHeight = _pv.frame.size.height; CGRect pvRect = CGRectMake(xPos, yPos+height, width, pvHeight); _pv.frame = pvRect; } - (NSString *)pickerView:(UIPickerView *)pickerView titleForRow:(NSInteger)row forComponent:(NSInteger)component { return [_array objectAtIndex:row]; } - (NSInteger)numberOfComponentsInPickerView:(UIPickerView *)pickerView { return 1; } - (NSInteger)pickerView:(UIPickerView *)pickerView numberOfRowsInComponent:(NSInteger)component { return _array.count; } - (void) pickerView:(UIPickerView *)pickerView didSelectRow:(NSInteger)row inComponent:(NSInteger)component { _tf.text = [_array objectAtIndex:row]; [self dismissPickerView:self]; } @end

    Read the article

  • testing .mobile mime format with capybara / rspec

    - by Chris Beck
    For detecting and responding to mobile user agents, I'm using Mime::Type.register_alias "text/html", :mobile and the approach I'm wondering what is the best approach to test with capybara. This article suggests setting up an iphone driver with Capybara.register_driver :iphone do |app| http://blog.plataformatec.com.br/2011/03/configuring-user-agents-with-capybara-selenium-webdriver/ but I'd like a more flexible approach where the mime type is set via the url extension localhost/index.mobile and where I can do this visit user_path( format: :mobile) Rails understands the extension and sets the format in the params hash, but how do I get the url helper methods to add that to all urls as a file extension?

    Read the article

  • Defining - and dealing with - Evil

    - by Chris Becke
    As a software developer one sometimes gets feature requests that seem to be in some kind of morally grey area. Sometimes one can deflect them, or implement them in a way that feels less 'evil' - sometimes - on reflection - while the feature request 'feels' wrong theres no identifiable part of it that actually causes harm. Sometimes one feels a feature is totally innocent but various anti virus products start tagging one as malware. For example - I personally consider EULAs to (a) hopefully be unenforceable and (b) a means by which rights are REMOVED from consumers. However Anti Virus scanners frequently mark as malware any kind of download agent that does not display a EULA. Which to me is the result of a curious kind of double think. What I want to know is - are there any online (or offline) resources that cover evil software development practices? How can I know if a software practice that I consider dodgy is in fact evil enough to consider fighting?

    Read the article

  • LGPL library with plugins of varied licenses

    - by Chris
    Note: "Plugins" here refers to shared objects that are accessed via dlopen() and friends. I'm writing a library that I'm planning on releasing under the LGPL. Its functionality can be extended (supporting new audio file formats, specifically) through plugins. I'm planning on creating an exception to the LGPL for this library so that plugins can be released under any license. So far so good. I've written a number of plugins already, some of which use LGPL and some of which use GPL libraries. I'm wary of releasing them with the main library, however, due to licensing issues. The LGPL-based ones would generally be fine, but for my "any license" clause. Would distributing these LGPL-based plugins with the library require the consent of the other license holders to create this exception? Along the same lines, would the inclusion of GPL-based plugins with my library force the whole thing to go GPL? I could also release the plugins separately. The advantage, I presume, is that the plugins an d library will now not be distributed together, creating more separation. But this seems to be no different, really, in the end. Boiled down: Can I include, with my LGPL library, plugins of varied licenses? If not, is it really any different releasing them separately? And if so, there's no real need to create an exception for non-LGPL plugins, is there? It's LGPL or nothing. I'd prefer asking a lawyer, of course, but this is just a hobby and I can't afford to hire a lawyer when I don't expect or want monetary compensation. I'm just hoping others have been in similar situations and have insight.

    Read the article

  • Valgrind 'noise', what does it mean?

    - by Chris Huang-Leaver
    When I used valgrind to help debug an app I was working on I notice a huge about of noise which seems to be complaining about standard libraries. As a test I did this; echo 'int main() {return 0;}' | gcc -x c -o test - Then I did this; valgrind ./test ==1096== Use of uninitialised value of size 8 ==1096== at 0x400A202: _dl_new_object (in /lib64/ld-2.10.1.so) ==1096== by 0x400607F: _dl_map_object_from_fd (in /lib64/ld-2.10.1.so) ==1096== by 0x4007A2C: _dl_map_object (in /lib64/ld-2.10.1.so) ==1096== by 0x400199A: map_doit (in /lib64/ld-2.10.1.so) ==1096== by 0x400D495: _dl_catch_error (in /lib64/ld-2.10.1.so) ==1096== by 0x400189E: do_preload (in /lib64/ld-2.10.1.so) ==1096== by 0x4003CCD: dl_main (in /lib64/ld-2.10.1.so) ==1096== by 0x401404B: _dl_sysdep_start (in /lib64/ld-2.10.1.so) ==1096== by 0x4001471: _dl_start (in /lib64/ld-2.10.1.so) ==1096== by 0x4000BA7: (within /lib64/ld-2.10.1.so) * large block of similar snipped * ==1096== Use of uninitialised value of size 8 ==1096== at 0x4F35FDD: (within /lib64/libc-2.10.1.so) ==1096== by 0x4F35B11: (within /lib64/libc-2.10.1.so) ==1096== by 0x4A1E61C: _vgnU_freeres (vg_preloaded.c:60) ==1096== by 0x4E5F2E4: __run_exit_handlers (in /lib64/libc-2.10.1.so) ==1096== by 0x4E5F354: exit (in /lib64/libc-2.10.1.so) ==1096== by 0x4E48A2C: (below main) (in /lib64/libc-2.10.1.so) ==1096== ==1096== ERROR SUMMARY: 3819 errors from 298 contexts (suppressed: 876 from 4) ==1096== malloc/free: in use at exit: 0 bytes in 0 blocks. ==1096== malloc/free: 0 allocs, 0 frees, 0 bytes allocated. ==1096== For counts of detected errors, rerun with: -v ==1096== Use --track-origins=yes to see where uninitialised values come from ==1096== All heap blocks were freed -- no leaks are possible. You can see the full result here: http://pastebin.com/gcTN8xGp I have two questions; firstly is there a way to suppress all the noise? --show-below-main is set to no by default, but there doesn't appear to be a --show-after-main equivalent.

    Read the article

  • Is it necessary to create ASP.NET 4.0 SQL session state database, distinct from existing ASP.NET 2.0

    - by Chris W. Rea
    Is the ASP.NET 4.0 SQL session state mechanism backward-compatible with the ASP.NET 2.0 schema for session state, or should/must we create a separate and distinct session state database for our ASP.NET 4.0 apps? I'm leaning towards the latter anyway, but the 2.0 database seems to just work, though I'm wondering if there are any substantive differences between the ASPState database schema / procedures between the 2.0 and 4.0 versions of ASP.NET. Thank you.

    Read the article

< Previous Page | 90 91 92 93 94 95 96 97 98 99 100 101  | Next Page >