Rails 5: Send reload to set of clients on data update - ruby-on-rails

I'm using Rails 5 to make a simple turn based game tracker for an in-person social game (via phones/tablets/etc..)
I want to have all the 'players' in the game (list of sessions/users/...) to reload their browsers automatically once a player has taken an action.
I know that there are live update capabilities such as AJAX and websockets, but they all seem far too weighty for what seems to be a simple problem. Furthermore, I want to update other clients pages, not just the client initiating the action.
Is there a simple solution to send a reload? Or do I need to code something up in one of the more complicated APIs?

For the simple trouble, you still can use AJAX to reload user client by making interval request for each XX seconds. The server can return the last action time which can be used for client to determine that it should reload itself or not.
For example, on the controller
# SomeController
def get_last_action_time
# Return the timestamp of the last action
render json: {last_action_time: "2017-12-29 10:00:42 UTC"}
end
on the client
function getLocalLastAction () {
/* return timestamp of the last action on local */
}
function setLocalLastAction (time) {
/* Store the `time` to somewhere, ex: localStorage */
}
function checkLastAction () {
$.getJSON("/get_last_action_time", function (data) {
if (getLocalLastAction() < data.last_action_time) {
/* destroy the interval */
setLocalLastAction(data.last_action_time)
/* do the reload page */
} else {
/* do nothing */
}
})
}
// Check every 1 second, shouldn't be too short due to performance
var checking = setInterval(checkLastAction, 1000)
Then when user A do an action, the server last_action_time will change, hence client of other users will be reloaded at most after 1 second.
This way is old but quite easy to do in some simple case, and when you implement together with actions caching, the performance of app still acceptable. In the more complicated cases, I suggest using WebSocket solution for
Full control
Low latency
Better performance for app

Thanks to #yeuem1vannam's answer, here is the final code I used that helps avoid the race condition of a page loading old information while the time is being updated and then the javascript updating the time and getting the new time, and hence missing the reload.
The javascript code:
var actionChecker;
function doneChecking () {
clearInterval(actionChecker);
}
function checkLastAction () {
// Get the game ID from the html access span
var dataId = document.getElementById('javascript_data_access');
if (!dataId) return doneChecking();
var initActionTime = dataId.getAttribute('init_last_action_time');
if (!initActionTime) return doneChecking();
dataId = dataId.getAttribute('game_number');
if (!dataId) return doneChecking();
// Get the last action time
var ret = $.getJSON("/get_last_action_time/"+dataId, function (data) {
var lastActionTime = data.last_action_time;
if (!lastActionTime) return doneChecking();
if (lastActionTime>initActionTime) {
location.reload();
}
})
}
window.onload = function() {
// Check every 1 second, shouldn't be too short due to performance
actionChecker = setInterval(checkLastAction, 1000);
}
The controller's action:
def get_last_action_time
last_time = nil
begin
#game = Game.find_by_id(params[:id])
# Return the timestamp of the last action
last_time = (#game && !#game.endTime) ? #game.reloadTime.to_i : 0
rescue ActiveRecord::RecordNotFound
last_time = 0
end
# Stop bugging us after 30m, we should have moved on from this page
last_time==0 if (last_time!=0 && (milliseconds - last_time)>30*60*1000)
render json: {last_action_time: last_time}
end
And then in the html.erb:
<span id='javascript_data_access' game_number=<%= params[:id] %> init_last_action_time=<%= #game.reloadTime %>></span>
Obviously you need to add reloadTime to your model and also endTime if there's a time you no longer want to check for reloads anymore.
Seems to be working fine so far, you have to make sure that you're careful about who is in charge of setting reloadTime. If two pages set reloadTime everytime they reload, you'll be stuck in a reload loop battle between the two pages.

Related

How to cancel all running sagas when navigating between page views

I'm trying to find a simple and easy way to cancel all running sagas within a "page" when the user decides to navigate to another "page" within the app... We are not using routing, but instead each "page" is its own widget within a larger host application that is responsible for creating and loading each page when the user navigates...
Currently, we are using redux-saga and have setup logic like so (simplified for brevity) when a page widget is created and loaded...
// page-sagas
export function* rootSaga() {
const allSagas = [
// ... all sagas used by page (example) ...
// function* watchFoo() {
// yield takeEvery(FooAction, foo);
// }
];
yield all(allSagas.map((saga) => call(saga)));
}
// page-widget
onLoad = () => {
const sagaMiddleware = createSagaMiddleware();
const store = createStore(reducer, initState, applyMiddlware(sagaMiddleware));
sagaMiddleware.run(rootSaga);
}
Ideally, I'd prefer to avoid having to add forking logic to every single saga in every single page-widget, and looking at the Redux-Saga Task API, it says you can cancel a task returned by the call to middleware.run, but I'm wondering if this propagates down to all nested / child sagas that are currently in progress, or if there are any issues / gotcha's I should be aware of:
Example:
// page-widget
onLoad = () => {
...
this.task = sagaMiddlware.run(rootSaga);
}
destroy = () => {
this.task.cancel();
}

How Do I Make A Cache In Grails?

Here it is what I'm trying or doing, I'm just displaying say around 100 tasks instance in a page. The displayed tasks are the links (click-able), clicking on it will take to that particular tasks instance show page. From that show page, the user can go back to the page he/she was before, i.e to the page of displaying 100 tasks.
Now this time again i'm fetching those 100 tasks instance from DB, I need a way to "cache" them, so that there is no need to fetch again. Is there a way to do so?
Edit:
def user = User.get(springSecurityService.principal.id)
sleep(1000)
params.max = Math.min(params.max ? params.int('max') : 10, 100)
def tasks = Tasks.findAllByIsReadAndUser(false,user,[cache:true],params)
def tasksCount = Tasks.createCriteria().list(max: params.max as Integer, offset: params.offset as Integer) {
and {
eq('user',user)
eq('isRead',false)
}
}
if(request.xhr) {
render(view:'scroll', model:[userTasks:tasks,tasksCount: tasksCount.getTotalCount()])
}
else
{
[userTasks: tasks, tasksCount: tasksCount.getTotalCount()]
}
}
Notably I have this code, which works if the user hits scroll bar with bottom of his page and fetches more data :
(function() {
jQuery(function() {
return $(window).scroll(function() {
var url;
url = $('.pagination .nextLink').attr('href');
if (url && $(window).scrollTop() > $(document).height() - $(window).height() - 50) {
$('.pagination').show();
$('.pagination').text('Fetching more data please wait...');
return $.get(url, function(data) {
$('#scrolling').append(data);
return $('.pagination').hide();
});
}
});
});
}).call(this);
Now the problem is that, even though adding cache:true in my controller code and cache is been taken place, due to this jquery code I'm again fetching the data!
How do I stop this?
Thanks in advance.
Grails supports hibernate second-level cache. You can enable it in domain object, e.g.:
class Book {
…
static mapping = {
cache true
}
}
For details take a look into Database Mapping - cache or 5.5.2.2 Caching Strategy
The cache:true parameter caches the domain objects in Hibernate's L2 cache. This is a cache between the application and the database. In other words, if you fetch the same objects from Grails again it will get the objects from memory (the L2 cache) and not query the database again.
This will not cache requests between the browser and the application. Fortunately browsers already have a cache for this purpose.

Streaming Standard Output of a console application to an ASP.NET MVC View

What I am looking to do is:
1) From an MVC View, Start a long running Process. In my case, this process is a seperate Console Application being executed. The Console Application runs for potentially 30 minutes and regurlarily Console.Write's its current actions.
2) Back on the MVC View, periodically poll the server to retrieve the latest Standard Out which I have redirected to a Stream (or anywhere I can get access to it for that matter). I'll append newly retieved standard output to a log textbox or something equivalent.
Sounds relativly easy. My client side programming is a bit rusty though and I'm having issues with the actual streaming. I would assume this is not an uncommon task. Anyone got a decent solution for it in ASP.NET MVC?
Biggest issue seems to be that I cant get the StandardOutput until the end of execution, but I was able to get it with an event handler. Of course, using the event handler seems to lose focus of my output.
This is what I was working with so far...
public ActionResult ProcessImport()
{
// Get the file path of your Application (exe)
var importApplicationFilePath = ConfigurationManager.AppSettings["ImportApplicationFilePath"];
var info = new ProcessStartInfo
{
FileName = importApplicationFilePath,
RedirectStandardError = true,
RedirectStandardInput = true,
RedirectStandardOutput = true,
CreateNoWindow = true,
WindowStyle = ProcessWindowStyle.Hidden,
UseShellExecute = false
};
_process = Process.Start(info);
_process.BeginOutputReadLine();
_process.OutputDataReceived += new DataReceivedEventHandler(_process_OutputDataReceived);
_process.WaitForExit(1);
Session["pid"] = _process.Id;
return Json(new { success = true }, JsonRequestBehavior.AllowGet);
}
void _process_OutputDataReceived(object sender, DataReceivedEventArgs e)
{
_importStandardOutputBuilder.Insert(0, e.Data);
}
public ActionResult Update()
{
//var pid = (int)Session["pid"];
//_process = Process.GetProcessById(pid);
var newOutput = _importStandardOutputBuilder.ToString();
_importStandardOutputBuilder.Clear();
//return View("Index", new { Text = _process.StandardOutput.ReadToEnd() });
return Json(new { output = newOutput }, "text/html");
}
I haven't written the client code yet as I am just hitting the URL to test the Actions, but I'm also interested how you would approach polling for this text. If you could provide the actual code for this too, it would be great. I would assume you'd have a js loop running after kicking off the process that would use ajax calls to the server which returns JSON results... but again, its not my forte so would love to see how its done.
Thanks!
Right, so from the couple of suggestions I received and a lot of trial and error I have come up with a work in progress solution and thought I should share with you all. There are definitely potential issues with it at the moment, as it relies on static variables shared across the website, but for my requirement it does the job well. Here goes!
Let's start off with my view. We start off by binding the click event of my button with some jquery which does a post to /Upload/ProcessImport (Upload being my MVC Controller and ProcessImport being my MVC Action). Process Import kicks off my process which I will detail below. The js then waits a short time (using setTimeout) before calling the js function getMessages.
So getMessages gets called after the button is clicked and it does a post to /Upload/Update (my Update action). The Update action basically retrieves the status of the Process and returns it as well as the StandardOutput since last time Update was called. getMessages will then parse the JSON result and append the StandardOutput to a list in my view. I also try to scroll to the bottom of the list, but that doesn't work perfectly. Finally, getMessages checks whether the process has finished, and if it hasn't it will recursivly call itself every second until it has.
<script type="text/javascript">
function getMessages() {
$.post("/Upload/Update", null, function (data, s) {
if (data) {
var obj = jQuery.parseJSON(data);
$("#processOutputList").append('<li>' + obj.message + '</li>');
$('#processOutputList').animate({
scrollTop: $('#processOutputList').get(0).scrollHeight
}, 500);
}
// Recurivly call itself until process finishes
if (!obj.processExited) {
setTimeout(function () {
getMessages();
}, 1000)
}
});
}
$(document).ready(function () {
// bind importButton click to run import and then poll for messages
$('#importButton').bind('click', function () {
// Call ProcessImport
$.post("/Upload/ProcessImport", {}, function () { });
// TODO: disable inputs
// Run's getMessages after waiting the specified time
setTimeout(function () {
getMessages();
}, 500)
});
});
</script>
<h2>Upload</h2>
<p style="padding: 20px;">
Description of the upload process and any warnings or important information here.
</p>
<div style="padding: 20px;">
<div id="importButton" class="qq-upload-button">Process files</div>
<div id="processOutput">
<ul id="processOutputList"
style="list-style-type: none; margin: 20px 0px 10px 0px; max-height: 500px; min-height: 500px; overflow: auto;">
</ul>
</div>
</div>
The Controller. I chose not to go with an AsyncController, mainly because I found I didn't need to. My original issue was piping the StdOut of my Console application to the view. I found couldn't ReadToEnd of the standard out, so instead hooked the event handler ProcessOutputDataReceived up which gets fired when standard out data is recieved and then using a StringBuilder, append the output to previously received output. The issue with this approach was that the Controller gets reinstantiated every post and to overcome this I decided to make the Process and the StringBuilder static for the application. This allows me to then receive a call to the Update Action, grab the static StringBuilder and effectivly flush its contents back to my view. I also send back to the view a boolean indicating whether the process has exited or not, so that the view can stop polling when it knows this. Also, being static I tried to ensure that if an import in in progress, don't allow other's to begin.
public class UploadController : Controller
{
private static Process _process;
private static StringBuilder _importStandardOutputBuilder;
public UploadController()
{
if(_importStandardOutputBuilder == null)
_importStandardOutputBuilder = new StringBuilder();
}
public ActionResult Index()
{
ViewData["Title"] = "Upload";
return View("UploadView");
}
//[HttpPost]
public ActionResult ProcessImport()
{
// Validate that process is not running
if (_process != null && !_process.HasExited)
return Json(new { success = false, message = "An Import Process is already in progress. Only one Import can occur at any one time." }, "text/html");
// Get the file path of your Application (exe)
var importApplicationFilePath = ConfigurationManager.AppSettings["ImportApplicationFilePath"];
var info = new ProcessStartInfo
{
FileName = importApplicationFilePath,
RedirectStandardError = true,
RedirectStandardInput = true,
RedirectStandardOutput = true,
CreateNoWindow = true,
WindowStyle = ProcessWindowStyle.Hidden,
UseShellExecute = false
};
_process = Process.Start(info);
_process.BeginOutputReadLine();
_process.OutputDataReceived += ProcessOutputDataReceived;
_process.WaitForExit(1);
return Json(new { success = true }, JsonRequestBehavior.AllowGet);
}
static void ProcessOutputDataReceived(object sender, DataReceivedEventArgs e)
{
_importStandardOutputBuilder.Append(String.Format("{0}{1}", e.Data, "</br>"));
}
public ActionResult Update()
{
var newOutput = _importStandardOutputBuilder.ToString();
_importStandardOutputBuilder.Clear();
return Json(new { message = newOutput, processExited = _process.HasExited }, "text/html");
}
}
Well, that's it so far. It works. It still needs work, so hopefully I'll update this solution when I perfect mine. What are your thoughts on the static approach (assuming the business rule is that only one import can occur at any one time)?
Look into long poll. Basically you can open an ajax request and then hold onto it inside the controller.
Sample of long poll
This is something that you will want to do Async or you will possibly have issues with thread starvation.
Consider writing a service that runs on a server somewhere and pipes its output to a file/db accessible by your web server. Then you can just load the generated data in your website and returning them to your caller.
Understand that tying up your web server's threads for extended periods of time can result in thread starvation and make it look like your website has crashed (even though it's acutally just busy waiting for your console app to run).

Capturing twitter status timeline updates/infinite scroll updates

I'm using KRL to inject elements into twitter timeline statuses similar to Jesse Stay's TwitterBook. The problem I have is that these elements are only associated with statuses that are currently visible when the bookmarklet is initiated. If a new status is added through the 'new tweet' updated via Ajax or through status updates via infinite scroll, these new statuses are untouched.
Is there a way to either poll for new statuses or sense a twitter status update event via KRL in order to inject elements only into those newly added statuses?
The example posted at
http://kynetxappaday.wordpress.com/2010/12/25/day-21-modifying-facebook-stream-with-kynetx/
works with the Facebook stream but the concept is the same
create setTimeout infinite loop to look for stream items
only select stream items not marked as processed
process stream items
rinse and repeat
Code example from post
ruleset a60x512 {
meta {
name "MikeGrace-status-update-translator"
description <<
MikeGrace-status-update-translator
>>
author "Mike Grace"
logging on
}
global {
datasource insult:HTML <- "http://www.pangloss.com/seidel/Shaker/index.html?" cachable for 1 second;
}
rule find_status_updates_by_mike_grace {
select when pageview ".*"
{
notify("Starting to look for status upates by Mike Grace","");
emit <|
// get app object to raise web events
app = KOBJ.get_application("a60x512");
// function that finds FB status updates by Mike Grace
function findMikeGrace() {
// loop through each stream item on the page that hasn't been processed already by the app
$K("li[id^=stream_story]:not(li[kfbt])").each(function() {
var currentStreamItem = this;
// grab the current stream item posters name
var name = $K(currentStreamItem).find(".actorName").text();
// mark the stream item as being processed to reduce future processing times
$K(currentStreamItem).attr("kfbt","y");
// is the stream item by the perpetrator?
if (name == "Michael Grace") {
// strikethrough the original update
$K(currentStreamItem).find(".messageBody").wrap("<strike />");
// get selector to return translation of status update
var returnSelector = $K(currentStreamItem).attr("id");
returnSelector = "li#"+returnSelector+" .messageBody";
// raise web event to get translation for non geeks
app.raise_event("get_insult", {"returnSelector":returnSelector});
} // end of checking name
}); // end of looping through unprocessed stream items
// call myself again later to process new items on the page
setTimeout(function() {
findMikeGrace();
}, 9000);
}
// start the process of finding the perpetrator
findMikeGrace();
|>;
}
}
rule get_insult {
select when web get_insult
pre {
selector = event:param("returnSelector");
insulter = datasource:insult("#{selector}");
foundInsult = insulter.query("font");
singleInsult = foundInsult[0];
}
{
emit <|
console.log(singleInsult);
$K(selector).parent().after("<br/>"+singleInsult);
|>;
}
}
}

ASP.NET MVC ajax chat

I built an ajax chat in one of my mvc website. everything is working fine. I am using polling. At certain interval i am using $.post to get the messages from the db. But there is a problem. The message retrieved using $.post keeps on repeating. here is my javascript code and controller method.
var t;
function GetMessages() {
var LastMsgRec = $("#hdnLastMsgRec").val();
var RoomId = $("#hdnRoomId").val();
//Get all the messages associated with this roomId
$.post("/Chat/GetMessages", { roomId: RoomId, lastRecMsg: LastMsgRec }, function(Data) {
if (Data.Messages.length != 0) {
$("#messagesCont").append(Data.Messages);
if (Data.newUser.length != 0)
$("#usersUl").append(Data.newUser);
$("#messagesCont").attr({ scrollTop: $("#messagesCont").attr("scrollHeight") - $('#messagesCont').height() });
$("#userListCont").attr({ scrollTop: $("#userListCont").attr("scrollHeight") - $('#userListCont').height() });
}
else {
}
$("#hdnLastMsgRec").val(Data.LastMsgRec);
}, "json");
t = setTimeout("GetMessages()", 3000);
}
and here is my controller method to get the data:
public JsonResult GetMessages(int roomId,DateTime lastRecMsg)
{
StringBuilder messagesSb = new StringBuilder();
StringBuilder newUserSb = new StringBuilder();
List<Message> msgs = (dc.Messages).Where(m => m.RoomID == roomId && m.TimeStamp > lastRecMsg).ToList();
if (msgs.Count == 0)
{
return Json(new { Messages = "", LastMsgRec = System.DateTime.Now.ToString() });
}
foreach (Message item in msgs)
{
messagesSb.Append(string.Format(messageTemplate,item.User.Username,item.Text));
if (item.Text == "Just logged in!")
newUserSb.Append(string.Format(newUserTemplate,item.User.Username));
}
return Json(new {Messages = messagesSb.ToString(),LastMsgRec = System.DateTime.Now.ToString(),newUser = newUserSb.ToString().Length == 0 ?"":newUserSb.ToString()});
}
Everything is working absloutely perfect. But i some messages getting repeated. The first time page loads i am retrieving the data and call GetMessages() function. I am loading the value of field hdnLastMsgRec the first time page loads and after the value for this field are set by the javascript.
I think the message keeps on repeating because of asynchronous calls. I don't know, may be you guys can help me solve this.
or you can suggest better way to implement this.
Kaivalya is correct about the caching, but I'd also suggest that your design could and should be altered just a tad.
I made a very similar app recently, and what I found was that my design was greatly enhanced by letting the controllers work with the fairly standard PRG pattern (post-redirect-get). Why enhanced? well, because POST methods are built to add stuff to an app, GET methods are supposed to be used to get information without side effects. Your polling should be just getting new messages w/o side effects.
So rather than your $.post call expecting data and handling the callback, what I'd recommend is having your controller expose a method for creating new chat messages via POST and then another method that get the last X chat messages, or the messages since a certain timestamp or whatever.
The javascript callback from the post action, then can update some variables (e.g. the last message id, timestamp of the last message, or even the whole URL of the next message based on the info contained in a redirect, whatever).
The $.post would fire only in response to user input (e..g type in a box, hit 'send') Then, you have (separately) a $.get call from jquery that's set up to poll like you said, and all it does is fetch the latest chat messages and it's callback updates the chat UI with them.
I got my answer here: ASP.NET AJAX CHAT
The names below i am referring to are from above link.
i think the actual problem was with the timestamp thing and asynchronous behaviour of $.post. after calling "GetMessages()" method, even if the previous request to retrive chat message was not complete anathor call to same method used to fire due to setting timeout for "GetMessages()" method outside the $.post method. In my question you can see that timeout for "GetMessages()" method is set outside the $.post method. Now i set the timeout for "GetMessages()" method inside the $.post method. so that next call to "GetMessages()" only occur after 3 seconds of completion of current $.post method. I have posted the code below.
var t;
function GetMessages() {
var LastMsgRec = $("#hdnLastMsgRec").val();
var RoomId = $("#hdnRoomId").val();
//Get all the messages associated with this roomId
$.post("/Chat/GetMessages", { roomId: RoomId, lastRecMsg: LastMsgRec }, function(Data) {
if (Data.LastMsgRec.length != 0)
$("#hdnLastMsgRec").val(Data.LastMsgRec);
if (Data.Messages.length != 0) {
$("#messagesCont").append(Data.Messages);
if (Data.newUser.length != 0)
$("#usersUl").append(Data.newUser);
$("#messagesCont").attr({ scrollTop: $("#messagesCont").attr("scrollHeight") - $('#messagesCont').height() });
$("#userListCont").attr({ scrollTop: $("#userListCont").attr("scrollHeight") - $('#userListCont').height() });
}
else {
}
t = setTimeout("GetMessages()", 3000);
}, "json");
}
I addition to that i also changed few things. As suggested by ignatandrei i placed $("#hdnLastMsgRec").val(Data.LastMsgRec); immediately after function(Data) {.
and also
as said by MikeSW i changed the data retrieval process. Previously i was extracting data on the basis of timespan(retrieve all the data associated with
this room id that has greater timespan than last data retrieved message timespan) but now i keep track of the messageid. Now i retrieve only those data that
has message id greater than last retrieved message id.
and guess what no repeataion and perfectly working chat application so far on my intranet.
I still got to see it's performance when deployed on internet.
i think it solved my problem.
i will still test the system and let u guys know if there is any problem.
By default $.post() caches the results
You can either call $.ajaxSetup ({ cache: false}); before JS GetMessages function call to ensure caching is disabled or change the $.post to $.ajax and set cache attribute to false. In the end $.post() is a short cut to this:
$.ajax({
type: 'POST',
url: url,
data: data,
success: success
dataType: dataType
});

Resources