AutoConnectAux context changes in response to EdgeDriver enabler

[Hieromon]

This topic will examine the dynamic attach and detach availability of EdgeDriver as proposed by @FRANAIRBUS in Discussoins#1.

We use EdgeDriverEnabler.ino as the source code for this discussion. It is a skeleton to discover the appropriate EdgeUnified API and will not hesitate to break it.

So @FRANAIRBUS, I prepared EdgeDriverEnabler.ino example above based on your comment. It's based on my perception of a use case for keeping EdgeDriver's enabled state and menu display constant during the loop() function. It has 3 virtual EdgeDrivers and displays stats for each. The root / URI is the so-called menu panel, which you can use to select the enabled state of the EdgeDrivers. It is not fundamentally different from the control scheme you have presented in this comment.

The logic at the core of its control is quoted below.

String aux_drivers(AutoConnectAux& aux, PageArgument& args) {
  String  referer = String(portal.where());
  if (!referer.length())
    referer = "/";
  AutoConnectAux& request = *(portal.aux(referer));

  drivers.data.d1 = request["d1"].as<AutoConnectCheckbox>().checked;
  request["d1stats"].as<AutoConnectSubmit>().enable = drivers.data.d1;

  drivers.data.d2 = request["d2"].as<AutoConnectCheckbox>().checked;
  request["d2stats"].as<AutoConnectSubmit>().enable = drivers.data.d2;

  drivers.data.d3 = request["d3"].as<AutoConnectCheckbox>().checked;
  request["d3stats"].as<AutoConnectSubmit>().enable = drivers.data.d3;

  return String();
}

void driversProcess(void) {
  d1.enable(drivers.data.d1);
  if (drivers.data.d1) {
    if (!portal.aux("/d1-stats")) {
      Edge.join(STATS_D1, aux_d1stats);
      Serial.printf("Free heap: %" PRIu32 "\n", ESP.getFreeHeap());
    }
  }
  else if (portal.aux("/d1-stats")) {
    Edge.release("/d1-stats");
    Serial.printf("Free heap: %" PRIu32 "\n", ESP.getFreeHeap());
  }

  d2.enable(drivers.data.d2);
  if (drivers.data.d2) {
    if (!portal.aux("/d2-stats")) {
      Edge.join(STATS_D2, aux_d2stats);
      Serial.printf("Free heap: %" PRIu32 "\n", ESP.getFreeHeap());
    }
  }
  else if (portal.aux("/d2-stats")) {
    Edge.release("/d2-stats");
    Serial.printf("Free heap: %" PRIu32 "\n", ESP.getFreeHeap());
  }

  d3.enable(drivers.data.d3);
  if (drivers.data.d3) {
    if (!portal.aux("/d3-stats")) {
      Edge.join(STATS_D3, aux_d1stats);
      Serial.printf("Free heap: %" PRIu32 "\n", ESP.getFreeHeap());
    }
  }
  else if (portal.aux("/d3-stats")) {
    Edge.release("/d3-stats");
    Serial.printf("Free heap: %" PRIu32 "\n", ESP.getFreeHeap());
  }
}

The above code added AutoConnectAux release to the logic you devised to have the effect of reducing heap consumption. As you can see from the code above, it is simple but involves many steps. We have the possibility to discover there is an API that EdgeUnifed should be equipped with.

Break this skeleton by creating a new EdgeUnified API while considering your use case.

[FRANAIRBUS]

Hello @Hieromon good morning

I think I've gotten a little lost with this example and I haven't been able to check it or understand it very well yet.

1st question
what do you mean by
EdgeDriver's enabled state and menu display constant during the loop() function. ?

the condition of my implementation was that it does not appear in the menu

I think I understand that you want to be able to disable the driver without the need for a reboot of the ESP board, and in a way that saves as much as possible the size of the dynamic memory?

It seems great to me, although in my use cases I understand that when restarting and not loading, let's say, the setup of that edge driver, it will not consume memory (is it correct?)

the following video shows freeharp in my sketch with:
Step 1 using LH_MySQL
Step 2 disabling LH_MySQL with the enable option
Step 3 restarting LH_MySQL with the if option.

img-8178_ubFZU5Y8.mp4

by the way
missing an s

https://github.com/Hieromon/EdgeUnified-eval/blob/master/examples/EdgeDriverEnabler/EdgeDriverEnabler.ino#:~:text=aux%5B%22d3tats%22%5D.value%20%3D%20String(d3.data.stats)%3B

is 3

https://github.com/Hieromon/EdgeUnified-eval/blob/master/examples/EdgeDriverEnabler/EdgeDriverEnabler.ino#:~:text=Edge.join(STATS_D3%2C%20aux_d1stats)%3B

[FRANAIRBUS]

Hello thank you very much for the libraries and for your patience with me

sometimes the simplest things become the most complicated

I was using WiFi.status() in the main one and it did not work for me but I have realized that it is better that each driver compares it since not all the drivers will need a connection

let's forget my condition from the time of the dinosaurs and use this in the driver

void processSQLC() {

 if (WiFi.status() == WL_CONNECTED) {

  if (sqlc.data.serverNameSQLC.length()) {
    Serial.println("Starting Process SQLC");
}

or just this in the main

void loop() {

  mysql.enable(menu.data.Use_MySQL); 
  sqlc.enable(menu.data.Use_SQLC && WL_CONNECTED);

jejejejjejejej

Even so, I lack a stable condition for no internet since all the programs (mqtt, mysql, sqlc) block the loop and do not allow me to access the menus or the personalized pages
@Hieromon how about a ping to 8.8.8.8? would be the fastest right?

[Hieromon]

@FRANAIRBUS

how about a ping to 8.8.8.8? would be the fastest right?

Hmmm, I wonder. The fastest way to do this is to send only one packet connectionless via UDP without going through TCP. A significant service that uses UDP with Internet transparency that could use this trick might be DNS. If DNS packet Question section has only www.google.com, the DNS query packet length will be within a few tens of bytes. But the destination of its query must be your Internet provider's DNS address. If you simply use the gethostbyname (Similar to WiFi.hostByName), it may return a value based on your WiFi router's DNS cache. It does not guarantee Internet transparency.

In the first place, wouldn't WiFiClient::connect alone have faster throughput if the situation accepts waiting for a timeout of Ping?

[FRANAIRBUS]

hello @Hieromon

I don't know why all the drivers I have in my tests, your mqtt example, and my mysql and sqlc examples generate a loop lock for a few seconds multiplied by 3 driver and with those locks the custom pages don't load, in other words autoconnect It does not load me data due to the waiting time of each driver. I would perhaps have to delve a little deeper into each bookstore to check your comment.

In the first place, wouldn't WiFiClient::connect alone have faster throughput if the situation accepts waiting for a timeout of Ping?

when it does not find a server, it has its condition, retries and waiting times
when you don't have wifi it's the best that works hehehehe it goes immediately to the loop
when no internet was the problem

I have half solved it in the end as follows

void loop() {

  if (millis() - SENDloop > TimerSendOKloop) {        
       Serial.println("---------------------SYSTEM CHECK INTERNET----------------------");
         internet = (Ping.ping(ip_ping, 1));
         //Serial.println(internet);
            SENDloop = millis();
  }

 
  mysql.enable(menu.data.Use_MySQL && internet); 
  sqlc.enable(menu.data.Use_SQLC && internet);
  mqtt.enable(menu.data.Use_MQTT && WL_CONNECTED && internet);
  gpio.enable(menu.data.Use_BUILTIN);
  

  Edge.process(portal);
  mDNSUpdate();
  portal.handleClient();

}

if you try my example you will check it, when
mqttClient.connect
http.begin(wifiClient...
conn.connect
try to connect a delay occurs even lowering the timeout times. I can't download them much because I'm thinking about the possible lag of my clients or users

when I think of myself it is easier since my servers are what they are and I make it more specific but when it comes to turning it into a standard....

In the end I'm using 8.8.8.8, surely I'll add it as a menu for possible blockages (China).. my DNS server is MOVISTAR's (80.58.61.250) but I hardly ever use that

As for host resolve it takes more time.... and a ping only one packet and I set it to 500 ms

[FRANAIRBUS]

Perhaps later it can be solved better and you know more about this than I do.
perhaps as you say a own dns server or in a stand a second test to the own server for the databases of my clients, at the moment a ping to tokyo has just given me a result of Ping ms 284
but as we know this all depends on the connection of each individual and other factors.
you are just on the other side of the world, if I make a straight hole in the ground I think I reach japan without detours so the ping was quite fast to the first server I found, I don't even want to think about a client with an adsl and rural connection :)

The important thing as a measurement device is that the loop is not blocked in the local network and since it never rains to everyone's taste, I cannot set it as a standard either, since there will be people who want to send data to their own local server.

[Hieromon]

@FRANAIRBUS I have an idea, please go to this topic.