fixed network scan

[pljakobs]

esp_wifi_scan_get_ap_records() clears the ap data structure, so calling esp_wifi_scan_get_ap_num() after it will return zero. Switching the order fixes the error of Sming WifiStation.scan() not returning any networks on the esp32 platform

[what-the-diff]

PR Summary

• Improved Order of Function Calls
  The sequence of operations has been updated. Now, the first operation is gauging the number of access points (the places from which users connect to our system), known as 'ap_count'. The second operation involves collecting detailed information about these access points 'ap_info'. This change enhances system flow, verifying the quantity before delving into specifics.

[pljakobs]

this should fix #2905

[mikee47]

The call to esp_wifi_scan_get_ap_num is actually redundant since it's returned from esp_wifi_scan_get_ap_records:

memset(ap_info, 0, sizeof(ap_info));
ESP_ERROR_CHECK(esp_wifi_scan_get_ap_records(&number, ap_info));
// TODO: Handle hidden APs
for(unsigned i = 0; i < number; i++) {
  list.addElement(new BssInfoImpl(&ap_info[i]));
}

[pljakobs]

i was wondering about this and under which conditions the number of detected SSIDs and APs are different, especially, when the number of APs would be less than the number of SSIDs, but since this is the code that the ESP_IDF suggests, I thought it was "the right way"
I can drop the esp_wifi_scan_get_ap_num() call and replace ap_count with number where necessary (or actually use a variable that is a bit more descriptive)

[mikee47]

I think this loop should use number instead of event->number as it's the more recently aquired value. Presumably these values should always be the same, so could include an assert(number == event->number) but I think that's overkill.

[pljakobs]

ack, especially as it's unclear if esp_wifi_scan_get_ap_records changes number and what it would mean if those diverge

[mikee47]

Lovely, thank you for the fix!

[pljakobs]

Lovely, thank you for the fix!

thank you for your support