MMM-Scenes2

hrmax23

I may have spoken too soon. It has been working for about 12 hours but just froze on MMM-MyStandings so maybe that module has a similar issue.

sdetweil

@hrmax23 many of the modules have a design flaw copied from one of the original modules.

the web side provides the timing control to request data, which is reset when the helper responds.

but if the helper encounters an error( network glitch, timeout… ), it does not respond. and so the main side is waiting and the helper thinks it is done… deadlock not recoverable until restart

mumblebaj

@sdetweil Can I ask a stupid question? When I initially started coding my first module I was advised to handle the suspend and resume in the module and picked the samples up from I think your sample module or maybe it was another module. Apologies if it was not yours.

Are we saying it is bad practice to use the two together? I.e. handling the suspend and resume etc. in module and should rather leave it out completely?

sdetweil

@mumblebaj well, it depends…

if you want you module to continue to get data in the background when suspended, then no need to hook suspend and resume…

but those calls will count against your data cap at the provider…
my code doesn’t ‘handle’ the suspend/resume, just stays informed…

also, if you had a long refresh interval (once every hour for example) , the data might be stale if you don’t refresh on wake up… like overnight.
my system suspends all (hides) to blackout the screen as my monitors don’t support off/on…
in this case the calls made while suspended would be wasted work…

mumblebaj

@sdetweil Thanks Sam for the explanation. So it depends on the module design and data provider etc.

MMRIZE

@mumblebaj
Usually, implementing the suspend and resume methods of the MM module is useful when your module depends on expensive resources. When your module is inactivated, it would be better to stop consuming resource.
But that mechanism is somehow annoying for me, so I ignore those methods. :D

sdetweil

@mumblebaj said in MMM-Scenes2:

So it depends on the module design and data provider etc

correct… other users might not know or understand the data provider issues, so in my opinion, if you are posting for other users consumption, you should consider this as a design requirement… (to be debugged later if you don’t do it from the beginning)

sdetweil

@MMRIZE said in MMM-Scenes2:

But that mechanism is somehow annoying for me, so I ignore those methods

what could be annoying… they call, you set a flag.
you save any timer handle when u start it anyhow…

so on sleep you stop it, and on resume you restart the timer… what could be so hard
if you get data back while suspended, you don’t call updateDom(), and don’t restart the data request

tiny amount of code…

mumblebaj

@sdetweil and @MMRIZE Thanks for the useful information.

MMRIZE

@sdetweil

what could be annoying… they call, you set a flag.

It may be a relative point of view.

Let’s assume my module is using a paid API that has a daily quota. To save the consumption API quota on unused time, it would be better to suspend/resume the process of the node_helper(usually node_helper takes a charge to consume API)

However, some API needs an expensive re-establishment fee for their usage. For example, some Google APIs might have short valid lifetimes of retrieved tokens and data. So, it needs to begin with handshaking again at some point. It may require the additional cost of the connection/auth/indexing process. If some 3rd party module tries to resume/suspend my module in a habitual rapid cycle, It might be another abused waste of consumption and, even worse, unpredictable.
I said this would be annoying for me.

Of course, I might also be able to manage the lifecycle of resource consumption, but I think it is not worth it, at least based on my experience with MM modules. (Yup, I’m the one who made most many modules in this scene.)
I prefer background-alive with silence than suspend/resume modules frequently. It is more stable. (at least for me.)