• Replies 7 replies
  • Subscribers 18 subscribers
  • Views 1834 views
  • Users 0 members are here
Options
Related

Temperamental CIQ development environment

GHG23
I an using Eclipse (Oxygen) on Windows 10. What happens every now and then is that running my Datarun premium datafield (https://apps.garmin.com/nl-NL/apps/360a26c4-e2bb-45b7-8f97-2694ccf74eef) ends in a "out of memory" error in the simulator. Mostly after hitting the start button of the virtual device. The next day it just runs well.
Today it didn't again, but after exporting and updating it runs fine on my FR935. I am running the latest SDK. Is there a fix for this and are more people affected?
  • How close to max memory are you? Look at peak memory in the memory viewer. Changing app settings will increase the peak, so anything like that?
  • After starting the simulator and simulating data, the reported peak memory is 28.6kB and after starting the activity it crashes:
    "Failed invoking <symbol>
    Out Of Memory Error"

    This is with standard settings. It runs fine on the watch with the same settings. And it was working last week in the simulator as well with the same settings....
  • What is the idle memory in the sim? (the bottom line of the sim's window or memory usage in view memory). At 28.6 for peak, you're right at the edge with the max being 28.7.

    If you comment out some code to drop the peak a bit, do you still crash? (maybe try to reduce it by .2-.4k as a test)

    Since it worked last week, has anything changed? SDK version, a couple lines of code, etc.
  • Thanks Jim for replying. I will look at it this further weekend. Noone of my users has reported crashes since I uploaded the new version to the app store (as it didn't crash on my watch), and a few have already installed it.
    So that is the real worrying thing for me; that something can give an "out of memory" in the simulator, but not in real life. It used to be a guarantee until now.
  • Unfortunately, the ConnectIQ simulator doesn't perfectly emulate the device software. We do our best to get them as close as is reasonable, but it isn't perfect. If we can get details on conditions where our simulator and the device behavior varies, we will typically investigate and make changes.

    I know it is a big ask, but if you could provide a test case that demonstrates the inconsistency between the simulator and device I'd be happy to file an issue to investigate it.
  • I know it is a big ask, but if you could provide a test case that demonstrates the inconsistency between the simulator and device I'd be happy to file an issue to investigate it.


    Wouldn't it be great if you could publish a list of known inconsistencies, so we poor developers would know if we're encountering a known or new issue?