Status Update
Comments
sh...@gmail.com <sh...@gmail.com> #2
sh...@pinterest.com <sh...@pinterest.com> #3
hu...@google.com <hu...@google.com> #4
I am also experiencing this issue. I have 9 chromecasts V3, all bought at the same time and on 2 of them I was not able to cast images using either the
default or styled media reciever.
When comparing the debugs of a working and non working device I found that the non working one is using the canary framework instead of the V3
First of all i checked and this device doesn't have the Preview program on, how come it gets beta code ?
The error is as stated above on the cast.o.media.Il.create function
Non working canary
c = c.startsWith("image/") ? cast.o.media.Il.create(b) : .L(new hu(b,ku))
Working V3
c = c.startsWith("image/") ? .K(new ot(b)) : _.K(new Nt(b,Qt))
sh...@pinterest.com <sh...@pinterest.com> #5
hu...@google.com <hu...@google.com> #6
Can you also confirm whether this is happening with our latest version? You can try the preview URL that can be found here:
as...@gmail.com <as...@gmail.com> #7
So I created a custom reciever app pointing to the preview cast reciever version and I have the same error
It looks like the bug is also present in the preview version.
The problem I have is even if I specify to use the v3 cast reciever (
Why this chromecast in particular has this behavior ?
serial number is: 1125ADSFDMJ
Is there another version of the cast_receiver_framework.js that i could use ?
Thanks,
hu...@google.com <hu...@google.com> #8
Can you please confirm if this preview version is the same one announced on [Feb 10th (https://groups.google.com/g/cast-sdk-announcements/c/-dLdTPHTvDc) that is now in canary testing?
There are disconnects between the dates in the
as...@gmail.com <as...@gmail.com> #9
The SDK version is CAST SDK CAF Version: 3.0.0095 the shaka player loaded is 3.0.10. So apparently not the same as the one announced on Feb 10th (3.0.8).
I attached the whole debug output with the preview framework (
How come some devices use canary and some do not ?
Thanks,
Joris
hu...@google.com <hu...@google.com> #10
To answer your questions, as we roll out the releases, some devices will see the new release (canary) and others will see the old release until the new release is fully ramped up in production. In Feb we found an issue with Shaka player version 3.0.8 which was fixed in version 3.0.10. We delayed the rollout of the receiver release to get the updated Shaka player 3.0.10 into the release, and that's what's currently in canary (and in preview). (The preview URL will always point to the newest build about to be released.)
For the cssting photos bug reported here, engineering has implemented a fix, and we hope to update the release with the fix tomorrow. We'll do our best to update this bug as soon as the fix is in canary.
sh...@pinterest.com <sh...@pinterest.com> #11
We're in the process of updating the preview URL with the fix, which should be in place in an hour or so. Joris or John, once that's in place, would you be able to do quick check against the preview URL with the development version of your apps to verify that this has fixed the problems reported?
hu...@google.com <hu...@google.com> #12
The
sh...@pinterest.com <sh...@pinterest.com> #13
Thanks
hu...@google.com <hu...@google.com> #14
I just tested it and it works now !!
Thanks a bunch for your quick response.
We'll do some more testing in the next few days. We'll let you if we encounter other issues.
Thanks,
Joris
as...@gmail.com <as...@gmail.com> #15
Great -- thanks for the quick confirmation! We'll be rolling this fix out to production soon. (Full ramp will be early next week.)
hu...@google.com <hu...@google.com> #16
I've tested and confirm that the issue is resolved in preview mode, also appreciate your rapid feedback.
As I'm new to building and support Chromecast based applications I'd like to better understand an aspect of the release and testing process.
Was the issue we experienced present in the preview release of Feb 1st or Feb 10th, or were unannounced updates to the preview version made after the fact.
What I want to understand is if we monitor preview releases and test our application regularly with them can we avoid this type of issue impacting us in the future? Based on our admittedly very limited sample of Chromecasts it looks like Canary was in use on 10-20% of devices before this issue was identified.
as...@gmail.com <as...@gmail.com> #17
Thanks for testing, John! Yes, this issue has been present since the initial preview release was released in February. You should have been able to see this in the initial preview that was released back then. So going forward, testing with the preview URL prior to a new release will be the best way to ensure that you can catch issues affecting your app in advance of a new release.
an...@google.com <an...@google.com> #18
Excellent, thanks for the confirmation. I wasn't working in this space back when the preview release came out. It's good to know that if we subscribe and monitor the cast SDK announcements we'll be able to detect issues early and ensure the ongoing functionality of our product.
Thanks again to you and all the team for the excellent support we've received on this issue.
hu...@google.com <hu...@google.com> #19
We've updated the canary with the fix -- can you verify that you're now seeing that the problem is fixed?
Description
(I'm creating this bug based on Internal Issue 234414461 .)
Steps to reproduce
app
, setminSdk >= 24
(whenminSdk < 24
, dexing transforms are not incremental)lib
, setcompileOption.targetCompatibility = "11"
(in lower Java versions, nest members do not exist)lib
, write a Java class with nested classes./gradlew :app:mergeLibDexDebug --build-cache
at that location../gradlew :app:mergeLibDexDebug --build-cache
. Many tasks/transforms should have cache hits../gradlew :app:mergeLibDexDebug --build-cache
again. It will fail with:Root cause
Dexing transforms use a desugaring graph to store dependencies among class files for incremental dexing/desugaring.
The desugaring graph's nodes are currently absolute paths, which means after a remote cache hit, it will not be usable and will result in subsequent failed incremental builds.
We were aware of this issue when designing the graph and used a workaround , but (I think) at some point Gradle changed the behavior and it no longer worked.
Potential fix
Make the desugaring graph relocatable. It is a bit difficult because there isn't one single root path to normalize the absolute file paths (instead, there are multiple root paths).
(If we can't work out a solution soon, the temporary measure is to ensure in the first build after a remote cache hit, affected transforms will be non-incremental.)