Status Update
Comments
al...@google.com <al...@google.com> #2
Apologies Jeremy, I realized we have a tracking ticket for the sheet APIs specifically:
al...@google.com <al...@google.com> #3
Sounds good. I will make it a dup.
al...@google.com <al...@google.com> #4
This is a Lint issue.
al...@google.com <al...@google.com> #5
Moving back to root component again for triage.
cm...@google.com <cm...@google.com>
tn...@google.com <tn...@google.com> #6
I'd be happy to just apply that patch (shown in #1) if that's what you want; I think AndroidX is the main or only client of the prefix type (it was added to support AndroidX) -- just checking that that's exactly what you want. Is there a particular test case you want to pass and one you want to fail?
al...@google.com <al...@google.com> #7
Is there a particular test case you want to pass and one you want to fail?
The androidx.arch.core
from androidx.paging
scenario captures it well. This should pass.
Something like accessing androidx.arch.core
from com.google.android.material
should fail.
au...@google.com <au...@google.com> #8
How far up should it go?
Should com.google.android.material be able to access com.android.tools?
al...@google.com <al...@google.com> #9
Ooh, yeah, good point. I think the Material folks would expect something like com.google.android.material
or com.google.android
as the scope. Which indicates @RestrictTo(prefix = "androidx")
would be useful. This would also be useful for ErrorProne's implementation since it's not aware of Maven group IDs.
al...@google.com <al...@google.com> #10
Though at that point the restriction is not LIBRARY_GROUP_PREFIX
, it's more of a relaxed package
visibility. Equivalent for androidx, since we require the Java package to match the Maven group ID, but not generally true.
Description
Currently, the implementation of
sameLibraryGroupPrefix
warns whenandroidx.paging
uses aLIBRARY_GROUP_PREFIX
-scoped API fromandroidx.arch.core
. This seems to be an oversight, given theTODO
in the method and the original request forLIBRARY_GROUP_PREFIX
being specifically targeted to AndroidX usage.What we really want, I think, is just
firstIndexOf
rather thanlastIndexOf
: