Fixed issue #314 - include information about view/serializer in warnings #327
Conversation
Codecov Report
@@ Coverage Diff @@
## master #327 +/- ##
==========================================
- Coverage 98.22% 98.08% -0.14%
==========================================
Files 53 54 +1
Lines 4729 4919 +190
==========================================
+ Hits 4645 4825 +180
- Misses 84 94 +10
Continue to review full report at Codecov.
|
…in warnings.
The main technique for doing this in deeply nested places from which we call
`error()` or `warn()`, is to a stack of traces that get added as prefices to the
message (rather than having to pass this context information through many layers
of code).
This also fixes a set of consistency issues with warning/error messages:
- Always start message with `{serializer/view name}: ` if possible.
This meant removing this info from the body of the message
where we are now adding it as a prefix.
- First sentence shouldn't have a capital letter
(the message is printed after colon from the prefix)
- Subsequent sentences should have a capital letter.
spookylukey
force-pushed
the
better_traceability_in_warnings_alt1
branch
from
37f8b0b
to
52b1c88
Compare
|
i think the difference is very manageable.
which imho is a lot better. always having a full trace including the view is more consistent and may also help when the same serializer name is used in different apps. i could already see the next person asking for this 😄 thanks again! i hope you get that number down to zero fast. |
This is a replacement for #322, with changes as suggested in the review there, squashed to reduce noise in the history.
I ran this against my clients project. With #322, I was getting:
With this I'm getting:
There are a few more unique because there are a few more places where are we are adding the
MyViewSet:prefix. The bigger change is that most of my messages now haveMyViewSet: MySerializerinstead of justMySerializer:prefix.