Ask debug adapter for raw strings #14501
Conversation
Codecov Report
@@ Coverage Diff @@
## release-2020.10 #14501 +/- ##
================================================
Coverage 59.19% 59.19%
================================================
Files 721 721
Lines 40228 40228
Branches 5828 5828
================================================
Hits 23811 23811
Misses 15155 15155
Partials 1262 1262
Continue to review full report at Codecov.
|
| return _VSCODE_json.dumps(result) | ||
| stringified = _VSCODE_json.dumps(result) | ||
| # use `encode` to obtain a bytestring for increased portability | ||
| return stringified.encode() |
There was a problem hiding this comment.
These files are used in the normal case too. (Jupyter kernel I mean). There were no changes required there as well?
There was a problem hiding this comment.
Was testing and noticed this broke variable explorer, as values come back from Jupyter with a 'b' in front (bytestring) which is not JSON parseable. Just reverted this. It's not strictly required, just recommended by Pavel, and we should probably still do this properly, just not needed now.
joyceerhl
force-pushed
the
fix-escapes
branch
from
32ad978 to
e3a1cd0
Compare
|
Kudos, SonarCloud Quality Gate passed!
|
4 participants
For https://github.com/microsoft/vscode-python/issues/14441
With our previous custom evaluate request, debugpy was returning the result of calling
repr()on our variable value, which is what caused the extra backslashes. By sending "format": "rawString" with our request (implemented as a VS extension: https://github.com/microsoft/VSDebugAdapterHost/wiki/Stack-Trace-and-Value-Formatting), debugpy returns us the raw stringified JSON object, which we can then parse as JSON. The alternative would've been to parse Python string literals ourselves.package-lock.jsonhas been regenerated by runningnpm install(if dependencies have changed).