Not sure how…
Description
Details
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | • GWicke | T112330 In current VRS deployment scenarios, error logging should not assume Parsoid is the source of all errors. | |||
| Resolved | Krenair | T112339 Change VisualEditor's error logging to not assume all VRS errors are Parsoid errors |
Event Timeline
I left a bunch of comments on the CX extension code .. https://gerrit.wikimedia.org/r/#/c/237421/
Looking at the VE code here .. https://gerrit.wikimedia.org/r/#/c/217995/9/ApiVisualEditor.php .. I think many of those comments are applicable to this code as well.
In addition, also worth checking why the error reporting on lines 111-115 didn't report the RESTbase error raised in https://github.com/gwicke/restbase/commit/51bbb98aa0dd1a784c63a137f0478ba71f0acfb2 ... @GWicke, @cscott could also weigh in on where that error report got lost and transformed into the generic 'Something went wrong' error message that clients saw.
Change 238177 had a related patch set uploaded (by Alex Monk):
Change a bunch of parsoid references to acknowledge that RESTBase is sometimes in the middle
In addition, it makes sense for the VE extension to log any 4xx errors it gets on the save / review wikitext requests. Right now, they are being lost.
In the interim, I've asked RESTBase folks to enable logging for 4xx errors (they aren't logged right now because they are info level and production logging level is warn). We'll do the same in Parsoid. We need this extra info so we can figure out the remaining 'something went wrong' errors that are being reported.
In the longer term, without logging these 4xx reports, we won't be able to figure out what exactly (beyond 'something') went wrong.
Change 238177 merged by jenkins-bot:
Change a bunch of parsoid references to acknowledge that RESTBase is sometimes in the middle