Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (f530e92) 97.42% compared to head (a5307db) 97.48%.

@@            Coverage Diff             @@
##           master    #8049      +/-   ##
==========================================
+ Coverage   97.42%   97.48%   +0.06%     
==========================================
  Files         107      107              
  Lines       32598    32600       +2     
  Branches     3800     3800              
==========================================
+ Hits        31757    31781      +24     
+ Misses        634      616      -18     
+ Partials      207      203       -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

We could really do with a way to also get the traceback added in StreamReader.set_exception(). Can't seem to figure it out at the moment, but ideally we'd grab the stack trace in that function and then add it the exc.__cause__ somehow.

Maybe at worst we can just add:

try:
    raise Exception()
except Exception as e:
    exc.__cause__ = e

Though feels like quite an awkward way to achieve it.

Maybe at worst we can just add:

try:
    raise Exception()
except Exception as e:
    exc.__cause__ = e

Though feels like quite an awkward way to achieve it.

That seems awkward for sure. traceback.extract_tb might work but its also really expensive.. sys._getframe is cheap but only works on cpython.

Setting __cause__ is cheap and works everywhere.. its just a lot of copy pasta.

Backport to 3.9: 💚 backport PR created

✅ Backport PR branch: patchback/backports/3.9/a379e6344432d5c033f78c2733fe69659e3cff50/pr-8049

Backported as #8050

🤖 @patchback
I'm built with octomachinery and
my source is open — https://github.com/sanitizers/patchback-github-app.

Backport to 3.10: 💚 backport PR created

✅ Backport PR branch: patchback/backports/3.10/a379e6344432d5c033f78c2733fe69659e3cff50/pr-8049

Backported as #8051

🤖 @patchback
I'm built with octomachinery and
my source is open — https://github.com/sanitizers/patchback-github-app.

I think, this one may need a change note.
Also, I seem to recall finding other places where it's unset, in that old issue. Have you looked into it?

Not that I've seen, I just noticed a lack of information while debugging a couple of issues. The set_exception()s are the main other ones, but as described above it's a bit awkward to figure out how to put the traceback in. Best left to a separate PR if we do it.

@Dreamsorcerer I was thinking about all the other places as mentioned @ #4581 (comment). There's a set_exception() helper that isn't always used, it seems, and could be extended to be slightly more generic per my comment there.

Also, why didn't you reuse my patch @ #4581 (comment)? It's almost the same, but with better var names and the original exception repr backed into the message.

I was thinking about all the other places as mentioned @ #4581 (comment).

Oh, I totally forgot about that conversation. I was just debugging something on my own yesterday and found that one missing. Doesn't look like you solved the .set_exception() yet though.

Doesn't look like you solved the .set_exception() yet though.

@Dreamsorcerer I ended up pushing #8089. I haven't tried it, though. And didn't add any tests. Plz check it for obvious mistakes.

Setting __cause__ is cheap and works everywhere.. its just a lot of copy pasta.

@bdraco my PR unifies that a bit. Interestingly, I've found the use of __context__ in one place that I suspect may be problematic...