Fix more operations that can return small ints #17785
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #17785 +/- ##
==========================================
- Coverage 98.39% 98.37% -0.02%
==========================================
Files 171 171
Lines 22257 22265 +8
==========================================
+ Hits 21899 21904 +5
- Misses 358 361 +3 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
Code size report: |
Signed-off-by: Jeff Epler <jepler@gmail.com>
Since adafruit#9531 a few ways remained to produce long ints that were not reduced to small ints. Explicitly introduce "mp_obj_new_long_from_ll" and make the "new_int" functions always produce small ints when possible. At sites where a long int is really intended (or will eventually be reduced to a small int if the result fits), the called function is changed to be the new _long_ function. Signed-off-by: Jeff Epler <jepler@gmail.com>
Signed-off-by: Jeff Epler <jepler@gmail.com>
|
the nrf port is failing due to a "403 Forbidden" error downloading some files from nordic: furthermore, the failure is not detected at the time, but the build continues until this message is shown with any luck this is a transient error and nordic will fix the download link. |
|
The same error popped up at PR #17365. |
CircuitPython GitHub actions builds are also seeing this. It looks like |
There was a problem hiding this comment.
This looks good to me, thanks @jepler.
I tried to think of ways to combine mp_obj_new_int, mp_obj_new_int_from_ll, and mp_obj_new_int_from_ull into a single inner static implementation but they were all going to involve some level of compiler crimes I think. 😆
I've linked this PR to close the issue as IIUC it addresses everything enumerated there, but let me know if I'm missing a case.
| mp_obj_t mp_obj_new_int_from_ull(unsigned long long val); // this must return a multi-precision integer object (or raise an overflow exception) | ||
| mp_obj_t mp_obj_new_int_from_ll(long long val); // returns a small int if value fits | ||
| mp_obj_t mp_obj_new_int_from_ull(unsigned long long val); // returns a small int if value fits | ||
| mp_obj_t mp_obj_new_long_from_ll(long long val); // this must return a multi-precision integer object (or raise an overflow exception) |
There was a problem hiding this comment.
It's fully separate to this PR, but at some point we should probably rename all the longint references in MicroPython to bigint to avoid this ambiguity about what long means.
Summary
#17730 identified some code paths that could still return long ints when small ints would suffice. Fix them.
Testing
I wrote a new test based on sites that called new_int_from_{u,}ll, including the case I'd noticed of pow3 (already reported in #9531 as well).
Trade-offs and Alternatives
I haven't seen the code size change summary yet, but this is likely to increase code size. I'm open to suggestions about how to reduce the increase.
Making the new
long_fromfunctions always return longs, and changing the existingnew_int_fromfunctions to return ints when possible is just one alternative. Another alternative would be to add calls tomp_int_maybe_narrowin paths that usednew_int_from, but this seemed like the worse alternative, with more added code.