Can't convert -0x80000000 to int32 with mp_obj_int_get_checked #2868

Activity

1. ghseb

   changed the title [-]Possible [/-] [+]Invalid conversion to int32 in mp_obj_int_get_checked[/+] on Feb 13, 2017
2. dpgeorge commented on Feb 14, 2017

   

   dpgeorge

   on Feb 14, 2017

   Member

   More actions

   This is an artefact of how mpz's (bignums) are converted to C-level ints: the conversion is done using positive numbers, and then only at the end is is made negative if the mpz's negative flag is set. The overflow is checked during the conversion to a positive number, and since 0x80000000 is too large for a signed 32-bit int, it exits with an overflow.

   You could try using mp_obj_int_get_truncated instead.
3. ghseb

   closed this as completedon Aug 28, 2017
4. Jongy commented on Jul 17, 2019

   

   Jongy

   on Jul 17, 2019

   Contributor

   More actions

   You could try using mp_obj_int_get_truncated instead.

   Cool! This works nicely.

   Anyway, I found this issue while looking around with the exact same problem.

   @dpgeorge :
   How about we document mp_obj_int_get_truncated so others know it can be used this way? Perhaps even create a nice wrapper mp_obj_int_get_uint?

   Actually I see there's a mpz_as_uint_checked, only that it's unused. So I'll create a PR with mp_obj_int_get_uint_checked.
5. tannewt

   added a commit that references this issue on May 26, 2020

   Merge pull request micropython#2868 from xobs/fix-usb-sleep-nrf

   ...

   eea01c3