Message384065
| Author |
Frederik Rietdijk |
| Recipients |
Alexandru Ardelean, Frederik Rietdijk, Ray Donnelly, barry, benjamin.peterson, bmwiedemann, brett.cannon, dstufft, eric.araujo, eric.smith, eschwartz, jefferyto, mcepl, methane, miss-islington, sascha_silbe, vstinner |
| Date |
2020年12月30日.13:08:45 |
| SpamBayes Score |
-1.0 |
| Marked as misclassified |
Yes |
| Message-id |
<1609333727.85.0.60854765511.issue29708@roundup.psfhosted.org> |
| In-reply-to |
| Content |
Building Python packages reproducibly has now basically been resolved with the reproducible bytecode as well as changes in tools such as pip.
Unfortunately, the interpreters do not yet seem to be reproducible. After certain changes, a Nixpkgs build of 3.9 shows several tiny bytecode differences. What could have caused these differences? Please see the attached diffoscope report.
As part of installation all bytecode is force regenerated using compileall. This is using the default checked-hash. |
|
History
|
|---|
| Date |
User |
Action |
Args |
| 2020年12月30日 13:08:47 | Frederik Rietdijk | set | recipients:
+ Frederik Rietdijk, barry, brett.cannon, vstinner, eric.smith, benjamin.peterson, mcepl, eric.araujo, sascha_silbe, methane, dstufft, bmwiedemann, Alexandru Ardelean, Ray Donnelly, eschwartz, miss-islington, jefferyto |
| 2020年12月30日 13:08:47 | Frederik Rietdijk | set | messageid: <1609333727.85.0.60854765511.issue29708@roundup.psfhosted.org> |
| 2020年12月30日 13:08:47 | Frederik Rietdijk | link | issue29708 messages |
| 2020年12月30日 13:08:47 | Frederik Rietdijk | create |
|