Closed gmt closed 7 years ago
Why do need cypari2-1.0.0? Can't you use cypari2-1.1.1? Is there a dependency combination where you need cypari2-1.0.0 and pari-2.10 at the same time (I am sure I haven't allowed that yet)?
¯\_(ツ)_/¯
(dev-python/cypari2-1.1.1:0/0::sage-on-gentoo, ebuild scheduled for merge) conflicts with
~dev-python/cypari2-1.0.0[python_targets_python2_7(-),python_targets_python3_6(-),-python_single_target_python2_7(-),-python_single_target_python3_6(-)] required by (sci-mathematics/sage-8.0:0/0::sage-on-gentoo, installed)
OK, but you are using sage 8.0 which wants cypari2-1.0.0 (I may relax that) but you shouldn't also pull pari-2.10. I'll check dependencies.
Thanks for looking into it!
Ha, I see. I added a runtime dependency in 9999 that should also be in 8.0. Also cypari2-1.0.0 shouldn't allow you to pull pari-2.10. It will take a couple of hours to fix (because of kids). In the meantime just mask pari-2.10.
Will do, thanks!
edit: seems to have done the trick for now so please feel free to rear your children :P
Ok, should be all fixed now. pari-2.10 shouldn't come up in your dependency tree anymore. Thank you for the report. You can close the issue if you are all satisfied with the outcome. Just pipe again if I didn't fix it.
I'll give it a go in a few hours, thanks!
Depsolving after removing my mask and syncing up my repo didn't change anything so I am pretty sure that equates with success. Thanks again!
Since upstream cypari2 has moved on ahead of our pari version, I guess we sort-of have to branch cypari2-1.0.0? I'm very sure I don't understand the actual dependency constraints the overlay faces for sage-8.0. But assuming there's no better option, comparing ebuild prepare results from pari and cypari2 from 9.16.2017 overlay versions, this seemed like a reasonable patch to apply to cypari2-1.0.0, to get things building again: