search

actionless / pikaur

AUR helper with minimal dependencies. Review PKGBUILDs all in once, next build them all without user interaction.
GNU General Public License v3.0
862 stars 87 forks source link

Resolving dependencies issue #27

Closed plague-doctor closed 6 years ago

plague-doctor commented 6 years ago

pikaur has problem with dependencies resolution for some packages.

Installing with pikaur

pikaur -S seafile-client

resolving dependencies...
Reading repository package databases...
Reading local package database...
Version mismatch:
 seafile-client depends on: 'seafile=6.1.5'
 found in 'aur': '6.1.5-2'

same installed with pacaur:

pacaur -S seafile-client
:: Package seafile-client not found in repositories, trying AUR...
:: resolving dependencies...
:: looking for inter-conflicts...

AUR Packages  (4) ccnet-6.1.5-3  libsearpc-1:3.0.8-2  seafile-6.1.5-2  seafile-client-6.1.5-2
Repo Packages (1) python2-simplejson-3.13.2-1

Repo Download Size:   0.00 MiB
Repo Installed Size:  0.37 MiB

:: Proceed with installation? [Y/n]
[...]
actionless commented 6 years ago

yeah, technically 6.1.5-2 is newer than 6.1.5 so if it would be criteria seafile>=6.1.5 it would match, isn't it?

plague-doctor commented 6 years ago

Technically it is. Maybe a switch to allow 'fast-forward' packages in case a package version is the same but the release has progressed? Just a thought, as there is no such problem with pacaur, and it is rather difficult to chase up every aur-package provider...

actionless commented 6 years ago

hm, i can strip everything in the version after the dash (ie pkgrel) when comparing deps/conflicts versions

plague-doctor commented 6 years ago

That would be very nice and useful, but I would rather make it on a request (prompt or switch). Sometimes the patches (new releases) can screw the things up badly...

actionless commented 6 years ago

idk, mb adding interactive prompt :: ignore version mismatch? [y/N] could be a way to go

polygamma commented 6 years ago

actually it is 6.1.5-2 == 6.1.5 - https://www.archlinux.org/pacman/vercmp.8.html

actionless commented 6 years ago

thanks, mb i'll just call vercmp instead of implementing that logic