Summary: | dev-ros/global_planner-1.14.0 : CMake Error at /.../costmap_2dConfig.cmake:106 (message): | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Toralf Förster <toralf> |
Component: | Current packages | Assignee: | Robot Operating System team <ros> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | pms |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
CMakeOutput.log
dev-ros:global_planner-1.14.0:20160811-023238.log emerge-history.txt environment |
Description
Toralf Förster
2016-08-11 06:44:23 UTC
Created attachment 443004 [details]
CMakeOutput.log
Created attachment 443006 [details]
dev-ros:global_planner-1.14.0:20160811-023238.log
Created attachment 443008 [details]
emerge-history.txt
Created attachment 443010 [details]
environment
we would need to rebuild costmap_2d whenever sci-libs/pcl switches vtk useflag on or off; that could be a := dep on pcl, but pcl would need to have some kind of dynamic slot... ccing pms team to see if they have a better solution. The only solution for flag-Dependant ABI is to use USE-deps to enforce a specific variant, i.e. If you'd like to support both ABI variants, you need to 'copy' the flag into revdep, and use 'foo='. a useflag is clearly an overkill here since it doesnt reflect any choice and is just a workaround a := dep is overkill too: costmap_2d doesnt need to be rebuilt in order to run; one needs to rebuild it in order to build against it... no clear solution then (In reply to Alexis Ballier from comment #7) > a useflag is clearly an overkill here since it doesnt reflect any choice and > is just a workaround How is it overkill if it solves a problem that we have no other way to solve? The vtk flag is a choice at the costmap_2d level, and that choice has implications that bleed through to the revdeps. (In reply to Zac Medico from comment #8) > (In reply to Alexis Ballier from comment #7) > > a useflag is clearly an overkill here since it doesnt reflect any choice and > > is just a workaround > > How is it overkill if it solves a problem that we have no other way to > solve? The vtk flag is a choice at the costmap_2d level, and that choice has > implications that bleed through to the revdeps. it is a choice at the pcl level (and is already an useflag), which exports include dirs, which are inlined by costmad_2d fooConfig.cmake, which is then used by global_planner. everything but the first one has no need for an useflag. *** Bug 591274 has been marked as a duplicate of this bug. *** IMO adding USE flags to the revdeps is much more elegant than letting users endure build failures until an EAPI extension lands in a number of years. *** Bug 591298 has been marked as a duplicate of this bug. *** We have a similar issue with Qt and certain revdeps. While it may not be ideal, USE flags work fine until such a time as there's an alternative mechanism to trigger a rebuild. should be fixed in the 1.16 series -- I don't see any vtk dep |