RFC: Using formfactor data in other recipes
Hi all,
Andrea has been doing some cleanup on meta-handheld, and he and I have come across what seems to be a bit of a shortcoming with formfactor over the variable-based alternative that was used in OE-Classic (MACHINE_GUI_CLASS, MACHINE_DISPLAY_WIDTH_PIXELS, MACHINE_DISPLAY_HEIGHT_PIXELS etc.). In meta-opie, the latter variables are still being used in a machine-specific packagegroup recipe to select the appropriate set of pixmaps to be installed for each machine based on the screen resolution, so you don't have to have pixmaps installed that you're never going to use (which would waste storage space).
formfactor is the current method supported in OE-Core for representing differences in machine display/input characteristics, however it seems only queryable at runtime since its files are not written into the sysroot. We could change it to stage to the sysroot, and therefore the data could be queried at least at packaging time in other recipes that depend on formfactor. Does this sound like a reasonable solution for this kind of problem, or is there a better solution?
Cheers,
Paul
Signature · 2 lines
Paul Eggleton Intel Open Source Technology Centre
Quoted 12 lines · Paul Eggleton
Hi all,
Andrea has been doing some cleanup on meta-handheld, and he and I have come across what seems to be a bit of a shortcoming with formfactor over the variable-based alternative that was used in OE-Classic (
MACHINE_GUI_CLASS,MACHINE_DISPLAY_WIDTH_PIXELS,MACHINE_DISPLAY_HEIGHT_PIXELSetc.). In meta-opie, the latter variables are still being used in a machine-specific packagegroup recipe to select the appropriate set of pixmaps to be installed for each machine based on the screen resolution, so you don't have to have pixmaps installed that you're never going to use (which would waste storage space).
Quoted 7 lines
formfactor is the current method supported in OE-Core for representing differences in machine display/input characteristics, however it seems only queryable at runtime since its files are not written into the sysroot. We could change it to stage to the sysroot, and therefore the data could be queried at least at packaging time in other recipes that depend on formfactor. Does this sound like a reasonable solution for this kind of problem, or is there a better solution?
Seems reasonable. We tend not to install some things to the sysroot as they're pointless (e.g. man pages or a lot of etc data) but in this case it would make sense to override the defaults as its information we can use.
Cheers,
Richard