Oh, wait can't have that kind of logic: the PKG_BUILD_DIR needs to match
whatever the tarball extracts as, which probably depends on PKG_NAME not
the source package name; in fact I'm not sure why the patch for
libltdl/libtool makes the change it does to PKG_BUILD_DIR. Will have to
review.
Regards,
Daniel
On 09/01/16 08:28 PM, Daniel Dickinson wrote:
Actually, this means that PKG_NAME is irrelevant to the source package
name issue; source package name *is* is the subdir name and PKG_NAME is
not related the source package name, so really the better solution is to
alter PKG_BUILD_DIR to use source package name instead of PKG_NAME
(since that is currently the main issue with having a source package
name different than the subdir name).
This nicely solves some outstanding issues with source package names,
PKG_NAME, and dependencies.
Regards,
Daniel
On 09/01/16 08:13 PM, Daniel Dickinson wrote:
I have discovered there is some other dependency, which I have not yet
determined, for source package name == name of package directory. In
particular if the SDK tries to rebuild any of the kernel packages, it
fails due to the kernel package in the base feed being in subdir linux,
but the build system is looking for subdir kernel (the source package
name).
This probably only shows up now, for me, due to switching metadata.pm to
use PKG_NAME as source package name instead of subdir name.
Regards,
Daniel
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
