buildlink breakage on libwebp?

["Dr. Thomas Orgis" <thomas.orgis%uni-hamburg.de@localhost>]

Hi,

with current CVS (updated from snaphsot, that's the version in the
paths), I'm trying to build gdal-lib, which needs libwebp. That cracks
up with:

Making all in man
Making all in extras
--- webp_quality-webp_quality.o ---
--- get_disto-get_disto.o ---
--- libwebpextras_la-extras.lo ---
--- libwebpextras_la-quality_estimate.lo ---
--- webp_quality-webp_quality.o ---
  CC       webp_quality-webp_quality.o
--- get_disto-get_disto.o ---
  CC       get_disto-get_disto.o
--- libwebpextras_la-extras.lo ---
  CC       libwebpextras_la-extras.lo
--- libwebpextras_la-quality_estimate.lo ---
  CC       libwebpextras_la-quality_estimate.lo
--- get_disto ---
  CCLD     get_disto
--- libwebpextras.la ---
  CCLD     libwebpextras.la
--- webp_quality ---
  CCLD     webp_quality
--- get_disto ---
libtool:   error: cannot find the library '/tmp/work-with-system-gcc/pkgsrc/cvs-20200227/bw/graphics/libwebp/work/.buildlink/lib/libwebp.la' or unhandled argument '/tmp/work-with-system-gcc/pkgsrc/cvs-20200227/bw/graphics/libwebp/work/.buildlink/lib/libwebp.la'
*** [get_disto] Error code 1

bmake[1]: stopped in /tmp/work-with-system-gcc/pkgsrc/cvs-20200227/bw/graphics/libwebp/work/libwebp-1.1.0/extras
1 error

bmake[1]: stopped in /tmp/work-with-system-gcc/pkgsrc/cvs-20200227/bw/graphics/libwebp/work/libwebp-1.1.0/extras
*** [all-recursive] Error code 1

bmake: stopped in /tmp/work-with-system-gcc/pkgsrc/cvs-20200227/bw/graphics/libwebp/work/libwebp-1.1.0
1 error

bmake: stopped in /tmp/work-with-system-gcc/pkgsrc/cvs-20200227/bw/graphics/libwebp/work/libwebp-1.1.0
*** Error code 2

Stop.
bmake[1]: stopped in /stuff/build/env/system-gcc/pkgsrc/devel/pkgsrc/graphics/libwebp
*** Error code 1

Stop.
bmake: stopped in /stuff/build/env/system-gcc/pkgsrc/devel/pkgsrc/graphics/libwebp


This looks like extras subdir wants to link to the very libwebp that
just got built, but wrappers point to the .buildlink directory, where
it is not present.

Isn't this a rather standard failure mode with a standard solution?


Regards,

Thomas

-- 
Dr. Thomas Orgis
HPC @ Universität Hamburg