Re: [Discuss-gnuradio] gr-air-modes/issues/81

[Nick Foster]

Thanks for the detective work, Maitland. I'll cave to convention and move the filename.

Fix pushed to master.

--n

On Sun, May 1, 2016 at 6:21 AM A. Maitland Bottoms <address@hidden> wrote:

(
 discuss-gnuradio readers: this is about this issue
 https://github.com/bistromath/gr-air-modes/issues/81
)

Nick Foster writes:
 > Thanks Maitland. Do you happen to know offhand if other OOT modules are
 > affected?

I have not seen any like this. Lucky me, I saw Sesame Street in my youth,
so I can play "One of these things is not like the other". I came up with
noticing all the working OOT modules have a swig/<modname>_swig.i file.

For example in gr-osmosdr has a swig/osmosdr_swig.i
while gr-air-modes has a swig/air_modes.i

So currently gr-air-modes breaks because of some implicit deriving
of the SWIG_MODULE_NAME from the .i file name. I have not tracked
down exactly when behaviour changed. Clearly Debian stable with
cmake 3.0.2 builds fine without changes, and Debian unstable last
year with cmake 3.4.1 was also fine. I think the failures coincide
with cmake 3.5.0, continuing in the current cmake 3.5.2.

Attached are two solutions.
Both yield builds on both Debian stable and Debian unstable.
You can take your pick.

fix-issue81-by-file-rename.patch:
My first attempt, just making gr-air-modes look more like other
OOT projects by renaming the .i file.

fix-issue81-by-PROPERTY-SWIG_MODULE_NAME.patch:
A one-liner, after noticing cmake 3.5.2's UseSWIG.cmake has some
documentation of SWIG_MODULE_NAME.

Seems that gr_modtool has been creating the swig/modulename_swig.i
file name since before it got merged into gnuradio. So gr_modtool
users are less likely to stumble accross this issue.

-Maitland

enc: