Review/help request: Chicken package in OpenWRT

[Jeronimo Pellegrini]

Hello!

I have been maintaining Chicken package in OpenWRT [1,2,3]. I have just 
made a PR to
upgrade it from 5.2.0 (which had problems with the compiler when 
installed on OpenWRT)
to 5.4.0 (which, as far as I could test, works fine):

https://github.com/openwrt/packages/pull/25007

However,

chicken-config.h is written during the Chicken compilation process, but 
things go wrong
in the OpenWRT build system: the built Chicken compiler will try to use, 
on the target,
flags that only make sense in the host (because they were hardcoded in 
the binary, which
includes chicken-config.h). So I made a patch to defaults.make which 
changes it to
remove those.

The excluded flags are -ldl and -fmacro-prefix-map=. Here:
https://github.com/openwrt/packages/pull/25007/files#diff-1deb1947a2148d3624133c082612ab9b5dbbf7f49c437feb37801c08eaed7dda

I feel like this hack is really not nice -- see that the test bot 
complained that I
included a binary patch, which is not really true -- and I was wondering 
if there is
a better way to do this. (Is there?)

Also, if there are more people who know the OpenWRT build process, I'd 
appreciate if you
could review the packaging.

Also - if anyone would like to use Chicken on devices running OpenWRT, 
it seems to work
fine (depending on available storage on the device -- a wireless router 
will likely need
an external root on a USB stick).  With enought room (I do this on a 
Netgear router), one
can use csc with gcc and chicken-install. Also available memory may 
affect the possibility
to compile large packages. But... It otherwise works :)

J.

[1] http://openwrt.org
[2] https://gitlab.com/jpellegrini/openwrt-packages
[3] https://openwrt.org/packages/pkgdata/chicken-scheme-full11
    https://openwrt.org/packages/pkgdata/chicken-scheme-interpreter11