New package: srain-1.0.1

[ghost]

No description provided.

[ghost]

It seems that glib-compile-resources cannot be found on the failing arches.
How should that be handled?

[travankor]

you usually need build_helper=gir

[ghost]

Thanks, let's see!
Also, don't merge yet, there was a new release made two hours ago.

[travankor]

build-helper=gir implies that glib-devel is in host

[ghost]

Do I need to take it out of makedepends, then?

[travankor]

you need it in both places

[travankor]

I think there's something wrong in the makefile/build flags. I ran into this issue before, but I forgot how I fixed it.

[ghost]

The errors now:

gcc: error: unrecognized command line option '-mfpu=vfpv3'
gcc: error: unrecognized command line option '-mfloat-abi=hard'

and

cc1: error: bad value ('armv8-a') for '-march=' switch
cc1: note: valid arguments to '-march=' switch are: nocona core2 nehalem corei7 westmere sandybridge corei7-avx ivybridge core-avx-i haswell core-avx2 broadwell skylake skylake-avx512 cannonlake icelake-client icelake-server cascadelake bonnell atom silvermont slm goldmont goldmont-plus tremont knl knm x86-64 eden-x2 nano nano-1000 nano-2000 nano-3000 nano-x2 eden-x4 nano-x4 k8 k8-sse3 opteron opteron-sse3 athlon64 athlon64-sse3 athlon-fx amdfam10 barcelona bdver1 bdver2 bdver3 bdver4 znver1 znver2 btver1 btver2 native

[ghost]

That looks like an issue upstream, right? I haven't found those flags when I grepped them, though.

[travankor]

yes, I would report it upstream

[travankor]

xbps-src does some automagic to get a cross-compile environment set up and minimal makefiles sometimes do not play well with it

[ghost]

Then isn't it on us to find out what is missing and then open a PR with a fix upstream?

[travankor]

Well, you can check on Void's IRC, there's usually one or two people that will help you.

I don't have time to look into this now, sorry.

[ghost]

Oh, yes, no I didn't expect us two to fix it now! Thanks for your help so far, we've progressed!

[ghost]

Ok, I think I get it. The errors look like an host compiler is trying to be used, and not the target compiler. Perhaps the Makefile does not let the compiler to be changed, and it does look like it.

[ghost]

I've added a patch now, to let the CC be overriden by what xbps-src wants. But now apparently the environment gets clobbered and headers/dependencies are no longer found.

This was my patch:

--- Makefile.in~	2020-03-14 10:55:04.950418920 +0100
+++ Makefile.in	2020-03-14 20:12:10.715364752 +0100
@@ -32,7 +32,7 @@
 default: src/Makefile
 	- mkdir -p $(BUILD_DIR)
 	$(MAKE) -C data resources
-	$(MAKE) -C src
+	$(MAKE) -C src -e
 
 .PHONY: install
 install: install-bin install-data install-po

[ghost]

Alright, using gnu-makefile build style seems to have the same effect as my patch had.
Now I noticed that all the CFLAGS have gone missing, which is why the headers are no longer found. Who let the flags out?

[ghost]

I was confused about the meaning of -e in make. I thought it is necessary for it to use environment variables, but it is only about precedence. The CFLAGS didn't get used because -e overrode them. make worked correctly without -e, using the env var and having flags appended from the Makefile.

The actual problem I had was being unable to set CC by environment, because the Makefile already set CC explicitly. The fix was to set CC conditionally based on whether it has its default value, or it has been redefined — basically falling back to a default if not otherwise specified.

[ghost]

Done.

[ghost]

Ping. According to me, it's good to merge.