[OE-core] [PATCH 1/2] go: Add recipes for golang compilers and tools

[Maciej Borzęcki]

On Mon, Dec 12, 2016 at 7:02 PM, Khem Raj <raj.khem at gmail.com> wrote:
>
> On Dec 12, 2016, at 4:36 AM, Gary Thomas <gary at mlbassoc.com> wrote:
>
> Looking at these recipes and they seem like a good starting point.
> I did have some issues though:
>
> * bitbake go - didn't create any target installable packages, although the
>  .../image directory seemed to be well populated.
>
>
> Yeah mostly, I was concentrating on cross builds but this is something to
> look at.
>
> * bitbake go-helloworld - had these errors:
> ERROR: go-helloworld-0.1-r0 do_populate_lic: QA Issue: go-helloworld:
> LIC_FILES_CHKSUM points to an invalid file: /MIT [license-checksum]
>
>
> Thats new, somehow it did not happen for me. I will check again.
>
> ERROR: go-helloworld-0.1-r0 do_compile: Function failed: do_compile (log
> file is located at
> /build/p0382_2016-01-13/tmp/work/cortexa9hf-neon-amltd-linux-gnueabi/go-helloworld/0.1-r0/temp/log.do_compile.31170)
> ERROR: Logfile of failure stored in:
> /build/p0382_2016-01-13/tmp/work/cortexa9hf-neon-amltd-linux-gnueabi/go-helloworld/0.1-r0/temp/log.do_compile.31170
> Log data follows:
> | DEBUG: Executing shell function do_compile
> | stat helloworld.go: no such file or directory
> | WARNING: exit code 1 from a shell command.
> | ERROR: Function failed: do_compile (log file is located at
> /build/p0382_2016-01-13/tmp/work/cortexa9hf-neon-amltd-linux-gnueabi/go-helloworld/0.1-r0/temp/log.do_compile.31170)
> ERROR: Task
> (/local/poky-cutting-edge/meta/recipes-extended/go-examples/go-helloworld_0.1.bb:do_compile)
> failed with exit code '1'
>
> I forced the compiler to my target (rsync is _my_ friend) and was
> able to run 'helloword.go'.  I've not tested much past this point,
> but I am enthusiastic about getting 'go' into OE-core
>
>
>
> Thanks Gary for your feedback it will help. I know I have a rev2 due for
> this patchset. Haven’t yet includes all needed changes for v2.

I would like to revive the discussion on including Go toolchain in
OE-core. Recipe specific sysroot changes have shown that external Go
layers may be a bit fragile. At least oe-meta-go, which we are using
right now, required some changes. Hopefully, the patches which were
posted today will get merged soon and everything will be back to
normal again.

There were some concerns previously about including Go in OE-core.
I''m not sure if these were eventually resolved. If not OE-core, how
about meta-go in OE layers? Maybe it would be a better match for the
already existing meta-{python,ruby,perl} layers. As long as there's a
single go-to layer for Go, everyone would benefit from this work.

Khem, do you need any help with v2 of the patchset you posted
previously? I think we should be able to spare some hours helping out
with this in Mender.

Cheers,
-- 
Maciej Borzecki
RnDity