Would the attached change to erlang.port.mk address the issue in a more
general way?
On 3/7/23 10:24, Omar Polo wrote:
On 2023/03/07 09:10:15 +0000, Stuart Henderson <s...@spacehopper.org> wrote:
On 2023/03/07 10:07, Omar Polo wrote:
On 2023/03/06 19:42:15 +0000, Stuart Henderson <s...@spacehopper.org> wrote:
On 2023/03/04 19:22, Omar Polo wrote:
could set TEST_TARGET=tests; tests don't currently work (they call
rebar3 and not rebar3-${MODERL_VERSION} and try to fetch stuff from
the internet). I've ran the tests outside of the ports infrastructure
and are all passing!
Maybe symlink it in to make it easier to run tests?
ln -fs ${LOCALBASE}/bin/rebar3-${MODERL_VERSION} ${WRKDIR}/bin/rebar3
that's an option, yes, but since it will break immediately after
(trying to use git to fetch deps) I just haven't bothered. I have on
my todo list to send a patch upstream to allow specifying the rebar3
(and pass CFLAGS to cc) if I don't forget...
It's not uncommon to have to allow pbuild to have network access for
tests to run, I'm flipping between not allowed and allowed quite often
when building/testing things, if we can make it so things work when the
build user does have network access it's helpful.
Sure then, here's an updated tarball that have `make test' working if
you allow _pbuild to talk to the internet.
Index: erlang.port.mk
===================================================================
RCS file: /cvs/ports/lang/erlang/erlang.port.mk,v
retrieving revision 1.31
diff -u -p -r1.31 erlang.port.mk
--- erlang.port.mk 16 Jan 2023 18:27:29 -0000 1.31
+++ erlang.port.mk 7 Mar 2023 09:30:35 -0000
@@ -43,6 +43,7 @@ REBAR_BIN ?= ${LOCALBASE}/bin/rebar3-${
# Make sure rebar gets called as 'rebar3', otherwise escript tries to call the
# binary name (e.g. rebar3-25) as the script entrypoint.
_MODERL_LINKS += rebar3-${MODERL_VERSION} rebar
+_MODERL_LINKS += rebar3-${MODERL_VERSION} rebar3
.endif
# Append the flavor to all the Erlang dependencies
