| Age | Commit message (Collapse) | Author |
|---|
|
Thanks Ilya Khaprov for the fix.
|
|
This is the same as `check`. The files of the cross-platform-actions VM
are synchronised in both ways, so we can re-use the same configuration.
|
|
`vmactions` VM fails to boot quite frequently and we have no visibility
on the reason of the failure.
The only failure I got out-of-the-box with cross-platform-actions VM was
the `core-deps-rel` test case which hangs. It comes from the use of
`erl_call` to detect if the Erlang node was started correctly. A comment
in the generated release starts scripts says:
# users who depend on stdout when running rpc calls must still use nodetool
# so we have an overload option to force use of nodetool instead of erl_call
If `$USE_NODETOOL` is set, it will use `relx_nodetool` instead. Indeed,
this fixed the hang.
|
|
It looks like the VM is sometimes stuck booting, long before executing
any tests. The default timeout of 6 hours blocks other CI jobs for a
very long time. Because the VM won't boot anyway, we reduce this time to
one hour.
One hour should be long enough for a successful slow execution of the
tests.
|
|
Tests are executed in a FreeBSD VM using the latest stable release. The
default version of Erlang and Elixir from the Ports tree are being used;
currently Erlang/OTP 26 and Elixir 1.17 as of this commit.
Otherwise, tests are split the same way as for Linux/Windows/OSX. That
said, the list is duplicated, so a bit annoying to maintain.
|
|
On FreeBSD, libsodium's `sodium.h` header is installed in
`/usr/local/local`. The Makefile already adds
`/usr/local/include/sodium` to the compiler's `-I` search path, but it
doesn't cover the FreeBSD case.
This patch adds `-I /usr/local/include` to the CFLAGS on the make(1)
command line to let libsalty2's Makefile pick it up. This fixes the
build on FreeBSD.
|
|
Its Makefile called sed(1) in a way that was only compatible with GNU
sed. As a consequence, the build failed with BSD sed.
Now, the test case fetches dependencies and patches `reloader`'s
Makefile to use a sed(1) script compatible with both implementations of
sed(1).
|
|
Use "sed ... > ..." + "mv ..." instead.
The problem is that the `-i` argument is not part of POSIX and is
implemented differently betwee GNU sed and BSD sed. The previous use was
ok on Linux but failed on FreeBSD for instance.
|
|
Using the "latest" version doesn't work because it fails to determine
a version for Erlang/OTP.
To work around this, we pin the version of Elixir to "1.18.4-otp-27".
While here, update asdf to version 1.18.0.
|
|
With the parallel execution of tests, I think it's possible that an
Erlang VM starts to initialize the missing cookie while a parallel VM
reads an yet-to-be-populated empty file. This leads to the following
crash:
Slogan: Kernel pid terminated (application_controller) ("{application_start_failure,kernel,{{shutdown,{failed_to_start_child,net_sup,{shutdown,{failed_to_start_child,auth,{\"Too short cookie string\",[{auth,init_no_setcookie,0,[{file,\"auth.erl\"},{line,360}]},...
|
|
This is the standard name and is available everywhere, unlike "netcat".
|
|
If `V=...` is set on the command line to increase verbosity while
debugging, the first grep(1) command will also catch the debug output of
make(1) which contains the executed command line and thus mentions
"hello".
The fixed grep(1) command line explicitly grep for a line starting with
"hello".
|
|
|
|
... instead of https://github.com/hypernumbers/hypernumbers which is no
longer available apparently.
|
|
parse_transform output
At least for me with Erlang/OTP 28, parse_transform messages end up on
stderr.
Let's always redirect stderr to stdout. It will work for anyone with any
versions of Erlang.
|
|
|
|
If the user running the tests has set `init.defaultBranch` to something
else in their `.gitconfig`, the tests won't have the expected branch
name.
This patch explicitly sets the branch name to make sure the tests are
not affected by a user setting.
|
|
|
|
Should make x509 and others work.
|
|
By default Erlang projects should not have Elixir enabled.
|
|
|
|
|
|
|
|
This commit also includes a way to completely disable Eunit
as that is generally desirable for Elixir-only projects.
|
|
Was failing on Windows in lz4-erlang without it.
|
|
|
|
This allows overriding the default which is to use the same
explicit version as the dep_* line, and instead specify
ranges of versions.
|
|
|
|
We don't need the tarball in that scenario.
|
|
|
|
|
|
|
|
While using "-mode minimal" removes too much that we need
(such as inet_db), we can safely disable the distribution
by default and limit the number of processes and sockets
to a maximum of 1024.
|
|
|
|
Fixes #1004
|
|
|
|
The scrape file was updated to only seek "rebar3"
tagged packages. There are too many Elixir-only
packages tagged as "make".
|
|
|
|
Templates now no longer use Make variables for substitution
but instead replace strings with their equivalent:
template_name: Corresponds to n=template_name
project_name: Corresponds to $(PROJECT) or in=project_name
This allows defining templates outside of Makefiles. For
example an external plugin could define their templates
in templates/my_template.erl and then have the following
in the included Makefile:
tpl_my_template = $(file < $(THIS)/templates/my_template.erl)
By default the created file will be in src/template_name.erl.
This can be overriden with the tplp_* variable:
tplp_my_template = src/model/my_template.erl
Substitution is applied both to the template contents and
to its path.
In addition, attempting to overwrite an existing file when
creating a template will result in failure.
|
|
|
|
It will be removed at a later date.
|
|
This includes the SSL security fix for OTP-25+.
|
|
We use $1 instead of $(1) or $p instead of $(p) when possible,
except when building variable names (pkg_$(p)_name for example,
but also $(dep_fetch_$(1))).
|
|
This method dates back from the beginnings of Erlang.mk
and should not have been used for around a decade by now.
Adding "git" to the offending dep lines is enough to upgrade.
|
|
Use the query_* equivalents instead.
|
|
This should both be more solid and more secure.
|
|
|
|
|
|
|
|
|
