ocaml 4.13 changed capitalization of some of the lib files which confuses
some of the findlib build tests (and then leads to fallback code which
depends on ocamlbuild).
A newer version of findlib addresses this.
In addition, findlib 1.9.4 added proper DESTDIR support, so we can drop
FAKE_FLAGS from the port Makefile as part of this update.
Diff from Volker Schlecht; and tested by myself on amd64.
For devel/ruby-sorted_set and graphics/ruby-rqrcode-core, do not do FLAVORed
builds by default, as these are pure ruby libraries with no native component.
--ignore-permission-errors didn't work in all cases, so I fixed one.
Testing my PR on top of 0.12.0 as well and not just master revealed another
bug, which lead to truncated filenames.
This is fixed in master already.
Do not wait for another release and apply both now, so backups no longer
abort or corrupt in those use cases.
All 57 (58 with this diff) bats(1) tests pass on amd64.
OK bket
`make test' only runs unit tests at the moment, but the bats(1) suite (for
which the unused TEST_DEPENDS is already there) covers more and will gain a
new test for a new fix soon, so actually run cli tests.
No package change.
OK bket
changelog:
* Add single compactor http client for delete and gennumber clients.
This fixes a bug caused by the accidental introduction of different
HTTP clients for compactor and gennumber operations that resulted
in 404s when only the gennumber middlewares were enabled
* Change default deletion mode to filter-only on account of the bug
* Delete can now be performed over GRPC
ok landry@
Now that vmd(8) supports communicating e820 ranges via fw_cfg, we
No longer need the patches that make SeaBIOS identify the guest
memory via the nvram registers.
vmd(8) already communicates etc/screen-and-debug via fw_cfg as well,
so drop that patch, too.
OK sthen@, mlarkin@
Turns out restic-generate(1) is clever enough to omit the reference to
rest-self-update(1) if built without the tag.
Same story for mount: it isn't compiled but we ship the manual.
This way we'll always have code and docs in sync.
OK sthen
- tweak recent change to use a normal patch instead of sed, the filename
is not version-dependent so no benefit of sed
- don't install the restic-self-update manpage now that it's been disabled
- install shell completion scripts
- update DESCR
Fixing the go build directive was enough for the default to kick in,
no need to explicitly disable it through flags.
Should be no package change, but bump revision for clarity.
Seems like newer shellcheck complains more, so disable another false positive
and move `disable' directives into the script.
Replace [ with [[ and (( while here.
Driver-side changes will be required for making use of the new firmware
images. The existing images we currently use do not change, and no
change in behaviour is expected for now.
The .pnvm files do change. They are still being parsed the same way
in Linux, which implies that these new files should just work.
I have verified that my AX210 device is still happy with new pnvm.
ok sthen@
changelog:
- rinstall: ignore environment variable http_proxy
- rinstall: use shell getopts to parse arguments
- raise error if a label alias begins with a space
portwise: no need to the pre-test ruby symlink hack: now it's possible
to specify the ruby version with the RUBY make var.
