The :latest tag points to a very old image versions, which are not
consistent with regards to the base-image. This is causing differences
in library versions across different target platforms.
Once a v0.3.0 release is available, :main can be changed into :0.3.0.
Closes#421.
NPM uses the .gitignore file, which includes the generated JS code. This
caused the v4.8.0 JS packages to be completely empty. Adding an empty
.npmignore file resolves the issue as if a .npmignore file is detected,
the .gitignore file is ignored.
This reverts commit 4f1a1bbce5. This was
intended to improve the development speed, but it breaks the build. This
is because even when cross-compiling, the Rust compiler does compile
some code for the host target for code-generation.
In this case, we return the (primary) NetID of the ChirpStack instance.
Supporting the HomeNSReq is useful in the case the roaming ChirpStack
instances are setup to resolve the servers using DNS. In case of OTAA,
the fNS resolves the JoinEUI to a hostname to which it must make the
HomeNSReq. This returns the HNetID, which then can be used to resolve
the hostname of the hNS to which the join-request must be forwarded.
onFocus is only triggered when when dropdown doesn't have focus where
onDropdownVisibleChange is always triggered when opening / closing the
dropdown.
Closes#334.
This makes it a lot easier to iterate over the intervals, as we no
longer have to take into account DST changes that could either result in
an invalid or ambiguous date, or not incrementing by the expected
interval. E.g. incrementing by 1 day resulting in a 23 hour increment
because or DST change.
On returning the metrics, we try to convert the NaiveDateTime back into
a DateTime<Local>, failing that, we skip it rather than failing on it.
Closes#415.
In general ChirpStack is configured with a rx1_delay configuration
matching the network latency (e.g. on cellular, one might want to set
this to rx1_delay=3). However, this does not take into account a
possible latency of the end-application. Handling the uplink and
enqueueing a downlink might take more time than the configured rx1_delay
(and get_downlink_data_delay) allows.
This option makes it possible to increase the RX1 Delay in the
device-profile. If the RX1 Delay has between increased relative to
the system default, then the get_downlink_data_delay will be
incremented with the same amount of seconds.
This needs to be tracked by git to make cargo publish work. If this
folder is in the .gitignore, then cargo publish will ignore this folder
as well and the publish command will fail because of missing .proto
files. If we would temporarily remove / rename the .gitignore file, then
cargo publish will error because the git state is dirty.
All these files can be generated using the `make api` command and there
is no real need to commit these into the repo. Only the api/go files
need to be comitted of how the Go import system works.
This also updates the Rust, Go, JS and gRPC-web (JS) code generation and
UI build to use the nix-shell environment instead of using Docker.
The z-index of the header was set to 1000 and the drawer header is behind the UI header so single frame download and title not displaying
Setting the z-index to 1002 is enough to fix the Drawer header issue.
This fallback was already implemented for reading the private key when
generating MQTT client-certificates, but it was not used when reading a
private key for connecting to the MQTT broker.
While it is recommended to use PKCS#8 private-keys, this will perform an
auto conversion for RSA PKCS#1 and EC P256 SEC1 private-keys to PKCS#8.
