profile
viewpoint
Paul Bovbel paulbovbel @locusrobotics Canada www.bovbel.com

crystaldust/unitree_a1_ros2 4

The UNOFFICIAL ROS2 driver for Unitree robotics dog A1

paulbovbel/ansible-personal 1

personal deployment scripts

huawei-ros/unitree_legged_sdk 0

SDK tools for control robots.

locusrobotics/actionlib 0

Provides a standardized interface for interfacing with preemptable tasks. Examples of this include moving the base to a target location, performing a laser scan and returning the resulting point cloud, detecting the handle of a door, etc.

locusrobotics/bloom 0

A release automation tool which makes releasing catkin (http://ros.org/wiki/catkin) packages easier.

locusrobotics/realsense-ros 0

Intel(R) RealSense(TM) ROS Wrapper for D400 series and SR300 Camera

locusrobotics/ros 0

Core ROS packages

locusrobotics/roslint 0

Lint macros for ROS packages

locusrobotics/ros_comm 0

ROS communications-related packages, including core client libraries (roscpp, rospy, roslisp) and graph introspection tools (rostopic, rosnode, rosservice, rosparam).

push eventlocusrobotics/catkin_virtualenv

push time in 13 days

pull request commentlocusrobotics/catkin_virtualenv

Rename process to program name

The setproctitle approach would fail because shebang size exceed 128 chars in CI - and potentially elsewhere. The exec -a approach is certainly a lot simpler, wish I saw that flag earlier.

paulbovbel

comment created time in 13 days

push eventlocusrobotics/catkin_virtualenv

Paul Bovbel

commit sha d3a9e6e3d0ee138db557a7cd2b3d782309f19571

Use exec -a flag to rename process

view details

push time in 13 days

Pull request review commentlocusrobotics/catkin_virtualenv

Rename process to program name

-#!/usr/bin/env bash-${${PROJECT_NAME}_VENV_DEVEL_DIR}/bin/python ${program_path} "$@"+#!${${PROJECT_NAME}_VENV_DEVEL_DIR}/bin/python

Rebased but breaks tests - in CI this shebang ends up longer than the 128 character limit due to nested paths

paulbovbel

comment created time in 13 days

PullRequestReviewEvent

push eventlocusrobotics/catkin_virtualenv

locus-services

commit sha 8022ab94199d873757dbd2b22531eaa80829f976

Tailor: Updating Jenkinsfile

view details

locus-services

commit sha b6a744a6b8751a7bb2ff6c19619979a16fe5ca9c

Tailor: Updating Jenkinsfile

view details

Paul Bovbel

commit sha ed437cf16b975423b2744a0e68004c2a5f3ad25d

Rename process to node name

view details

Paul Bovbel

commit sha d6e4a592c37513f23397b197aa5728553ceaf2d9

Rethink approach

view details

Paul Bovbel

commit sha f209801d64a407169a9b345adcc6e326f915e248

Mirror how native processes are named

view details

Paul Bovbel

commit sha 0bca349f2a0091d9a5bd12f64292123387af558e

Bump dependencies

view details

push time in 13 days

issue closedlocusrobotics/catkin_virtualenv

Workaround native library (un)availability for Python 3

tf2 (any any native modules inside the ROS ecosystem) have to be built with Python 3 libs for compatibility with PYTHON3 virtualenvs, which is not currently the case.

closed time in 13 days

paulbovbel

issue commentlocusrobotics/catkin_virtualenv

Workaround native library (un)availability for Python 3

Well, consider this one solved!

paulbovbel

comment created time in 13 days

pull request commentros-drivers/vrpn_client_ros

Ros2

Removed multiple sensor support. To be honest, I just don't know how it works and what are sensor id's and therefore I deleted it. If you can provide me some info about that, I can commit changes later.

I wish I could remember, but I imagine they existed for a reason. I haven't had access to a VICON system in years.

AlexKaravaev

comment created time in 16 days

Pull request review commentros-drivers/vrpn_client_ros

Ros2

 namespace vrpn_client_ros        tf_broadcaster.sendTransform(tracker->transform_stamped_);     }+++    if (tracker->calculate_twist_and_accel_ && tracker->previous_message_arrived_)

This is an interesting feature! Honestly I think it probably belongs as another node in the system, consuming 'pose' messages and spitting out 'twist' and/or 'accel'. that keeps the VRPN client 'simpler' in that it's responsible only for adapting the incoming data.

AlexKaravaev

comment created time in 16 days

PullRequestReviewEvent

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha e50a9f64541e4558757bc61268841dc9dc8e0358

Switch to mitogen strategy

view details

push time in 17 days

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha 07f01e4b95ac92c9e9b3c0e8dc0e9ef3c80d6118

Disable minecraft, valheim, csgo servers

view details

Paul Bovbel

commit sha 7273295f4149a1d06df3a263277deef437a2a8e5

Update media configuration

view details

Paul Bovbel

commit sha 428fa24a5ad8f78b0bb5c8641493794194e27ce7

Replace home with media server

view details

Paul Bovbel

commit sha 4272e5edfab4b1ffab253210f613da3254672814

Update to caddy 2.4.6

view details

Paul Bovbel

commit sha ce157fb349055b8b0053f75f6af706704c77fba5

Switch to mitogen strategy

view details

push time in 17 days

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha cf4a551acc8eeb71bff52dadaa2602b94dc49f6e

Fix mitogen strategy loading

view details

push time in 17 days

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha 5e2930dea3d2f5450e47740e37c0949582c45436

Remove home.bovbel.com from github actions

view details

push time in 17 days

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha 29eb9677c6603992f5b3795521479993f7aae2b2

Clean up caddy config

view details

Paul Bovbel

commit sha ba96454f61560cba94ef9996e685b2f32ef48be8

Switch to mitogen strategy

view details

push time in 17 days

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha af232a15b98dc69300290421e19a97c29979e7ae

Replace home with media server

view details

Paul Bovbel

commit sha 1fb7fed6b7ffe0a164871ee435f4d4da6ea0dbf9

Move game server config to a different group

view details

Paul Bovbel

commit sha 76b8c4568414be00d9f9e6733bfc374b7258caec

Update to caddy 2.4.6

view details

push time in 19 days

PullRequestReviewEvent

pull request commentros-drivers/vrpn_client_ros

Ros2

This is great, thanks. I'll circle back around to review this as soon as I can. Merging into foxy is fine, we can re-arrange branches once this is released anyways.

AlexKaravaev

comment created time in 24 days

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha 494f82805ee0159b2c86e09337198096435ec518

Update media configuration

view details

push time in a month

issue commentros-drivers/zbar_ros

Porting to ROS2

Hi @ijnek, thanks for reaching out! I've created a ros2 branch, please do PR your changes against that.

Looks like there's still no connect/disconnect callbacks for lazy pub/sub in ROS2 (https://docs.ros2.org/galactic/api/rclcpp/structrclcpp_1_1SubscriptionEventCallbacks.html seems like where it should exist..), so please feel free to leave a TODO for that.

I'd prefer if you kept throttle_repeated_barcodes in as I've found it useful in the past, but it's not strictly necessary as it's really something that could be implemented consumer-side.

ijnek

comment created time in 2 months

create barnchros-drivers/zbar_ros

branch : ros2

created branch time in 2 months

startedfacontidavide/Treexy

started time in 2 months

startedarun11299/cpp-subprocess

started time in 2 months

push eventpaulbovbel/ansible-personal

Paul Bovbel

commit sha f69240c265a611deda5fae6d7064e16d19f00fd3

Disable minecraft, valheim, csgo servers

view details

push time in 2 months

created taglocusrobotics/tailor-meta

tag0.1.18

created time in 2 months

push eventlocusrobotics/tailor-meta

Paul Bovbel

commit sha 7a982eb8a20811a350dcaa069e0cbdfdd59b5538

Bump cxx_standard to 20

view details

push time in 2 months

push eventpaulbovbel/cyclonedds

Paul Bovbel

commit sha 999d100626250415cf276b195e2f00a196b1b4a1

More threads on top of 0.8.0

view details

push time in 2 months

push eventpaulbovbel/cyclonedds

Erik Boasson

commit sha 25b7dad41798680345fee493b9ab937009b5cbea

Skip all work in tracing addrset calculation if no trace needed Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Christophe Bedard

commit sha 4d3e913f73c7b5962f0423b6a9ecadd0a3cd210a

Make sure to use buffer with truncated name for pthread_setname_np Signed-off-by: Christophe Bedard <bedard.christophe@gmail.com>

view details

Thijs Miedema

commit sha 05eada6793f8e06be3930aa83e47aa3f5bb5639d

Remove duplicate increment of input buffer position upon array key readout.

view details

Thijs Miedema

commit sha f87b11e2eaef58aa4dcef314f4c46a9d06b525c4

Fix some subtle ways the key calculation can go wrong in unlikely datatype scenarios. Allow small bounded strings to stay below 'MD5-hashing-threshold' in IDL.

view details

Erik Boasson

commit sha 430404dd76273c3cefe4d73b27f103b393fd6f8f

Fix domain_create test trace handling Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha fc5b4aa3ce68c2703a5f4e7068228c9e1e173606

Fix liveliness test output Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 638d48b7001272adc8a30844639192012a460259

Clean up/refactor/improve waitset tests Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 9d8f2fe597de165b6d56fe8de0d32e8fc21b4774

Disc stress test: more info on failing final check Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 7e3ffb8347d53c5e37ad5e473fa056e818b5661d

Eliminate dds_entity_trigger_set Used only in two places, with behaviour that is confusingly different in its side-effects from dds_entity_status_set. Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 654c0df8ad92106076d4943fb1eaaadd6cbeea5a

Allow mask = 0 for read/take status operation The documentation comment suggests it is allowed, though it said "NULL" and didn't say what exactly that means, but in any case 0 was a rather useless value as it would turn the operation into a no-op. The wisdom of returning an error if the mask contains status flags not relevant to the particular entity type is also doubtful, but if 0 is allowed to simply mean "everything", that is not really much of an issue. And with that change, dds_get_status_changes becomes an alias for dds_read_status with mask = 0. Deprecating should be considered, as the "get" is rather ambiguous and can just as easily be interpreted as an innocent "getter" or as a function that also reset the status. Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 10eb8881418f39df4e53240ce5323bd238d79cd1

Always update m_cb_pending_count before waiting Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 47cccd85f8ab8dc64b66c7fe4115f5932ca46b8c

PoC: Set statuses prior to invoking listener Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha b9a21cb8ebc93aba2394ab3212d735c65b8a8862

Only materialize DATA_ON_READERS when needed The vast majority of the DDS applications ignores DATA_ON_READERS, as it really is only useful for group ordered/coherent subscribers, and those weren't available in any DDS implementation for many years. Cyclone doesn't have them either. At the same time, DATA_ON_READERS is relatively costly, because the spec says it should always be raised when DATA_AVAILABLE is on any of the subscriber's readers, and reset when read/take is called on any of the same. That roughly means every single time a reader history cache is updated, the subscriber status has to be manipulated as well. It turns out that this has a significant performance impact once one allows the statuses to remain set even when invoking the listener, so that one can have listeners that monitor without any side-effects. (Significant here is 3.02MS/s -> 2.95MS/s in ddsperf -TOU pub/sub test on my laptop, and the significane is perhaps more in sinking below 3MS/s than in any absolute terms!) This commit addresses that be only materializing the DATA_ON_READERS state when the subscriber is attached to a waitset, and returning an approximation of it when querying the subscriber's status. The workaround for getting a materialized DATA_ON_READERS is to create an otherwise unused waitset with just the subscriber attached to it. Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha ae91dfeecc9082d403d0fa90e91723c10e0150e2

Add _arg version for listener setters/getters These extend on the existing setters/getters as follows: * Provide a return value for returning BAD_PARAMETER if the "listener" argument is a null pointer. * Allow overriding the listener argument pointer specified in the dds_create_listener call. This is quite useful but the workaround for the absence of this argument was a rather convoluted procedure combining getting, setting and merging listeners. * A new boolean argument whether (true) or not (false) the status corresponding to the listener should be reset prior to invoking the listener (or more precisely: never get set) as described in the DDS specification. Setting it to false allows installing listeners without side-effects. The old functions have been re-written in terms of the extended ones. Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha 2399a7f85499475235aed395b0db0b7348a68fcf

Leave status set on invoking listener if requested This provides the implementation of leaving the status flags set in the entity when invoking a listener if that listener was set with the "invoke_on_reset" flag clear (via the new, extended interface for setting listeners), and resetting it as required by the spec otherwise. Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha a3ac2fd83c2a087cd01d2118b7ff8070fe4b8984

Status mask should only affect waitset triggering This brings the behaviour of the status mask in line with the specification, previously masking an event (i.e., clearing its bit in the mask) would have two additional, impractical consequences: * It would prevent a listener from being invoked. If there is no need for a listener, don't set one. * It would force the corresponding status to 0 (not signalled), making it impossible to poll for events outside the status mask. E.g., with a waitset and a mask of DATA_AVAILABLE to wait for the arrival of data, it was impossible to observe any liveliness changed events on the same reader. Signed-off-by: Erik Boasson <eb@ilities.com>

view details

Erik Boasson

commit sha fb0c57ee07cd075f04dad6a7d6add92f170930b6

Add test for combination of waitset and listener This adds a test to verify that invoking a DATA_AVAILABLE listener configured not to auto-reset the status does indeed allow a waitset to trigger and that the waitset (generally, there is a short window in time during which the status can be observed) wakes up after the listener was invoked. It tests this both in the case of a listener that doesn't reset DATA_AVAILABLE internally, and in the case of a listener that does reset it by calling dds_take(). Signed-off-by: Erik Boasson <eb@ilities.com>

view details

T0ny Peng

commit sha f2bc7030d4c0979b214e8f12b05f3c092c095598

Fix the wrong description about the default value of ParticipantIndex

view details

Vidotto, Ettore

commit sha 449b569542d5d59be8de18b3f80143a6d8ea699c

Default message "using network interface..." is printed as info Fix for issue #867

view details

Erik Boasson

commit sha 468cedf6f2b0b133e0681b30ec43419d40e76d4b

Double unlock creating publisher with invalid QoS Signed-off-by: Erik Boasson <eb@ilities.com>

view details

push time in 2 months

PullRequestReviewEvent
more