Adding Executable description class #454

roger-strain · 2020-08-19T21:00:23Z

This is an initial commit toward the process described in ros2/design#272. It should not be merged to master yet, and probably needs a feature branch instead.

Haven't added unit testing yet, but also not as familiar as I'd like with the testing setup, so guidance toward getting appropriate tests would be greatly appreciated.

Part of implementation of refactor for #114

Distro A; OPSEC #2893

Signed-off-by: Roger Strain [email protected]

roger-strain · 2020-08-19T21:01:37Z

@ivanpauno @hidmic Getting started on the implementation of this. Wanted to just put a quick first cut of one class up first to make sure I get the process right. Looking forward to feedback and suggestions, especially with respect to testing.

ivanpauno

I've left some minor comments, looks good to me.

ivanpauno · 2020-08-24T16:47:54Z

launch/launch/descriptions/executable.py

+from launch.utilities import perform_substitutions
+
+_global_process_counter_lock = threading.Lock()
+_global_process_counter = 0  # in Python3, this number is unbounded (no rollover)


this is going to be a problem until ExecuteProcess doesn't internally use this new description, as there is going to be two global counters (and thus, names may collide).

This does not need to be fixed now.

I've changed the name to a local-specific name regardless. There's no need for those to conflict.

ivanpauno · 2020-08-24T16:49:00Z

launch/launch/descriptions/executable.py

+        process_event_args = self.__process_event_args
+        if process_event_args is None:


nit: this can be done in a single line

Suggested change

process_event_args = self.__process_event_args

if process_event_args is None:

if self.__process_event_args is None:

is this check even needed?

It doesn't seem like it. Also, as personal nit, I'd much rather have __expand_substitutions return something to apply_context instead of the former mutating state directly.

Removed the identified check. __expand_substitutions is still modifying the object, but it's doing is by adding newly-defined values instead of changing the existing ones. Not sure if it's really different enough for this concern, so if we want to continue tweaking it, please let me know. If we really would rather have a stateful object to pass around instead of changing this one, I think I can see the outlines of doing that, but I'm not sure if we're gaining much (other than complexity). Open to suggestions/thoughts/comments, of course.

ivanpauno · 2020-08-24T16:51:56Z

launch/launch/descriptions/executable.py

+
+    def __init__(
+        self, *,
+        cmd: Iterable[SomeSubstitutionsType],


A bit unrelated: maybe we should take advantage of this new description to fix:

Change cmd parameter in ExecuteProcess #263

Passing a string to ExecuteProcess arguments behaves unintuitively #394
?

Using a single string instead than a list of strings plays better with substitutions.

I've taken a stab at addressing this. Also added a few unit test cases to evaluate how it's handling the different cases described in #263 itself. I'd very much like to add more tests to make sure we're getting the expected results, so if you have any additional suggestions, I'd love to hear them.

ivanpauno · 2020-08-24T16:53:45Z

launch/launch/descriptions/executable.py

+        with _global_process_counter_lock:
+            global _global_process_counter
+            _global_process_counter += 1
+            self.__name = '{}-{}'.format(name, _global_process_counter)


nit: use f-strings

ivanpauno · 2020-08-24T16:55:36Z

I will create a branch feature/refactor-execute-process when this is ready, and we can start merging PRs there.

hidmic

Overall it's looking good @roger-strain. I don't have much more to say on top of what @ivanpauno already pointed out.

hidmic · 2020-08-24T21:54:52Z

launch/launch/descriptions/executable.py

+        process_event_args = self.__process_event_args
+        if process_event_args is None:


It doesn't seem like it. Also, as personal nit, I'd much rather have __expand_substitutions return something to apply_context instead of the former mutating state directly.

hidmic · 2020-08-24T22:03:05Z

launch/launch/descriptions/executable.py

+    @property
+    def name(self):
+        """Getter for name."""
+        return self.__name


In an unrelated (or maybe not quite unrelated) note, launch and launch_ros actions are inconsistent w.r.t how action properties interact with the action lifecycle:

launch.actions.ExecuteProcess.name gets you a normalized substitution, always.

launch_ros.actions.Node.node_name raises if you try to get it before executing the action.

launch_ros.parameter_description.Parameter.name, though arguably not an action, introduces yet another variation and gets you a normalized substitution before evaluating the parameter and the actual name after evaluating it.

@ivanpauno @wjwwood perhaps we should settle this here and stick to it.

@hidmic Would it be worth introducing more explicitly named parameters instead, to clarify which thing you're requesting? Maybe name always returns the value as it was originally specified, without applying substitutions, and applied_name returns the one with substitutions applied to it, raising an exception if substitutions haven't been performed yet?

@hidmic Would it be worth introducing more explicitly named parameters instead, to clarify which thing you're requesting? Maybe name always returns the value as it was originally specified, without applying substitutions, and applied_name returns the one with substitutions applied to it, raising an exception if substitutions haven't been performed yet?

I think that's ok.
bikeshed: I'm not convinced about naming, maybe final_name sounds better than applied_name. I don't mind much.

I've taken a stab at this in and updated. In reviewing how the objects were used, I don't think this was the right place to be creating the packed dictionary anyway, so I've removed that; it'll probably come back in ExecuteLocal. For all the parameters which can have substitutions applied, there's now a "final_" version which can be accessed. Right now, that isn't throwing an error if the substitutions haven't been applied, but I can make it do so if that's the standard approach.

It's OK for now. It's not up to you to solve these discrepancies anyways.

roger-strain · 2020-09-22T21:08:25Z

@ivanpauno I've taken a first stab at trying to refactor the main execute_process class into the new execute_local. I spent quite a while trying to figure out how to write some unit tests to actually verify it's doing what it should, but in all honesty, this part of the system is still pretty unfamiliar to me, and I wasn't able to come up with any good samples to follow. The code itself is by and large copied straight from the original class, so I'm very happy to try to clean things up and improve them from this point, so any direction there would be appreciated, as well as suggestions on how to put some useful tests together.

ivanpauno

This looks good to me.

The only missing thing for merging is refactoring ExecuteProcess to reuse ExecuteLocal and Executable. If all previous tests pass, I think this would be ready to merge.

It would also be great to have some direct testing of ExecuteLocal (and more extended testing of Executable)

ivanpauno · 2020-09-28T13:37:11Z

launch/launch/descriptions/executable.py

+        cmd = shlex.split(cmd)
+        self.__final_cmd = cmd


I think you cannot handle both the "old style cmd" and the "new style cmd" in the same argument.
If you want, you can ignore this change and keep the old cmd style, that can be changed later.

e.g.:

['cmd', TextSubstitution(text='my arg with spaces')]

was used like: ['cmd', 'my arg with spaces']
with the new logic, it should be used like: ['cmd', 'my', 'arg', 'with', 'spaces']

Changing the semantics of cmd from a "iterable of substitutions" to a single substitution is a breaking change.
With the new semantics, the user must quote "my arg with spaces" if a single argument is desired.

So, it we want to keep backwards compatibility, we need two arguments in the constructor:

cmd: SomeSubstitutionsType, cmd_old: Iterable[SomeSubstitutionsType]

Users would use cmd directly, and we would used cmd_old through ExecuteProcess to keep backwards compatibility (and of course, some logic is needed in the constructor to ensure only one was passed).

I think in the interest of keeping this PR simpler, I'll err on the side of just retaining the old style for cmd. I'd like to be able to help pick up the other issue as well, but it's probably better to let that be a separate task down the line.

ivanpauno · 2020-09-28T14:18:03Z

launch/test/test_executable.py

+def test_cmd_multiple_arguments_in_string():
+    exe = Executable(cmd=['ls', '-opt1', '-opt2', '-opt3'])
+    exe.apply_context(LaunchContext())
+    assert all([a == b for a, b in zip(exe.final_cmd, ['ls', '-opt1', '-opt2', '-opt3'])])


you could also have tests for final_name, final_cwd and final_env.

ivanpauno · 2020-09-28T14:18:50Z

Note: Linters aren't passing, check http://build.ros2.org/job/Rpr__launch__ubuntu_focal_amd64/82/.

roger-strain · 2020-09-28T15:44:29Z

The only missing thing for merging is refactoring ExecuteProcess to reuse ExecuteLocal and Executable. If all previous tests pass, I think this would be ready to merge.

That's the plan for the next step, but I wanted to make sure things looked sane to this point first.

It would also be great to have some direct testing of ExecuteLocal (and more extended testing of Executable)

I would absolutely love to get more testing in place, but I'm kind of lost on how to best accomplish that. If there are some samples I could work from that show testing execution of things, I would really love to dig into that so I can apply it to this. Actually spinning up external processes as part of unit testing seems a little dicey, especially considering the variety of platforms the system should work on. If you've got any guidance, it'd be a huge help.

roger-strain · 2020-12-01T21:55:41Z

@ivanpauno Pushed part of this a couple of weeks ago, and then commented on the wrong PR. Sorry about that. I think the unit tests are happy now; I'm still not 100% confident, but I'm not sure I ever will be. When you have a chance, give it a look over and see what you think. I'm going to take this out of draft status at this point.

hidmic · 2020-12-02T18:38:44Z

launch/launch/actions/execute_process.py

-        self.__stderr_buffer = io.StringIO()
-
-        self.__executed = False
+        self.__executable = Executable(cmd=cmd, prefix=prefix, name=name, cwd=cwd, env=env,


@roger-strain nit: it'd seem to me that self.__executable is redundant with self.process_description

hidmic · 2020-12-02T18:43:55Z

launch/test/launch/test_executable.py

+# Delivered to the U.S. Government with Unlimited Rights, as defined in DFARS
+# Part 252.227-7013 or 7014 (Feb 2014).
+#
+# This notice must appear in all copies of this file and its derivatives.


@tfoote would you give this preamble a look? I'm not well versed in such licensing qualifications.

@tfoote friendly ping

@roger-strain Does this notice have to be on these files? It is much easier for us if this is just not there; then it looks the same as all of the rest of the licensing in our code and isn't a special snowflake. If it absolutely has to be there for some reason, I'm going to have to check with some others about whether this is allowed in this codebase.

@clalancette I've asked back up my chain to revisit this. I will say that we tried to come up with something that would meet the requirements we were given, but also would be compatible with existing licensing here. That being said, we're looking again. Is there a specific part of it that's especially troublesome, or is the whole thing the issue? Might help to focus the discussions if I can target things appropriately.

It's really the whole thing. The basic problem is that we are not lawyers, and so that anything that deviates from the standard license text is a problem.

Please let me know what you decide. Once you've come to your decision, I'll pass it along to those more in the know on this side. Thanks.

Just to follow up, we're still in active discussions about how to sort this out. Appreciate your patience.

hidmic · 2020-12-02T18:45:11Z

launch/test/launch/test_executable.py

+def test_cmd_multiple_arguments_in_string():
+    exe = Executable(cmd=['ls', '-opt1', '-opt2', '-opt3'])
+    exe.apply_context(LaunchContext())
+    assert all([a == b for a, b in zip(exe.final_cmd, ['ls', '-opt1', '-opt2', '-opt3'])])


@roger-strain tests for other arguments besides cmd and/or using substitutions would be most welcome.

hidmic · 2020-12-02T18:46:03Z

launch/launch/descriptions/executable.py

+    @property
+    def name(self):
+        """Getter for name."""
+        return self.__name


It's OK for now. It's not up to you to solve these discrepancies anyways.

hidmic · 2020-12-02T18:47:45Z

launch/launch/actions/execute_local.py

+    def prepare(self, context: LaunchContext):
+        """Prepare the action for execution."""
+        self.__process_description.apply_context(context)
+        self.__expand_substitutions(context)


@roger-strain meta, safe to ignore: I'm not a huge fan of methods with side effects like this. Purity is more expensive but less surprising.

hidmic · 2020-12-02T18:49:21Z

@roger-strain I think we'll have a better grasp of the refactor once a somewhat larger portion of it lands. Are you looking forward to continue your contributions in the near future?

roger-strain · 2020-12-03T20:19:57Z

@roger-strain I think we'll have a better grasp of the refactor once a somewhat larger portion of it lands. Are you looking forward to continue your contributions in the near future?

@hidmic Yes, I'm hoping to get the next section of this next week. This is the only PR that will be targeting launch itself; the rest are going to be in the launch_ros side of things. When we were discussing the timeline/order of operations for the refactor, though, it was suggested I get this side of things sorted first, and backward compatible, so that was the first focus.

hidmic · 2021-01-11T21:32:02Z

@roger-strain friendly ping

roger-strain · 2021-01-11T21:38:21Z

@roger-strain friendly ping

Sorry, holidays caught me off guard and I'm now in the process of getting my laptop rebuilt. I believe I have the code finished for the next update on this and the first corresponding PR over in launch_ros, but I need to get more unit tests in place before I'm ready to officially push them up. Targeting later this week, assuming nothing else blows up on me in the process.

roger-strain · 2021-01-25T21:59:00Z

@hidmic Finally finished a big chunk of work over in ros2/launch_ros#215, and updated this one to go alongside it. Please take a look and see how you think it's looking. Still need to go in and do the bits for Composable and Lifecycle nodes, and I'm sure there are other bits that I've missed, but hopefully this is pretty close to what you're looking for otherwise.

roger-strain · 2021-02-09T17:45:40Z

@hidmic Some more updates over here; little bit of cleanup, and also part of the refactor discussed in ros2/launch_ros#215 regarding the proper place for arguments; they now are handled at the Executable level.

hidmic

@roger-strain my since apologies for not coming back to this earlier. Too many things to do.

Nothing specifically added by this patch calls my attention in a negative way. Its PR job doesn't seem happy though. There's plenty of failing tests. Mind to take a look?

hidmic

Thanks for the bump @mlanting. Overall LGTM pending green CI

Linux
Linux-aarch64
macOS
Windows

hidmic · 2021-10-22T17:26:27Z

launch/launch/descriptions/executable.py

+    """Describes an executable (usually a single process) which may be run by the launch system."""
+
+    def __init__(
+            self, *,


@mlanting this still applies.

hidmic · 2021-10-22T17:26:53Z

launch/launch/descriptions/executable.py

+        :param cmd: A list where the first item is the executable and the rest are
+            arguments to the executable, each item may be a string or a list of strings
+            and Substitutions to be resolved at runtime
+        :param: prefix a set of commands/arguments to preceed the cmd, used for


@mlanting nit:

Suggested change

:param: prefix a set of commands/arguments to preceed the cmd, used for

:param prefix: a set of commands/arguments to preceed the cmd, used for

hidmic · 2021-10-22T17:27:17Z

launch/launch/descriptions/executable.py

+        :param additional_env: Dictionary of environment variables to be added. If env was
+            None, they are added to the current environment. If not, env is updated with
+            additional_env.
+        :param: arguments list of extra arguments for the executable


@mlanting nit:

Suggested change

:param: arguments list of extra arguments for the executable

:param arguments: list of extra arguments for the executable

hidmic · 2021-10-22T17:29:33Z

launch/launch/event_handlers/on_process_exit.py

@@ -45,15 +45,15 @@ class OnProcessExit(BaseEventHandler):
    def __init__(
        self,
        *,
-        target_action: 'ExecuteProcess' = None,
+        target_action: 'ExecuteLocal' = None,


This still applies. OK to defer though.

hidmic · 2021-10-25T21:33:25Z

@mlanting the sole test failure is unrelated. Mind to address the nits? Then we can (finally) merge.

mlanting · 2021-10-27T19:26:35Z

@hidmic I made the suggested changes regarding the colon placement. The indentation in the launch/launch/descriptions/executable.py file seems correct to me, according to the PEP8 guidelines (https://www.python.org/dev/peps/pep-0008/#indentation) - since that is a hanging indent for function params, it should be indented an extra level. Given that, the lines in on_process_exit should be indented further as well, but I wanted to check which way you'd prefer before I changed it one way or the other.

I'll defer the abstract Execute action to when I add in the remote launching classes.

hidmic · 2021-10-28T18:17:39Z

The indentation in the launch/launch/descriptions/executable.py file seems correct to me, according to the PEP8 guidelines

@mlanting you're right about PEP8 indentation, but I will point out that Executable.__init__, and pretty much every function and method in launch, aligns the closing parenthesis with the def keyword in a new line. The readability issue that double indentation would solve is not there.

Linters are happy, but I'm somewhat inclined towards being consistent with the rest of the codebase.

I'll defer the abstract Execute action to when I add in the remote launching classes.

@mlanting Agreed.

BTW this needs a rebase!

Part of implementation of refactor for ros2#114 Distro A; OPSEC #2893 Signed-off-by: Roger Strain <[email protected]>

Modified handling of cmd parameter as described in ros2#263 Distro A; OPSEC #2893 Signed-off-by: Roger Strain <[email protected]>

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Distro A, OPSEC #4584. You may have additional rights; please see https://rosmilitary.org/faq/?category=ros-2-license Signed-off-by: matthew.lanting <[email protected]>

mlanting · 2021-10-28T19:44:00Z

@hidmic I've fixed the indentation and done the rebase.

hidmic

Alright, I think this is good to go. Thank you for your patience @mlanting!

On a related note, when do you expect moving forward with the rest? Humble is still quite a few months away but it'd be great to land the full refactor @roger-strain intended to land before then.

mlanting · 2021-10-29T13:36:20Z

I'm going to start moving forward with it immediately. Once this is merged, I'll start working on getting the launch_ros changes rebased so we can get those added, and then start work on the rest.

ivanpauno · 2021-10-29T14:00:43Z

launch/launch/actions/execute_process.py

 _global_process_counter_lock = threading.Lock()
 _global_process_counter = 0  # in Python3, this number is unbounded (no rollover)


nit: these are unused after this PR, and could be deleted

ivanpauno · 2021-10-29T14:01:08Z

launch/launch/actions/execute_local.py

+_global_process_counter_lock = threading.Lock()
+_global_process_counter = 0  # in Python3, this number is unbounded (no rollover)


these globals are not being used anywhere

Ahh, I overlooked those. Thanks for pointing it out @ivanpauno.

@mlanting mind to fix it?

wjwwood · 2023-08-16T21:45:22Z

launch/launch/descriptions/executable.py

+        :param additional_env: Dictionary of environment variables to be added. If env was
+            None, they are added to the current environment. If not, env is updated with
+            additional_env.
+        :param arguments: list of extra arguments for the executable


I know this issue is quite old, but what's the purpose of this argument? Why could they not be part of the cmd argument?

ivanpauno added the enhancement New feature or request label Aug 24, 2020

ivanpauno self-requested a review August 24, 2020 16:44

ivanpauno reviewed Aug 24, 2020

View reviewed changes

ivanpauno requested a review from hidmic August 24, 2020 16:54

hidmic reviewed Aug 24, 2020

View reviewed changes

roger-strain force-pushed the execute-process-refactor branch from 7693958 to 8b9f068 Compare September 22, 2020 21:08

ivanpauno reviewed Sep 28, 2020

View reviewed changes

roger-strain marked this pull request as ready for review December 1, 2020 21:55

hidmic reviewed Dec 2, 2020

View reviewed changes

clalancette assigned hidmic Dec 18, 2020

roger-strain force-pushed the execute-process-refactor branch from bc27324 to 265ed45 Compare January 25, 2021 21:49

roger-strain mentioned this pull request Jan 25, 2021

Execute process refactor ros2/launch_ros#215

Open

roger-strain force-pushed the execute-process-refactor branch 2 times, most recently from a408d92 to 7f21a4f Compare March 3, 2021 16:15

hidmic reviewed Mar 9, 2021

View reviewed changes

roger-strain force-pushed the execute-process-refactor branch from a601e00 to dc7f80b Compare March 10, 2021 14:20

mlanting force-pushed the execute-process-refactor branch from 972082b to bc185af Compare August 4, 2021 17:15

hidmic reviewed Oct 22, 2021

View reviewed changes

Roger Strain and others added 16 commits October 28, 2021 15:24

Adding Executable description class

980377a

Part of implementation of refactor for ros2#114 Distro A; OPSEC #2893 Signed-off-by: Roger Strain <[email protected]>

Cleaned up access to substituted values

3748c2d

Modified handling of cmd parameter as described in ros2#263 Distro A; OPSEC #2893 Signed-off-by: Roger Strain <[email protected]>

Initial implementation of execute_local

161a53c

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Fixed minor bugs to verify unit tests

0d5a63f

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Unit test fixes

62e92e7

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Adjust for launch_ros modifications, add unit tests

e17d4a6

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Apply minor updates to address feedback

3b0659d

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Refactor arguments to apply to executable

28720cf

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Fix order of parameters

dca16b9

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Fixed default log name prefix

265ea48

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Fix unit tests

59a0f2d

Distro A; OPSEC #4584 Signed-off-by: Roger Strain <[email protected]>

Removed some language from file headers

4bf31f4

Distro A, OPSEC #4584. You may have additional rights; please see https://rosmilitary.org/faq/?category=ros-2-license Signed-off-by: matthew.lanting <[email protected]>

Add prefix filtering to Executable description class.

106ac78

Distro A, OPSEC #4584. You may have additional rights; please see https://rosmilitary.org/faq/?category=ros-2-license Signed-off-by: matthew.lanting <[email protected]>

Formatting fixes.

2b16262

Distro A, OPSEC #4584. You may have additional rights; please see https://rosmilitary.org/faq/?category=ros-2-license Signed-off-by: matthew.lanting <[email protected]>

Fixed indentation in descriptions/executable.py.

cc8f930

Distro A, OPSEC #4584. You may have additional rights; please see https://rosmilitary.org/faq/?category=ros-2-license Signed-off-by: matthew.lanting <[email protected]>

Fixed namechange missed during rebase.

aa4d847

Distro A, OPSEC #4584. You may have additional rights; please see https://rosmilitary.org/faq/?category=ros-2-license Signed-off-by: matthew.lanting <[email protected]>

mlanting force-pushed the execute-process-refactor branch from 4effcb7 to aa4d847 Compare October 28, 2021 19:39

hidmic approved these changes Oct 28, 2021

View reviewed changes

hidmic merged commit 54f93b9 into ros2:master Oct 29, 2021

ivanpauno reviewed Oct 29, 2021

View reviewed changes

ivanpauno mentioned this pull request Oct 29, 2021

First prototype of native pytest plugin for launch based tests #528

Merged

This was referenced Apr 28, 2022

Remove unused variables. #612

Merged

ROS 2 Humble needed release notes ros2/ros2_documentation#2446

Closed

wjwwood reviewed Aug 16, 2023

View reviewed changes

		process_event_args = self.__process_event_args
		if process_event_args is None:

	process_event_args = self.__process_event_args
	if process_event_args is None:
	if self.__process_event_args is None:

	:param: prefix a set of commands/arguments to preceed the cmd, used for
	:param prefix: a set of commands/arguments to preceed the cmd, used for

	:param: arguments list of extra arguments for the executable
	:param arguments: list of extra arguments for the executable

		_global_process_counter_lock = threading.Lock()
		_global_process_counter = 0 # in Python3, this number is unbounded (no rollover)

Adding Executable description class #454

Adding Executable description class #454

Conversation

roger-strain commented Aug 19, 2020

roger-strain commented Aug 19, 2020

ivanpauno left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ivanpauno commented Aug 24, 2020

hidmic left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hidmic Aug 24, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

roger-strain commented Sep 22, 2020

ivanpauno left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ivanpauno commented Sep 28, 2020

roger-strain commented Sep 28, 2020

roger-strain commented Dec 1, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hidmic commented Dec 2, 2020

roger-strain commented Dec 3, 2020

hidmic commented Jan 11, 2021

roger-strain commented Jan 11, 2021

roger-strain commented Jan 25, 2021

roger-strain commented Feb 9, 2021

hidmic left a comment

Choose a reason for hiding this comment

hidmic left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hidmic commented Oct 25, 2021

mlanting commented Oct 27, 2021

hidmic commented Oct 28, 2021 • edited Loading

mlanting commented Oct 28, 2021

hidmic left a comment

Choose a reason for hiding this comment

mlanting commented Oct 29, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hidmic left a comment •

edited

Loading

hidmic Aug 24, 2020 •

edited

Loading

hidmic commented Oct 28, 2021 •

edited

Loading