Bug fix macros.measure with backendv2 #9987

to24toro · 2023-04-19T04:27:55Z

Summary

Currently, we cannot use macros.measure with backendV2, but it is clear that this is an emergency issue that needs to be resolved.
After the discussion, the plan is to update macros.measure significantly by adding a new class MeasureGrouping. However, this is a major API break issue and will not be ready for the recent 0.25.0 release.
Therefore, as a temporary, I made changes as a version that do not break the API.

Details and comments

fix #9488

Add meas_map as a new attribute of Target. There is another part of the code that creates an instance of target, however I have not changed it yet now.
In the measure in macors.py, dispatch measure_v1 and measure_v2 depending on the class of the backend.
FakeHanoiV2 is used as backendV2 in test.

…p_class

macros.py target delete raise statement in measure_v2 modify instructions.Acquire to Acquire

qiskit-bot · 2023-04-19T04:28:04Z

Thank you for opening a new pull request.

Before your PR can be merged it will first need to pass continuous integration tests and be reviewed. Sometimes the review process can be slow, so please be patient.

While you're waiting, please feel free to review other open PRs. While only a subset of people are authorized to approve pull requests for merging, everyone is encouraged to review open pull requests. Doing reviews helps reduce the burden on the core team and helps make the project's code better for everyone.

One or more of the the following people are requested to review this:

@Qiskit/terra-core
@nkanazawa1989

ikkoham · 2023-04-19T08:57:30Z

qiskit/pulse/macros.py

+        "Depricating ``backendV1`` as the type of measure's `backend` argument."
+    ),
+    additional_msg=("Instead use ``backendV2``as the type of measure's `backend` argument."),
+    since="0.25.0",


Suggested change

since="0.25.0",

since="0.24.0",

0.25 or 0.24?

TBH, I would just drop the since field as this is a pending deprecation it's not on any timer and by default isn't user facing so people knowing that we're planning to deprecate this in the future doesn't really have a start date as there is no timer associated with it yet.

That being said I don't think we need this at all. BackendV1 is still a support interface in Qiskit we haven't marked it as deprecated or anything yet so I would just remove the use of the deprecation decorator here since I'd expect the pulse builder to support both v1 and v2.

That's true

I deleted the deprecate decorator because "since" field must be provided.

@Eric-Arellano here is another case we need to consider :) just fyi

Yes, the Sphinx deprecated directive expects us to always have a since, so that is why it's required.

The docstring for since says this:

since: The version the deprecation started at. If the deprecation is pending, set the version to when that started; but later, when switching from pending to deprecated, update `since` to the new version.

I deleted the deprecate decorator because "since" field must be provided.

I encourage you to still use it and set pending=True. That has some benefits:

The message will be standardized

The pending deprecation will show up in our docs. (It will say its pending, not actually deprecated)

It's easy to switch from pending to deprecated. You delete pending=True and bump since to the version it became deprecated.

nkanazawa1989

This is awesome @to24toro ! We can support circuit scheduling with V2 backend with this PR! Proposed implementation looks good, but some minor edits are requested to approve. @mtreinish Can we still add this PR to 0.24 as a bugfix related to BackendV2?

nkanazawa1989 · 2023-04-19T11:04:40Z

qiskit/providers/backend_compat.py

@@ -56,7 +56,10 @@ def convert_to_target(
    # Parse from properties if it exsits
    if properties is not None:
        qubit_properties = qubit_props_list_from_props(properties=properties)
-        target = Target(num_qubits=configuration.n_qubits, qubit_properties=qubit_properties)
+        target = Target(


Please revert this change.

I fixed it at 3926200.

nkanazawa1989 · 2023-04-19T11:10:41Z

qiskit/pulse/macros.py

+    If the backend has an attribute ``target``, the function uses the measurement logic,
+    "_measure_v2" that takes ``target`` of the ``backend``, ``meas_map`` and ``qubit_mem_slots``
+    assignment.
+    Otherwise, if the backend is None or ``backendV1``, the function uses the
+    measurement logic, "_measure_v1" including ``instruction_schedule_map`` and ``meas_map``
+    as inputs.


Suggested change

If the backend has an attribute ``target``, the function uses the measurement logic,

"_measure_v2" that takes ``target`` of the ``backend``, ``meas_map`` and ``qubit_mem_slots``

assignment.

Otherwise, if the backend is None or ``backendV1``, the function uses the

measurement logic, "_measure_v1" including ``instruction_schedule_map`` and ``meas_map``

as inputs.

.. note::

This function internally dispatches schedule generation logic depending on input backend model.

For the :class:`.BackendV1`, it considers conventional :class:`.InstructionScheduleMap`

and utilizes the backend calibration defined for a group of qubits in the `meas_map`.

For the :class:`.BackendV2`, it assembles calibrations of single qubit measurement

defined in the backend target to build a composite measurement schedule for `qubits`.

This doesn't need to be written here. Note that this is API document that end-users read, so implementation details must be avoided.

I fixed at 4029849.

nkanazawa1989 · 2023-04-19T11:19:18Z

qiskit/pulse/macros.py

+    qubit_mem_slots: Dict[int, int],
+    measure_name: str = "measure",
+) -> Schedule:
+    """Return a schedule which measures the requested qubits according to the given


nkanazawa1989 · 2023-04-19T11:19:49Z

qiskit/pulse/macros.py

+        backend (Union[Backend, BaseBackend]): A backend instance, which contains
+            hardware-specific data required for scheduling.


Suggested change

backend (Union[Backend, BaseBackend]): A backend instance, which contains

hardware-specific data required for scheduling.

This doesn't exist.

I fixed at 4029849.

nkanazawa1989 · 2023-04-19T11:22:53Z

qiskit/pulse/macros.py

+                        channels.AcquireChannel(measure_qubit),
+                    ]
+                )
+        except exceptions.PulseError as ex:


Suggested change

except exceptions.PulseError as ex:

except KeyError as ex:

Target doesn't raise PulseError
https://github.com/Qiskit/qiskit-terra/blob/5128c6751fc2909131ab38c72358bb1e91c9fd84/qiskit/transpiler/target.py#L904-L906

I fixed at 4029849.

nkanazawa1989 · 2023-04-19T11:29:01Z

qiskit/pulse/macros.py

+                    channels.AcquireChannel(qubit_index),
+                    mem_slot=channels.MemorySlot(reg_index),
+                ),
+                inplace=True,


nkanazawa1989 · 2023-04-19T11:29:14Z

qiskit/transpiler/target.py

@@ -280,6 +280,7 @@ def __init__(
                matches the qubit number the properties are defined for. If some
                qubits don't have properties available you can set that entry to
                ``None``
+            meas_map (list): List of sets of qubits that must be measured together.


Please revert this file

I fixed it at 3926200.

nkanazawa1989 · 2023-04-19T11:30:20Z

qiskit/pulse/macros.py

+        qubit_mem_slots: Mapping of measured qubit index to classical bit index.
+
+    Returns:
+        A schedule remapped by qubit_mem_slots as the input provided.


Suggested change

A schedule remapped by qubit_mem_slots as the input provided.

A measurement schedule with new memory slot index.

I fixed at 4029849.

nkanazawa1989 · 2023-04-19T11:31:39Z

test/python/pulse/test_macros.py

@@ -91,6 +92,91 @@ def test_fail_measure(self):
        with self.assertRaises(PulseError):
            macros.measure(qubits=[0], inst_map=self.inst_map)

+    def test_measure_v2(self):
+        """Test macro - measure with backendV2."""
+        sched = macros.measure(qubits=[0], backend=self.backend_v2).filter(


Can you add another test with multiple qubits? Also it's better not to filter output (i.e. testing full schedule equality), because some future PR may break this function by injecting invalid instruction. This should be captured by this test.

I added a multiple test at 381a8e6.

nkanazawa1989 · 2023-04-19T11:33:17Z

test/python/pulse/test_macros.py

+            (
+                0,
+                Play(
+                    GaussianSquare(


You cannot hard-code these values because backend calibration snapshot may be updated in future.

I fixed the test at 381a8e6.

nkanazawa1989 · 2023-04-19T11:44:10Z

qiskit/pulse/macros.py

@@ -104,3 +232,33 @@ def measure_all(backend) -> Schedule:
        A schedule corresponding to the inputs provided.
    """
    return measure(qubits=list(range(backend.configuration().n_qubits)), backend=backend)
+
+
+def schedule_remapping_memory_slot(schedule: Schedule, qubit_mem_slots: Dict[int, int]) -> Schedule:


@TsafrirA are you happy with this API? This is public so we cannot easily change after release. If you have any concern we can turn this into protected.

I would make this private just at this point in the release cycle if you want to include this for 0.24.0. It's best not to commit to something like this when there is a pending deadline like this. If we make it _schedule_remapping_memory_slot this is gives a bit more time because there is no new public api addition so we can include this post RC1. Then we can always promote it to a public API in 0.25.0

Make perfect sense :)

I changed at b075cb3.

mtreinish

As an isolated bugfix this is fine to include in 0.24.0. Without any API additions it's something we can also merge for 0.24.0 post rc1. I left a couple of small inline comments. Besides that I think a bugfix release release note would be good to document that the schedule is fixed when being passed a backend v2 object.

mtreinish · 2023-04-19T12:04:44Z

qiskit/pulse/macros.py

+        "Depricating ``backendV1`` as the type of measure's `backend` argument."
+    ),
+    additional_msg=("Instead use ``backendV2``as the type of measure's `backend` argument."),
+    since="0.25.0",


TBH, I would just drop the since field as this is a pending deprecation it's not on any timer and by default isn't user facing so people knowing that we're planning to deprecate this in the future doesn't really have a start date as there is no timer associated with it yet.

That being said I don't think we need this at all. BackendV1 is still a support interface in Qiskit we haven't marked it as deprecated or anything yet so I would just remove the use of the deprecation decorator here since I'd expect the pulse builder to support both v1 and v2.

mtreinish · 2023-04-19T12:10:03Z

qiskit/pulse/macros.py

@@ -104,3 +232,33 @@ def measure_all(backend) -> Schedule:
        A schedule corresponding to the inputs provided.
    """
    return measure(qubits=list(range(backend.configuration().n_qubits)), backend=backend)
+
+
+def schedule_remapping_memory_slot(schedule: Schedule, qubit_mem_slots: Dict[int, int]) -> Schedule:


I would make this private just at this point in the release cycle if you want to include this for 0.24.0. It's best not to commit to something like this when there is a pending deadline like this. If we make it _schedule_remapping_memory_slot this is gives a bit more time because there is no new public api addition so we can include this post RC1. Then we can always promote it to a public API in 0.25.0

nkanazawa1989 · 2023-04-20T03:08:26Z

qiskit/pulse/macros.py

+    for t0, inst in schedule.instructions:
+        if isinstance(inst, instructions.Acquire):
+            qubit_index = inst.channel.index
+            reg_index = qubit_mem_slots.get(qubit_index, 0)


The unittest you wrote made me think this should be

Suggested change

reg_index = qubit_mem_slots.get(qubit_index, 0)

reg_index = qubit_mem_slots.get(qubit_index, qubit_index)

because all acquisition instruction might store results in the slot0 when you specify qubit_mem_slots={0:0} (i.e. you may provide the slot only for qubit of interested). This causes data conflict on the slot and you may get meaningless outcome from experiment.

nkanazawa1989

Thanks @to24toro this looks almost good. Just minor comments. I'm happy to approve once these are addressed.

nkanazawa1989 · 2023-04-20T05:30:11Z

qiskit/pulse/macros.py

+    meas_group = sorted(list(meas_group))
+
+    meas_group_set = set(range(max(meas_group) + 1))
+    unassigned_qubit_indices = list(meas_group_set - qubit_mem_slots.keys())


Suggested change

unassigned_qubit_indices = list(meas_group_set - qubit_mem_slots.keys())

unassigned_qubit_indices = list(set(meas_group) - qubit_mem_slots.keys())

I think this should work because qubit outside this group doesn't appear.

nkanazawa1989 · 2023-04-20T05:33:34Z

qiskit/pulse/macros.py

+    unassigned_reg_indices = sorted(list(meas_group_set - set(qubit_mem_slots.values())))
+    if set(qubit_mem_slots.values()).issubset(meas_group_set):
+        for qubit in unassigned_qubit_indices:
+            qubit_mem_slots[qubit] = unassigned_reg_indices.pop(0)


This is not blocking but usually pop(0) is slow. For better performance you need to reverse sort unassigned_reg_indices and just do .pop() here.

nkanazawa1989 · 2023-04-20T05:35:57Z

Could you please also write bugfix note?

nkanazawa1989

LGTM

* create measuregrouping class * add meas_map.setter in MeasureGrouping class * macros.measure * get_qubit_groups * generate_schedule * target.add_measuregrouping in backend_compat * target.add_measuregrouping in backend_converter * reformat and add docs * on the way of working on generate_schedule_in_measure * split measure into measure_v1 and measure_v2 macros.py target delete raise statement in measure_v2 modify instructions.Acquire to Acquire * macros.py * test_measuregrouping * bug fix schedule with backendV2 for 0.25.0 * modify comments * fix name of schedule in test_macros * delete meas_map as a Target attribute * minor changes in macros.py * add test to test_macros.py * make schedule_remapping_memory_slot private * delete since field from deprecate_arg * delete deprecate depcorator * black macros.py * revert about target * modify implementation of qubit_mem_slots * change the definition of meas_group_set * black macros.py * fix meas_group_set * fix qubit_mem_slots * reno * modify unassigned_qubit_indices * remove list() from unassigned_qubit_indices and unassigned_reg_indices

fvoichick · 2023-07-09T21:29:16Z

I'm relatively inexperienced with Qiskit and I think I'm still running into this issue even after building from source. Could someone please help me or direct me to the appropriate place to ask?

I'm running this code:

from qiskit import schedule
from qiskit.circuit import QuantumCircuit
from qiskit_ibm_provider import IBMProvider

provider = IBMProvider()
backend = provider.get_backend('ibmq_quito')
circuit = QuantumCircuit(1, 1)
circuit.measure(0, 0)
schedule(circuit, backend)

I'm getting this error message:

Traceback (most recent call last):
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/pulse/macros.py", line 129, in _measure_v1
default_sched = inst_map.get(measure_name, measure_group_qubits)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/pulse/instruction_schedule_map.py", line 195, in get
return self._get_calibration_entry(instruction, qubits).get_schedule(*params, **kwparams)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/pulse/instruction_schedule_map.py", line 219, in _get_calibration_entry
self.assert_has(instruction, qubits)
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/pulse/instruction_schedule_map.py", line 166, in assert_has
raise PulseError(
    qiskit.pulse.exceptions.PulseError: "Operation 'measure' exists, but is only defined for qubits [0, 1, 2, 3, 4]."

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "/home/finn/Desktop/qiskit/stretch/meas.py", line 9, in <module>
                                                              schedule(circuit, backend)
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/compiler/scheduler.py", line 101, in schedule
schedules = parallel_map(schedule_circuit, circuits, (schedule_config, method))
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
            File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/tools/parallel.py", line 152, in parallel_map
return [task(values[0], *task_args, **task_kwargs)]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/scheduler/schedule_circuit.py", line 59, in schedule_circuit
return methods[method](circuit, schedule_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/scheduler/methods/basic.py", line 103, in as_late_as_possible
circ_pulse_defs = lower_gates(circuit, schedule_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/scheduler/lowering.py", line 177, in lower_gates
circ_pulse_defs.append(get_measure_schedule(qubit_mem_slots))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/scheduler/lowering.py", line 98, in get_measure_schedule
meas_sched = measure(
             ^^^^^^^^
             File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/pulse/macros.py", line 78, in measure
return _measure_v1(
    ^^^^^^^^^^^^
    File "/home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages/qiskit/pulse/macros.py", line 131, in _measure_v1
raise exceptions.PulseError(
    qiskit.pulse.exceptions.PulseError: "We could not find a default measurement schedule called 'measure'. Please provide another name using the 'measure_name' keyword argument. For assistance, the instructions which are defined are: ['sx', 'cx', 'measure', 'id', 'x', 'rz']"

I followed the "Install from source" instructions at https://qiskit.org/documentation/getting_started.html, I'm using the updated main branch, and it looks like I'm using the right version:

$ pip show qiskit-terra
Name: qiskit-terra
Version: 0.25.0
Summary: Software for developing quantum computing programs
Home-page: https://github.com/Qiskit/qiskit-terra
Author: Qiskit Development Team
Author-email: [email protected]
License: Apache 2.0
Location: /home/finn/Desktop/qiskit/source/QiskitDevenv/lib/python3.11/site-packages
Requires: dill, numpy, ply, psutil, python-dateutil, rustworkx, scipy, stevedore, symengine, sympy
Required-by: qiskit-ibm-provider

This problem was fixed, right? Anyone have ideas about why this still isn't working? Let me know if there's somewhere else I should ask.

to24toro · 2023-07-10T02:39:27Z

@fvoichick
Thank you for asking and sorry for this bug.
Certainly your point is right and we have not fixed scheduling with some backends yet, I will work on some issues related to this bug.

[WIP]Schedule with backendV2 #10285 (this PR fixes the bug you pointed out)

We would appreciate it if you could wait a little longer.

to24toro added 15 commits April 5, 2023 10:44

create measuregrouping class

4bab6a6

add meas_map.setter in MeasureGrouping class

9d109d0

macros.measure

a40211c

get_qubit_groups

25d0719

generate_schedule

11f734b

target.add_measuregrouping in backend_compat

ca3f032

target.add_measuregrouping in backend_converter

2d2ff79

reformat and add docs

0d4b715

Merge remote-tracking branch 'origin/main' into feature/add_meas_grou…

d223f4f

…p_class

on the way of working on generate_schedule_in_measure

d29025f

split measure into measure_v1 and measure_v2

68ec213

macros.py target delete raise statement in measure_v2 modify instructions.Acquire to Acquire

macros.py

d19464b

test_measuregrouping

5930369

bug fix schedule with backendV2 for 0.25.0

e53f226

modify comments

c1a51d5

to24toro marked this pull request as ready for review April 19, 2023 04:29

to24toro requested review from a team, eggerdj, wshanks and jyu00 as code owners April 19, 2023 04:29

fix name of schedule in test_macros

f31ae23

to24toro changed the title ~~Feature/bug fix schedule with backendv2~~ Bug fix schedule with backendv2 Apr 19, 2023

ikkoham added this to the 0.25.0 milestone Apr 19, 2023

ikkoham added Changelog: Deprecation Include in "Deprecated" section of changelog Changelog: New Feature Include in the "Added" section of the changelog Changelog: Bugfix Include in the "Fixed" section of the changelog mod: pulse Related to the Pulse module labels Apr 19, 2023

ikkoham reviewed Apr 19, 2023

View reviewed changes

ikkoham removed the Changelog: Deprecation Include in "Deprecated" section of changelog label Apr 19, 2023

nkanazawa1989 requested changes Apr 19, 2023

View reviewed changes

nkanazawa1989 reviewed Apr 19, 2023

View reviewed changes

minor changes in macros.py

4029849

mtreinish reviewed Apr 19, 2023

View reviewed changes

to24toro added 6 commits April 19, 2023 22:59

add test to test_macros.py

381a8e6

make schedule_remapping_memory_slot private

b075cb3

delete since field from deprecate_arg

df4c4c7

delete deprecate depcorator

d4ff655

black macros.py

952f865

revert about target

f487cfe

nkanazawa1989 requested changes Apr 20, 2023

View reviewed changes

to24toro added 4 commits April 20, 2023 13:42

modify implementation of qubit_mem_slots

9fe1944

change the definition of meas_group_set

f35ed36

black macros.py

25f2ac9

fix meas_group_set

acb0c93

nkanazawa1989 reviewed Apr 20, 2023

View reviewed changes

to24toro added 2 commits April 20, 2023 14:43

fix qubit_mem_slots

9f973ea

reno

c2bed42

to24toro changed the title ~~Bug fix schedule with backendv2~~ Bug fix macros.measure with backendv2 Apr 20, 2023

to24toro added 2 commits April 20, 2023 17:35

modify unassigned_qubit_indices

f5fb919

remove list() from unassigned_qubit_indices and unassigned_reg_indices

9986e7f

nkanazawa1989 approved these changes Apr 20, 2023

View reviewed changes

nkanazawa1989 added this pull request to the merge queue Apr 20, 2023

Merged via the queue into Qiskit:main with commit 1203a3b Apr 20, 2023

chuanqixu mentioned this pull request Jun 16, 2023

Another Potential Solution to Schedule with V2 Fails #10298

Open

		backend (Union[Backend, BaseBackend]): A backend instance, which contains
		hardware-specific data required for scheduling.

	A schedule remapped by qubit_mem_slots as the input provided.
	A measurement schedule with new memory slot index.

	reg_index = qubit_mem_slots.get(qubit_index, 0)
	reg_index = qubit_mem_slots.get(qubit_index, qubit_index)

	unassigned_qubit_indices = list(meas_group_set - qubit_mem_slots.keys())
	unassigned_qubit_indices = list(set(meas_group) - qubit_mem_slots.keys())

Bug fix macros.measure with backendv2 #9987

Bug fix macros.measure with backendv2 #9987

Conversation

to24toro commented Apr 19, 2023 • edited Loading

Summary

Details and comments

qiskit-bot commented Apr 19, 2023

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

nkanazawa1989 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nkanazawa1989 Apr 19, 2023 • 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

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

to24toro Apr 19, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mtreinish Apr 19, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mtreinish left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mtreinish Apr 19, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nkanazawa1989 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nkanazawa1989 commented Apr 20, 2023

nkanazawa1989 left a comment

Choose a reason for hiding this comment

fvoichick commented Jul 9, 2023

to24toro commented Jul 10, 2023 • edited Loading

to24toro commented Apr 19, 2023 •

edited

Loading

nkanazawa1989 Apr 19, 2023 •

edited

Loading

to24toro Apr 19, 2023 •

edited

Loading

mtreinish Apr 19, 2023 •

edited

Loading

mtreinish Apr 19, 2023 •

edited

Loading

to24toro commented Jul 10, 2023 •

edited

Loading