Plugin: spectral extraction #1514

kecnry · 2022-07-20T16:54:49Z

Description

This pull request implements a new plugin in specviz2d for spectral extraction via specreduce. It currently only implements the functionality for Trace creation, offsetting, loading from the API, and visualizing in the 2d spectrum viewer. The UI elements for the other "steps" (background, extraction) are implemented (in order to plan for the necessary interaction between the steps), but are currently non-functional. (woops! - that was a nice plan, but planning for how things would interact with each other eventually became just an API call away from actually implementing and devs decided it would be easier to review/merge in a complete form anyways)

IMPORTANT: this PR fixes the units on the x-axis to be shown in pixels instead of erroneously in meters, but by doing so disables the line list and line analysis plugins when in pixels (whenever not manually loading a 1D spectrum in other units). Future work will either need to properly handle plotting in wavelength or update the logic in these plugins to handle converting on-the-fly using the wavelength solution, if available.

view specviz2d barebones docs and the spectral-extraction docs (full specviz2d docs are out-of-scope, but created the plugins page so links from jdaviz work).

NOTE this PR currently requires the following non-merged branches from other dependencies (and will likely fail CI tests until they are merged and pinned here):

~~glue-astronomy#72~~ (functionality stripped from this PR)
~~specreduce#115~~ (functionality stripped from this PR)

Once the upstream PRs above are released, we can re-introduce the functionality in this plugin as a follow-up PR.

Sorry dev-reviewers for another large diff 😬 ! But many of the new lines are just boilerplate new plugin and boilerplate docs for specviz2d, so hopefully it won't be too difficult to parse 🤞

Note: this screen recording was still in pixel space. This has since been updated to show in wavelength, when possible, but the linking from the example notebook is so non-linear that it's almost unusable. We should probably consider switching it out with data that works better and then open follow-up efforts to improve support for more general input data.

Screen.Recording.2022-07-25.at.1.43.13.PM.mov

In addition to the UI, once glue-astronomy#72 is released, we can implement support for importing traces created in the notebook via app.add_data (this functionality has been stripped from the PR so that we're not blocked waiting for that PR to be approved, merged, and released):

from specreduce import tracing
trace = tracing.FlatTrace(spectrum1d.data, 29)

specviz2d.app.add_data(trace, data_label='trace_imported')
specviz2d.app.add_data_to_viewer('spectrum-2d-viewer', 'trace_imported')

You can also use the plugin API to interact with specreduce directly with something like:

ext_plugin = specviz2d.app.get_tray_item_from_name('spectral-extraction')
ext_plugin.trace_pixel = 29
trace = ext_plugin.create_trace()

Note that if the live-preview of the extracted spectrum disappears, that is because an error is being raised by specreduce. When using auto (Kosmos) trace or a manual background, specreduce is often throwing a (probably incorrect) error detecting an overlap in background regions. These cases also result in increased lag as they take a while to compute through specreduce. These cases should be fixed in specreduce and are out-of-scope here.

TODO:

If not removing background/extraction steps into a separate follow-up PR:

test coverage for background/extraction steps
docs for background/extraction steps
fix plotting of background/subtracted image

Potential Future Work:

~~advanced mode toggle (when in simple, hide intermediate export expansion panels as well as all inputs that default to "From Plugin")~~ (ended up partially simplifying the options without losing too much flexibility, a separate toggle at this point is probably overkill)
improved trace plot options support (currently shows with markers, but no line, but with line-like choices)
default color of trace to be non-gray?
can we edit existing entry in place instead of adding new data entry (then would be able to keep plot options)? This would likely need to be done in the AddResults component itself.

Things brought up in other issues/tickets that are not (yet) directly addressed:

Parameters can be specified manually in GUI and exported in notebook for reproducibility (like get_model_parameters).
Ability to set same parameters for all the spectra in the Mosviz table.
spectrum_result= spec2d.extract_spectrum(2dspectrum_background_subtracted_already, trace_object, extractor_object)

Post-merge TODO

Add GitHub issue templates for Specviz2d.
Update change log template in release instructions to include Specviz2d.

Closes #809

Checklist for package maintainer(s)

This checklist is meant to remind the package maintainer(s) who will review this pull request of some common things to look for. This list is not exhaustive.

Are two approvals required? Branch protection rule does not check for the second approval. If a second approval is not necessary, please apply the trivial label.
Do the proposed changes actually accomplish desired goals? Also manually run the affected example notebooks, if necessary.
Do the proposed changes follow the STScI Style Guides?
Are tests added/updated as required? If so, do they follow the STScI Style Guides?
Are docs added/updated as required? If so, do they follow the STScI Style Guides?
Did the CI pass? If not, are the failures related?
Is a change log needed? If yes, is it added to CHANGES.rst?
Is a milestone set?
After merge, any internal documentations need updating (e.g., JIRA, Innerspace)?

kecnry · 2022-07-20T18:27:26Z

jdaviz/components/plugin_dataset_select.vue

-          <v-icon style='margin-right: 2px'>{{ data.item.icon }}</v-icon>
+          <v-icon v-if="data.item.icon" style='margin-right: 2px'>{{ data.item.icon }}</v-icon>


this just makes the left of the entries with and without icons line up better. I also applied this across all other instances, even though not all dropdowns use "manual entries".

kecnry · 2022-07-20T18:28:48Z

jdaviz/components/viewer_data_select_item.vue

-    <j-tooltip :tooltipcontent="'data label: '+item.name" span_style="font-size: 12pt; padding-top: 6px; padding-left: 6px; width: calc(100% - 140px); white-space: nowrap; cursor: default;">
+    <j-tooltip :tooltipcontent="'data label: '+item.name" span_style="font-size: 12pt; padding-top: 6px; padding-left: 6px; width: calc(100% - 80px); white-space: nowrap; cursor: default;">


this fixes an alignment issue with the "x" button. #1491 apparently didn't properly set the alignment for all cases.

kecnry · 2022-07-20T18:30:03Z

jdaviz/configs/default/plugins/plot_options/plot_options.py

-                                               state_filter=is_profile)
+                                               state_filter=not_image)


Lines used to just be spectra, but now we have a line that is in an image viewer and needs to expose the right plot options.

kecnry · 2022-07-20T18:31:21Z

jdaviz/core/template_mixin.py

@@ -1018,8 +1021,27 @@ class DatasetSelect(BaseSelectPluginComponent):
    """
    def __init__(self, plugin, items, selected,
                 filters=['not_from_plugin_model_fitting', 'layer_in_viewers'],
+                 default_text=None, manual_options=[],


We didn't have a use for manual options in dataset dropdowns until now, so they weren't exposed. This works the same way as all other Select components that inherit from BaseSelectPluginComponent.

kecnry · 2022-07-20T18:32:23Z

jdaviz/core/template_mixin.py

-        replace = self.viewer.selected_reference != 'spectrum-viewer'
+        if replace is None:
+            replace = self.viewer.selected_reference != 'spectrum-viewer'


replace used to be hardcoded to True for viewers other than spectrum-viewer, but in this new plugin, we definitely want to plot the trace on top of the existing image, not replacing it.

kecnry · 2022-07-20T18:32:48Z

setup.cfg

@@ -34,6 +34,7 @@ install_requires =
    voila>=0.3.5
    pyyaml>=5.4.1
    specutils>=1.7.0
+    specreduce>=1.0.0


specreduce will be an added dependency (although we might need to pin a newer version).

Yet another required dependency... Is @havok2063 okay with this?

If not, we could have it be required for the plugin to show up, but then users would need to install additional dependencies depending on which plugins they want.

Tricky business: #1399

I think I'm ok with it, since we will use this plugin in MAST. Does this plugin also provide users information on the default spectral extraction that occurs when I load a single s2d file? Or is it only for performing custom extractions?

As far as I know, the (current) plan is only for custom extraction and using jdaviz as a visualization layer to existing and upcoming functionality in specreduce.

User feedback said they wanted to know how the default extraction was made. If we switch the default extraction to use this plugin to perform the extraction on initial load, would that provide that info and satisfy this feedback?

That's an interesting idea. I think the current displayed spectrum is just collapsed across the entire image. But if we could come up with automated defaults as input into the specreduce functionality, use those for the initial collapse instead, and have those set as the default in the plugin, that would be slick! I'm guessing even if that extraction isn't perfect, it would be better than just the full collapse?

There already is a snackbar warning that says: "The collapsed 1D spectrum is for quicklook purposes only. A robust extraction should be used for scientific use cases." Perhaps this text should also be updated to direct the user to this new plugin? Or maybe the "Spectrum 1D" data label should be changed to "Collapsed Spectrum"?

Yeah I think it would be fine to define some reasonable defaults to perform a default extraction. We would still display the snackbar warning message. Yeah it's also a good idea to update the message to point people to the plugin. If it's too much work to switch over, then I am also ok just updating the label to say "Collapsed Spectrum" and updating the message to point people to the plugin.

PatrickOgle · 2022-07-21T14:18:25Z

@kecnry This looks really good. My biggest suggestion would be to make the trace and extraction update automatically as you change the parameters, the products should then only be written out at the end (bottom of extraction workflow), similar to other plugins. Background subtraction might be more difficult to dynamically update, since it involves a 2D rather than 1D product, but that is just my guess.

kecnry · 2022-07-21T17:28:28Z

@PatrickOgle

My biggest suggestion would be to make the trace and extraction update automatically as you change the parameters

This is definitely do-able (maybe using the same blue we do in line analysis to make a clear distinction between the live preview and actual result), but I'm a bit worried about all the things we'll want to visualize as we implement the rest of the plugin: the trace, the adopted regions for the background (some width and separation around the trace), the background image and/or subtracted image, the region adopted for the extraction (some width around the trace), and finally the extracted spectrum. Showing all of these at once in an intuitive way might be not be possible with our limited real estate.

I think the regions for background and extraction might be the most important to show in real-time since they're just visualizing the inputs rather than the outputs and would really help the user to understand what the parameters mean. But which of the others do we want to show without making it too cluttered?

One option might be to have the visual feedback be temporary - so that when you change options for the trace, there is some live visual "preview" feedback but that only stays on the screen for 5 seconds, or maybe stays until you make changes to one of the other steps, and then goes away so you can see feedback from the current step. I'd be happy to give this a try if people like the idea, but it'll take some framework implementation first.

the products should then only be written out at the end (bottom of extraction workflow)

I see and agree with your point that for simple cases this would be more convenient. But unless we really don't want to support more complex workflows (few cases I have in mind listed below), I'd like to try to keep support for importing/exporting at each intermediate step. What about something like this to make the simpler case the default while still having the flexibility for interacting in the UI or notebook at each step? If we wanted to simplify it even more, we could have an "simple/advanced" toggle at the top which hides the export expandable sections and all the dropdowns that default to "From Plugin" which would effectively enforce a flow from top to bottom without all the clutter.

Screen.Recording.2022-07-21.at.1.07.12.PM.mov

Cases where the more advanced workflow might be useful:

The point you brought up about the background/subtracted image. Won't the user want to inspect these before finalizing extraction? I'm not sure we can visualize those images on the fly since they'll need to be loaded as data entries into the app and manually toggled in the data menu or have their opacities changed to view multiple layers at once. I suppose these images could still be added to the app when pressing the final "extract" button at the end and just inspected by the user to make sure they make sense, but then the user can't visually see the extraction region on top of the subtracted image ever (unless maybe by re-running the plugin, choosing the subtracted image from the previous run as input, and then choosing to skip doing another round of background subtraction).
Complex UI workflows: outputting the trace to a data entry enables the workflow of using one trace to create the background subtracted image and then refining the trace from that subtracted image (and similar complex, non-linear, workflows). Having everything automatically update linearly from the top to the bottom of the plugin without manual outputs, probably prevents this type of (advanced) workflow.
Notebook workflows: part of the original criteria was to support loading/exporting a trace object via the API. I think we could still support that with a linear/automatic flow in the plugin, but it would then be the only reason to need support for trace objects as data entries. If we went in this direction, I'd probably suggest loading the trace directly into the plugin itself, but that seems a bit at odds with @eteq's request for high-level API access to loading/visualizing traces.

kecnry · 2022-07-21T20:36:35Z

@PatrickOgle - this is what I've managed to come up with as a prototype from your suggestions (both for an easier linear-flow and for visual feedback) - let me know what you think! Unfortunately, I don't think this will be capable of live-preview of the background/subtracted images, but does visualize all of the various input parameters and does so in a way that isn't too cluttered (and hopefully is still intuitive - cc @Jenneh?).

Screen.Recording.2022-07-21.at.4.12.00.PM.mov

A slightly more complex workflow where you might want to use separate traces for the background and extraction would then look something like this

Screen.Recording.2022-07-21.at.4.20.33.PM.mov

PatrickOgle · 2022-07-21T20:37:18Z

Here is an example of static matplotlib visualizations that I find useful when performing an extraction, top to bottom:

Trace and extraction window on original 2D spectrum. Background locations (not shown) would be useful here too.
Trace and extraction window on background-subtracted 2D spectrum
Left: Spatial (cross-dispersion profile, with median trace location indicated (KosmosTrace dotted vs. PolyTrace (solid)).
3: Right: Trace y location vs x-pixel (for 4 different traces).
The extracted spectrum (1D

spectrum viewer).
Ignore--specific to my use case.

We will want at minimum 1 and 2, with 3 as nice-to have. 4. is already implemented.

kecnry · 2022-07-21T20:43:37Z

Ok, that's very useful as a reference, thanks!

(1) is implemented above in the comment I posted almost exactly as you posted.
(2) is possible, but probably will require the user to export the subtracted image into the viewer and select the layer manually (or maybe we can have some control in the plugin to do that in one step).
(3) would probably require a small plot in the plugin (similar to Imviz's line profiles). Or maybe we just need a more general plugin/viewer for plotting the cross-dispersion in general? I can see how this would be quite useful though and specviz2d being restricted to the two viewers will be limiting to the user.

kecnry · 2022-07-21T21:02:25Z

jdaviz/configs/specviz2d/plugins/spectral_extraction/spectral_extraction.py

+        # TODO: range(len(...)) is a temporary hack for this data displaying in meters instead of pixels
+        self.marks['trace'].update_xy(range(len(trace.trace)), trace.trace)


will definitely want to fix this or find a workaround so that the underlying bug doesn't become a blocker to this PR 🐱

kecnry · 2022-07-22T18:29:07Z

@havok2063 - I managed to implement your suggestion to use this as the default extraction method (if the user only loads a 2D spectrum). It currently defaults to a simple extraction with the trace set somewhat intelligently based on median of the row number of the brightest pixel in each column and with a width of 10% the full image height. When you open the plugin, you'll then see the settings that were used to extract that spectrum, and the snackbar is updated to state: "The collapsed 1D spectrum was generated automatically. Use the spectral extraction plugin for custom extraction."

Screen.Recording.2022-07-22.at.2.25.29.PM.mov

havok2063 · 2022-07-22T19:51:14Z

@kecnry That looks great! This plugin vastly improves the reasons for using Specviz2d. Can we change the snackbar message to the following? "The collapsed 1D spectrum was generated automatically. See the spectral extraction plugin for details or to perform a custom extraction."

One follow-up question, if someone performs a custom extraction in the GUI, is there a GUI way to save/download it as a file. Thinking about the use case of using this in MAST.

pllim · 2022-07-22T19:53:54Z

I thought Specviz2D was a temporarily solution until "something" happens? We don't even really document it in the user docs. Should we start?

pllim · 2022-07-22T19:55:57Z

docs/index.rst

@@ -174,6 +174,7 @@ Using Jdaviz
  imviz/index.rst
  specviz/index.rst
  cubeviz/index.rst
+  specviz2d/index.rst


Need to add specviz2d stuff to API docs too?

Also add new Specviz2D section to change log template in the release instructions?

On GitHub itself, should I create a new Specviz2D label and add corresponding Issue and Bug templates?

Since specviz2d is currently used by MAST (which links to a non-existent RTD page), I think we should... and we can always merge with mosviz later if we decide to go that route. Full docs are out-of-scope for this PR, I think/hope, but I just put the bare minimum so that all the plugin links point somewhere appropriate.

I added a specviz2d label in GitHub already (for this PR) and added a custom section in the change log (but not the template). The helper was already in the docs, but I'll add this plugin as well, thanks for catching that!

The book-keeping stuff (templates, release instructions) can be done after this is merged. Thanks!

I'll let @rosteen decide if anything in JIRA needs updating to include specviz2d...

* plotting the 1D spectrum in wavelength/frequency and linking to the 2D spectrum viewer can have adverse affects when the mapping is highly non-linear * previous behavior in main was to plot in pixel-space but had incorrect wavelength label/units on the x-axis (and in the accessed spectrum1d object). This PR will partly fix that by being consistent with units, but no longer attempts to plot in actual wavelengths until a more general solution is implemented.

* UserWarning from traitlets was being caused by setting the x/y values on the bqplot line, by casting to an array, this goes away * now that extraction replaces collapse, the NAXIS meta is no longer set. As far as I can tell, this is unused so can be safely removed from the test. If that isn't the case, we can manually add to the meta-data when extracting instead.

* for now setting to False, but could see the argument that they should default to True to mimic the behavior of the UI buttons which do add data to the app

* can be re-implemented once the glue-astronomy PR is merged and released

Co-authored-by: Ricky O'Steen <39831871+rosteen@users.noreply.github.com>

* could cause confusion with the actual spectral line in the 2d spectrum viewer

* the "accordion" container and the mouseover text includes the necessary context

kecnry · 2022-08-05T13:12:13Z

Rebased on top of #1538 which should fix the bug found by @rosteen and @javerbukh.

rosteen · 2022-08-09T15:39:25Z

I think this looks good to me now, thanks for addressing those comments.

pllim · 2022-08-09T16:24:25Z

I feel like @ojustino should give the second approval since he has been very involved in specreduce.

javerbukh

Looks good to me, thanks for addressing all the feedback! I'm OK with @ojustino at least taking a look at this monster of a PR before merging 😄

ojustino · 2022-08-09T18:25:52Z

Even though it uses specreduce, my understanding of the point of constructing the plugin this way was that this step's development focus would solely be on the jdaviz side. Given that, and a low likelihood that I'll be able to review it today (though I do plan to catch up soon), please don't delay a merge on my account.

kecnry · 2022-08-09T18:29:14Z

Thanks for the thorough reviews! @ojustino - I'll go ahead and merge then - I know the plan is to continue to iterate on this, so whenever you do get a chance to try it, feel free to recommend or make any changes.

kecnry added plugin Label for plugins common to multiple configurations specviz2d labels Jul 20, 2022

kecnry added this to the 2.8 milestone Jul 20, 2022

github-actions bot added the specviz label Jul 20, 2022

kecnry force-pushed the plugin-specreduce-trace branch from 3efcde2 to 87eeb5f Compare July 20, 2022 16:59

kecnry commented Jul 20, 2022

View reviewed changes

pllim requested a review from eteq July 20, 2022 18:47

kecnry force-pushed the plugin-specreduce-trace branch 4 times, most recently from 7693b7a to 10eee9e Compare July 20, 2022 19:43

rosteen modified the milestones: 2.8, 2.9 Jul 21, 2022

kecnry commented Jul 21, 2022

View reviewed changes

kecnry force-pushed the plugin-specreduce-trace branch 4 times, most recently from 8f420b4 to 98f89a8 Compare July 22, 2022 16:52

kecnry changed the title ~~Plugin: spectral extraction (trace)~~ Plugin: spectral extraction Jul 22, 2022

kecnry force-pushed the plugin-specreduce-trace branch 2 times, most recently from c9011d3 to 4e65f50 Compare July 22, 2022 18:48

pllim reviewed Jul 22, 2022

View reviewed changes

kecnry and others added 12 commits August 5, 2022 09:07

disable line lists and line analysis plugins when spectrum in pixels

4f6eb1d

consistent default for add_data in API methods

c5cdfd2

* for now setting to False, but could see the argument that they should default to True to mimic the behavior of the UI buttons which do add data to the app

API to display marks without having to manually open the plugin

10f9d63

update changelog

5cd693c

remove support for importing/exporting trace

ed1c396

* can be re-implemented once the glue-astronomy PR is merged and released

fixed trace live-feedback updating when changing type

0ac7c8b

Apply suggestions from code review

6550135

Co-authored-by: Ricky O'Steen <39831871+rosteen@users.noreply.github.com>

remove white "shadows" from spectral extraction live-previews

31ff49e

* could cause confusion with the actual spectral line in the 2d spectrum viewer

change buttons from Background/Subtract to Export

74766f9

* the "accordion" container and the mouseover text includes the necessary context

improve comment language

c42f1af

kecnry force-pushed the plugin-specreduce-trace branch from 876a9b2 to c42f1af Compare August 5, 2022 13:08

rosteen approved these changes Aug 9, 2022

View reviewed changes

kecnry added the Ready for final review label Aug 9, 2022

fix two-sided to actually be two-sided

ffe90ea

javerbukh approved these changes Aug 9, 2022

View reviewed changes

kecnry mentioned this pull request Aug 9, 2022

spectral-extraction: API #1554

Merged

10 tasks

kecnry merged commit f7ee912 into spacetelescope:main Aug 9, 2022

kecnry deleted the plugin-specreduce-trace branch August 9, 2022 18:31

This was referenced Aug 9, 2022

spectral-extraction: support for peak_method in auto/kosmos trace #1555

Merged

spectral-extraction: support import/export trace via glue-astronomy #1556

Merged

Specreduce integration with jdaviz astropy/specreduce#109

Closed

kecnry mentioned this pull request Oct 7, 2022

Enable loading specific extension for 2D spectrum in Specviz2d #1705

Merged

kecnry mentioned this pull request Apr 13, 2023

specviz2d: plot spectrum in wavelength/frequency with non-linear axis #2150

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Plugin: spectral extraction #1514

Plugin: spectral extraction #1514

kecnry commented Jul 20, 2022 •

edited

Loading

kecnry Jul 20, 2022

kecnry Jul 20, 2022

kecnry Jul 20, 2022

kecnry Jul 20, 2022

kecnry Jul 20, 2022

kecnry Jul 20, 2022

pllim Jul 20, 2022

kecnry Jul 20, 2022

pllim Jul 20, 2022

havok2063 Jul 20, 2022

kecnry Jul 20, 2022

havok2063 Jul 21, 2022

kecnry Jul 21, 2022

havok2063 Jul 21, 2022

PatrickOgle commented Jul 21, 2022

kecnry commented Jul 21, 2022

kecnry commented Jul 21, 2022

PatrickOgle commented Jul 21, 2022 •

edited

Loading

kecnry commented Jul 21, 2022

kecnry Jul 21, 2022

kecnry commented Jul 22, 2022

havok2063 commented Jul 22, 2022

pllim commented Jul 22, 2022

pllim Jul 22, 2022

kecnry Jul 22, 2022

pllim Jul 22, 2022

pllim Jul 22, 2022

kecnry commented Aug 5, 2022

rosteen commented Aug 9, 2022

pllim commented Aug 9, 2022

javerbukh left a comment

ojustino commented Aug 9, 2022

kecnry commented Aug 9, 2022

		<v-icon style='margin-right: 2px'>{{ data.item.icon }}</v-icon>
		<v-icon v-if="data.item.icon" style='margin-right: 2px'>{{ data.item.icon }}</v-icon>

		<j-tooltip :tooltipcontent="'data label: '+item.name" span_style="font-size: 12pt; padding-top: 6px; padding-left: 6px; width: calc(100% - 140px); white-space: nowrap; cursor: default;">
		<j-tooltip :tooltipcontent="'data label: '+item.name" span_style="font-size: 12pt; padding-top: 6px; padding-left: 6px; width: calc(100% - 80px); white-space: nowrap; cursor: default;">

		# TODO: range(len(...)) is a temporary hack for this data displaying in meters instead of pixels
		self.marks['trace'].update_xy(range(len(trace.trace)), trace.trace)

Plugin: spectral extraction #1514

Plugin: spectral extraction #1514

Conversation

kecnry commented Jul 20, 2022 • edited Loading

Description

Post-merge TODO

Checklist for package maintainer(s)

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

PatrickOgle commented Jul 21, 2022

kecnry commented Jul 21, 2022

kecnry commented Jul 21, 2022

PatrickOgle commented Jul 21, 2022 • edited Loading

kecnry commented Jul 21, 2022

Choose a reason for hiding this comment

kecnry commented Jul 22, 2022

havok2063 commented Jul 22, 2022

pllim commented Jul 22, 2022

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

kecnry commented Aug 5, 2022

rosteen commented Aug 9, 2022

pllim commented Aug 9, 2022

javerbukh left a comment

Choose a reason for hiding this comment

ojustino commented Aug 9, 2022

kecnry commented Aug 9, 2022

kecnry commented Jul 20, 2022 •

edited

Loading

PatrickOgle commented Jul 21, 2022 •

edited

Loading