Observable-Community

Hi <@U048G5BG31S> thanks for reaching out, yes please email support@<http://observablehq.com|observablehq.com> and someone from the team will get back to you. 

You should be able to set them in the call to <https://developer.mozilla.org/en-US/docs/Web/API/Fetch_API|fetch>. Do you have a notebook you can share?

<@U0487FYPTBQ> You can't override either header in fetch(). You'd have to use a proxy server instead that sets them for you.

Thanks <@U048E2BJL9Z>. Then I don't use anymore time for this.

<@U0487FYPTBQ> Is your data source public?

I would expect that you'd only need a CORS proxy, and for that you can pass your request through <http://corsproxy.io|corsproxy.io>.

<@U0487FYPTBQ> Like Robert mentioned, if you share a notebook link we can make more specific suggestions. :slightly_smiling_face:

I tried to use cors-proxy and It worked with other fetches but the problem is this one post which requires certain Origin and Referer. I'm using API which is "public" but without any documentation. And the problem is logging in. The process uses Wicket-Ajax scripts.

Ah, too bad. For what it's worth, I've been using this glitch instance as a proxy that overrides the origin: <https://glitch.com/edit/#!/observablehq-api> (it's being used <https://observablehq.com/@mootari/sortable-notebook-list|here>)

Maybe you can adapt it to your requirements?

In Python the post requires the following headers:
```    extraHeaders = {  
         'Wicket-Ajax': 'true', 
         'Wicket-Ajax-BaseURL': 'login', 
         'Wicket-FocusedElementId': 'loginWithUserID5', 
         'X-Requested-With': 'XMLHttpRequest', 
         'Origin': '<https://authentication2.caruna.fi|https://authentication2.caruna.fi>', 
         'Referer': '<https://authentication2.caruna.fi/portal/login|https://authentication2.caruna.fi/portal/login>' 
     } 
      
     # Post a form 
     r = s.post("<https://authentication2.caruna.fi/portal|https://authentication2.caruna.fi/portal>"+action, data=svars, headers=extraHeaders)```

Did you retrieve those from the browser? Chrome's dev tools have a neat feature that lets you copy any request as a fetch() code snippet (not that it solves your problem)

I should check that. I have used Chrome's dev tools to figure out everything what is needed.

does the post request redirect you? does it set cookies which you need to pass along?

Yes. There is so many redirects and cookies which have to be used.

In APL I did it by using TCPClient so It was quite nice exercise. :slightly_smiling_face: I learned a lot about the cookies...

You've probably figured it all out already, but if you have use for a somewhat related Node.js example, this is what I wrote to handle some server-side auth via GitHub: <https://github.com/mootari/observable-client/blob/master/src/auth/github.ts>

The problem is just that one post. And cors of course. But cors is possible to handle with proxies but I just have to figure out this post. I will check the dev tools next.

Yesterday <https://observable-cors.glitch.me/> was working but not anymore. <@U048E2BJL9Z> is there some kind of limit for this?

Yes, and that particular instance is getting hammered all the time. You can easily fork it though and create your own free instance.

But there's really no good reason to use that instance anyway, afaik. Use <http://corsproxy.io|corsproxy.io> instead.

Thanks <@U048E2BJL9Z>. Now I have the fetch which is working in Chrome:
```{
  "headers": {
    "accept": "application/xml, text/xml, */*; q=0.01",
    "accept-language": "en-FI,en-US;q=0.9,en;q=0.8",
    "content-type": "application/x-www-form-urlencoded; charset=UTF-8",
    "sec-ch-ua": "\"Google Chrome\";v=\"107\", \"Chromium\";v=\"107\", \"Not=A?Brand\";v=\"24\"",
    "sec-ch-ua-mobile": "?0",
    "sec-ch-ua-platform": "\"Windows\"",
    "sec-fetch-dest": "empty",
    "sec-fetch-mode": "cors",
    "sec-fetch-site": "same-origin",
    "wicket-ajax": "true",
    "wicket-ajax-baseurl": "login",
    "wicket-focusedelementid": "loginWithUserID5",
    "x-requested-with": "XMLHttpRequest"
  },
  "referrer": "<https://authentication2.caruna.fi/portal/login>",
  "referrerPolicy": "strict-origin-when-cross-origin",
  "body": "usernameLogin4_hf_0=&amp;ttqusername=username&amp;userPassword=password&amp;userIDPanel%3AloginWithUserID=1",
  "method": "POST",
  "mode": "cors",
  "credentials": "include"
}```

you can remove most of the headers (everything with `sec-`, as well as accept*, content-type and x-requested-with probably as well

That's true. The harder part is the lower part.

yeah, I'm not sure it's worth pursuing a client-side implementation, tbh

That's true. I can fetch everything with Python script and convert to parquet-file with DuckDB and then just transfer everything to AWS S3 bucket and then use it directly in ObservableHQ. My problem is that the programming is my hobby so I can be a really stubborn. :grin:

Oh I know *exactly* what that feels like :grimacing:

The number of notebooks that I created just "because I can" ...

Actually I don't need Origin. Just referrer.

There’s also <https://observablehq.com/@tomlarkworthy/fetchp|this fetch proxy> to get around CORS issues. Not sure if it helps with the other issues, though.

Add this option to your table configuration:
```value: this?.value```

<@U0481G8N085> Are you looking for something like <https://observablehq.com/@saneef/clearable-store#cell-128> ?

<@U049ZHN9ELD> that’s exactly what I want! this is magic thanks.

<@U048E2BJL9Z> that was almost what I want, and very neat, thanks for sending it.  the issue I had was if I selected something, and then the search didn’t match, the values got cleared.  As in, let’s say I search “bee”, then select “Beethoven”, then clear search, it stays selected (which is great), however, if I then search “alex”, Beethoven isn’t in the list anymore, so it gets deselected.

<@U0481G8N085> If you want to retain the selection *and* be able to deselect, you can use the following pattern:
```mutable choices = []```
```viewof table = {
  const options = new Set(search);
  const value = this ? new Set([...mutable choices, ...this.value]) : [];
  mutable choices = [...value].filter(d =&gt; !options.has(d));
  return Inputs.table(options, {sort: "searchMatch", reverse: true, value, required: false});
}```
You can then get the entire selection via
```[...table, ...choices]```

That's roughly the pattern that I used to retain the selection in <https://observablehq.com/@mootari/page-metadata#props>

`choices` basically retains any values which have been previously selected _and_ aren't present in the current set (and thus cannot be deselected given the current options)

What’s making debugging even more difficult is that I’m also not really seeing any errors. Threebox is using the THREE.js GlTF loader under the hood, so I tried using the same technique as in this notebook <https://observablehq.com/@vicapow/example-of-loading-and-modifying-a-glb-file>

Please let me know if you have any idea on how to get it to work!

image.png

Ah sorry, my bad! There’s a 3d model supposed to show

Here’s the link to the official example <https://github.com/jscastro76/threebox/blob/master/examples/12-add3dmodel.html>

I'm getting `Error: The layer 'undefined' does not exist in the map's style and cannot be styled.`

I'd start by checking which of your callbacks are actually called

The example works locally, so I’m still guessing that it has something to do with how the model is loaded

the local example requires a relative URL

map.on("style.load", ...) doesn't get invoked for me

Alright, it seems that the way I was constructing the URL was causing the “load” event not to fire

But now I’m running into a CORS error, which I don’t really get since the model is hosted as a file attachment

as a general debugging help, you can inspect the failing request in the network tab and then check the "Initiator" tab. that lets you jump into the code and set breakpoints

hmm but the model is indeed a gltf model, not an mtl model

you have two places where you load models. I'd recommend to comment most of the code and work your way up from there

I was rather under the impression that the “No material file found” errors is produced because the model is not being loaded correctly

like, start with the map, define your layer callback in a separate cell, slowly start to add logic until things break again

&gt; you have two places where you load models.
actually, that seems to have been a bug in the multiplayer editor - i saw your code nested in itself :sweat_smile:

tbh, the threebox code looks superweird. looks like it's just trying to load materials no matter what, making the `mtl` option basically required?

btw, your mapbox-gl version seems pretty old? 2.2.0 vs 2.10.0

when you say it's working locally, then your bundler is definitely the problem here

might be relevant: <https://github.com/jscastro76/threebox/issues/265>

Hmm but issues seems to be confirming the CORS topic: The person that created the issue tried to run the examples without locally hosting a server, and subsequently ran into CORS issues. Since they got the same errors that I’m getting now in the notebook it does indeed seem to be linked

I’m mainly not getting this part: Why is observable not allowing “in-house” requests? I’ll try if I can fix this by hosting it myself

this is really just a symptom, i believe. you may have noticed the warning about a missing materials file.

which you need to specify via the .mtl option

Again: The exact same code works locally with THREE.js requesting the resource itself, no need to specify the .mtl

The only difference is that in the local version of the code, the model URL is a relative file path

added a config.js including the mapbox access token (you can use mine :blush:)

<http://localhost:3000/examples/12-add3dmodel>

You're including a different build in your notebook though

Can you check what resources are loaded when you run locally?

There should be something right before the glb file

The materials *are* missing, the only difference is how threebox goes about their absence

You need to set a URL *path* that's accessible, e.g.
```mtl: "<https://cdn.jsdelivr.net/npm/threebox-plugin@2.2.7/examples/models/radar/>",```

that's basically overriding the base path and lets the MTLLoader carry on

<@U049QMP0CTH> I've sent you a suggestion: <https://observablehq.com/compare/30464bced172eaf0...b4ba4d31828a6e1e>

the window.tb assignment is also important. Overall this plugin feels ... icky :sweat_smile:

Does the model actually look like this when you run locally, or is it just black? <https://observable-community.slack.com/archives/C03SEMPUF6F/p1667992043971979?thread_ts=1667991336.828949&amp;cid=C03SEMPUF6F>

that’s exactly what I wanted, now it also works with the FileAttachments

thanks so much for helping me :slightly_smiling_face:

Materials are there, the time of day was wrong (the location is in Australia)

I’ll try to integrate the time slider as well

but the material/model party was the tough part!

in summary, *what* you provide to the `mtl` option seems to be irrelevant - it only has to return a 200 status code

So it's probably fine to just set the obj url there again

that is soo weird, that entire block of code seems to lack an if statement actually checking if the model type is `mtl` …

yep, that's what I've been trying to convey throughout this entire conversation :grimacing:

ok I’m sorry :disappointed: but again thanks for finding the fix!

Hey, all good! Glad it's working for you now! :slightly_smiling_face:

it’s just also weird that it’s not required if you run it locally…

You may still want to keep your eyes peeled for alternatives

No, it's not. When you run a notebook in Observable, it defines the parent URL as document.baseURI. OTOH when you ran your local server, it probably just returned an index file for /

yes! My goal was to create a free Observable based wind turbine and solar plant 3d simulator

ahhh that makes sense now! thanks for the explanation

The behavior is the same, it's just that in Observable that turned into a forbidden CORS request

alright now the notebook also includes the time slider!

can you share your notebook? My guess would be to use a custom function for the tickFormat, such as `tickFormat: d =&gt; d%5===0 ? `${d}` : ""`

<https://observablehq.com/d/a92e76eab82e1fd9>
I hope this works. This is not my main notebook, but I reproduced the problem I have here. Thanks for the suggestion! I am not sure where would that tickFormat go exactly with the xScale here

in your case the value for x is a string such as "04:00:00"; so a suitable tickFormat might be to take the two first chars of that string: `d =&gt; d.slice(0,2)` ; you can do so here:
```  const xAxis = d3.axisBottom(xScale).tickSizeOuter(0).tickFormat(d =&gt; d.slice(0,2));```

That already looks much better! And with your previous tip, I managed to get only a few labels showing. Thank you, you really made my day :slightly_smiling_face:

Hey <@U04AK15MFL3>! Welcome to working with D3 and Observable! D3 is an amazing tool, but it can be quite challenging to get started — I took a few minutes to create an <https://observablehq.com/d/af78d5400afde1c8|alternative visualization> using Plot, an open-source visualization library built on top of D3 (by the same people!).

Let me know if you have any questions, and happy plotting!

No, but… you could create two scales outside of the plot, and use them to have both marks return scaled colors

Thanks, I've shared the notebook with you :pray:

in this case I'd recommend `fill:"type, fillOpacity: "count"`

How would I go about changing the domain in this case?

you can tweak all the parameters in the `opacity` scale

Alternatively, how can I enforce a minimum distance between ticks?

if the (top-level) height that is computed automatically doesn't work for you, you need to pass your own

But that would have to be a total height, correct? Meaning, that I would have to determine upfront how many rows (for example) I'll have?

you can do it with Plot if you want: `new Set(Plot.valueof(data, "type")).size`

I assume the total height gets calculated here? <https://github.com/observablehq/plot/blob/0df32cd992322619bab5344a60afbeb1ab655dd4/src/dimensions.js>

yes… there's this notebook that I mean to publish … one day: <https://observablehq.com/@observablehq/plot-default-height>

it's not published because <https://github.com/observablehq/plot/pull/837> is still pending

heh, yeah I just saw and commented :grin:

Precalculating the height would mean that I'd have to account for tick labels as well, right?

If I were to use a rect mark, could I enfore a certain aspect ratio?

feels a bit weird to have rx and ry for bar / rect / cell, but then the aspect isn't maintained

What's missing for this PR to get merged?

Screenshot 2022-11-16 at 11.53.09.png

according to <https://gis.stackexchange.com/a/160293>

Still getting the same maritime border. In <https://observablehq.com/@saneef/geojson-from-openstreetmap?location=Sri%20Lanka&amp;query=border_type=nation&amp;selectGeometryTypes=MultiPolygon-Polygon|this notebook> you can see by clicking ‘fetch GeoJSON’ button

<https://osm-boundaries.com/Documentation|Looks like>, we need to do the post processing ourselves to get the land only border. :disappointed:

<@U049ZHN9ELD> apparently you can filter out this tag: <https://wiki.openstreetmap.org/wiki/Tag:boundary%3Dmaritime>

Hm. According to <https://wiki.openstreetmap.org/wiki/Boundaries> "administrative" is supposed to return coastal borders ...

<@U049ZHN9ELD> You may want to inspect the OSM data in that area to figure out what you need to query

For example <https://tools.geofabrik.de/osmi/?view=coastline&amp;lon=80.02933&amp;lat=9.17282&amp;zoom=11&amp;baselayer=Geofabrik%20Standard&amp;overlays=coastline%2Ccoastline_error_lines%2Cline_not_a_ring%2Cline_overlap%2Cline_invalid%2Cline_direction%2Cquestionable%2Ccoastline_error_points%2Cunconnected%2Cintersections%2Cnot_a_ring%2Cdouble_node%2Ctagged_node%2Ccoastline%2Ccoastline_error_lines%2Cline_not_a_ring%2Cline_overlap%2Cline_invalid%2Cline_direction%2Cquestionable%2Ccoastline_error_points%2Cunconnected%2Cintersections%2Cnot_a_ring%2Cdouble_node%2Ctagged_node|https://tools.geofabrik.de/osmi/?view=coastline&amp;lon=80.02933&amp;lat=9.17282&amp;zoom=11&amp;basel[…]ed%2Cintersections%2Cnot_a_ring%2Cdouble_node%2Ctagged_node>

<@U048E2BJL9Z> Thanks a lot! ` natural=coastline`, I can get features: <https://observablehq.com/@saneef/geojson-from-openstreetmap?location=Sri+Lanka&amp;query=+natural%3Dcoastline&amp;selectGeometryTypes=LineString-Polygon>

I’ll try to intersect with the proper administrative borders.

<@U049ZHN9ELD> Check the relation here: <https://www.openstreetmap.org/way/715274216>

<https://www.openstreetmap.org/relation/10946441#map=8/7.882/80.769>

which administrative borders are missing for you?

I need `boundary=administrative and admin_level=5` boundaries. <https://observablehq.com/@saneef/geojson-from-openstreetmap?location=Sri+Lanka&amp;query=boundary%3Dadministrative+and+admin_level%3D5&amp;selectGeometryTypes=Polygon|See>. But those include maritime boundaries.

Yeah, looks like clipping is the way to go. Found <https://help.openstreetmap.org/questions/44591/coastline-relations> which has a similar problem, with rather unhelpful answers (do those people really not get the _intent_ of the question?)

I went with clipping GeoJSONs, <https://observablehq.com/@saneef/clip-geojson-files>

Thanks for this question <@U04BN5UP0GP>. We don’t yet have an Observable data connector for Databricks. This is on our roadmap and it is good to know that you are interested in using it. The mention of Databricks on that <https://observablehq.com/resources/use-cases|intro webpage> was definitely an error that we have now corrected. Thanks for letting us know.

If a node module connector is available, that should work. I’ll check with the team to see if we can figure out why @databricks/sql is not working.

Thanks <@U048YB4AXU0>! Looking forward to you all having an official connector, and thanks for looking into the node connector - if I can be of any help to the folks checking, I'm at your service.

If it's useful, here's a screenshot of what I tried (`connectionParams` contains the secrets):

<@U04BN5UP0GP> - While I hope that this Slack group gets more traction, you may also consider dropping your question into <http://talk.observablehq.com|talk.observablehq.com>, as I find questions there are read by a wider audience.

<@U048E2BJL9Z> pointed out that this is a node.js module that is not meant to run in the browser :disappointed: <https://docs.databricks.com/dev-tools/nodejs-sql-driver.html>

Thanks <@U048YB4AXU0> for chasing this down - I'll wait for good news about an official connector :slightly_smiling_face:

The cell has a .done property that you can check

```all_data = raw.done ? raw : invalidation```

easiest is probably to rename raw to something like raw_loading, and then create a new "raw" cell with the above pattern

it's just returning an unresolved promise

I'm using new Promise() in my debounce code, is it just a quicker way of doing same? Or is there …

like returning new Promise(), but making sure that the promise will resolve at some point. Courtesy of Mike Bostock. :slightly_smiling_face:

I’m using like this `raw = data_stream.done ? data_stream : (invalidation, [])` , should be okay right?

One trick I have is that if we're not .done, I sometimes (randomly) return the interim dataset, so that I have intermediate states during loading. It's nice but if the states are slow to compute, we need to adjust the random factor…

that would defeat the purpose saneef, it will make all your dependent cells rerun on an empty dataset

<@U049ZHN9ELD> no. (invalidation, []) will always return []

`raw = data_stream.done || Math.random() &lt;0.01 ? data_stream : invalidation`

would be perfect if you could consume the source page-wise

... I guess you can, if you run your query outside of an sql cell?

yes it works: interim values are yielded every 16ms until the sql query completes. But if my charts downstream take too much time to compute, this time is added to the interval, and the sql cell slows down. Which is why I'm trying to debounce… maybe we should pair in a notebook :smile:

Screenshot 2022-11-23 at 23.45.45.png

Screenshot 2022-11-23 at 23.45.28.png

Would you mind filing a brief issue in the feedback repo and adding your screenshots there? Thank you! :pray:

<https://github.com/observablehq/feedback/issues/507>

<@U049ZHN9ELD> Thanks again, we're looking into it!

Does <https://observablehq.com/d/7ce9514442ff0d8c#found|this method> works for you?

Thanks, that's a very functional way of doing it.

<https://bit.ly/WhatB-Movies>

No idea if this would help, but there is a notebook with a non-grid projection where the labels are correctly placed: <https://observablehq.com/@d3/graticule-labels-stereographic?collection=@d3/d3-geo>

... probably doesn't get you around the issue of where your intersection is off canvas

That is similar to the one in <https://observablehq.com/@d3/graticule-labels>, but on circular path. In both it’s drawing labels on well defined range, independent of graticules. Because the lines are predictable. In my case, we don’t before hand where the graticules as going cross. Or, at least I couldn’t find a way :disappointed:

I would recommend to add a clear example of what you expect the output to look like (ideally also as a screenshot).

The example that you referenced uses a zoom that shows the entire map with straight graticules, and also has the labels on the outside.

maybe this old notebook <https://observablehq.com/@fil/finding-intersections-method-2>

Oh my! That’s what I want. Now trying to decode the code :smile:

I don't remember how it works, but lmk if you have questions and I'll try to see if I can read my old writing :wink:

Thanks again, <@U048X1JDDHN>. It’s took me a while to understand how it works. The moment I understood, wow! It is an elegant solution! Just maths!

Also, I got it work <https://observablehq.com/d/9130687d8dbb603c|on a rectangular map canvas>.

You may want to review your CORS configuration. There are no CORS headers on the responses.

It seems that CORS support for R2 storage was only added as recently as late September, and may people struggle to enable it. Maybe this article can help? <https://kian.org.uk/configuring-cors-on-cloudflare-r2/>

I have used that blog to get the correct CORS settings and I have asked from their Discord what is the best way the check that CORS is working properly. My settings for CORS is quite straight forward:
```&lt;?xml version="1.0" encoding="UTF-8"?&gt;
&lt;CORSConfiguration xmlns="<http://s3.amazonaws.com/doc/2006-03-01/>"&gt;
    &lt;CORSRule&gt;
        &lt;AllowedMethod&gt;GET&lt;/AllowedMethod&gt;
        &lt;AllowedMethod&gt;HEAD&lt;/AllowedMethod&gt;
        &lt;AllowedMethod&gt;PUT&lt;/AllowedMethod&gt;
        &lt;AllowedOrigin&gt;*&lt;/AllowedOrigin&gt;
        &lt;ExposeHeader&gt;ETag&lt;/ExposeHeader&gt;
        &lt;MaxAgeSecond&gt;3000&lt;/MaxAgeSecond&gt;
        &lt;AllowedHeader&gt;content-type&lt;/AllowedHeader&gt;
    &lt;/CORSRule&gt;
&lt;/CORSConfiguration&gt;```

You can inspect the response headers via the Network tab in your browser's dev tools:

That's true but that doesn't help much. ;)

CORS problems would still let you access the file directly, which is not the case: <https://linna.78fa4b4b59d847246c0bfd2d91be662f.r2.cloudflarestorage.com/electricity.parquet?x-id=GetObject>

```&lt;Error&gt;
&lt;Code&gt;InvalidArgument&lt;/Code&gt;
&lt;Message&gt;Invalid Argument: Authorization&lt;/Message&gt;
&lt;/Error&gt;```

Or is the required authorization intentional?

... Not sure if it helps, but I get this occasionally when I move notebooks between Teams.  Did you happen to recently transfer from your personal account to a Teams account?

... ah, I see you've specified credentials

<@U0489M6NVGU> These aren't file attachments

<@U0487FYPTBQ> Are you able to access the file with a server-side client (e.g. via Python)?

Yes, I am. With `aws cli` and `boto3`which are basically the same tool and also with `postman`.

Seems to be an open issue with Cloudflare: <https://community.cloudflare.com/t/problem-with-settings-cors-policies-on-r2/432339/6>

Thanks <@U048E2BJL9Z>. I will ask there what is the situation.

Perhaps mentioning siddhant1 directly will help to get their attention.

Sid already responded on Discord channel. Some updates are in the pipeline.

Now it's working nicely. I put examples to <#C0483RB3QJX|bring-observable-to-work>

I am guessing it would look something like this, but I haven't gotten it yet: <https://observablehq.com/d/7050cb713a965597> ...or maybe it is working but it's only taking width at the time the notebook loads...

Can you share (or DM me) the notebook URL?

Hi Eli, sorry this is happening. To remedy the read-only issue we can get your students set up with individual edu licenses. I'll follow up via email.

Hi Max, the voronoi mark does not return coordinates, it just creates svg:path elements. I think this is a case where we have to write a custom transform <https://observablehq.com/@fil/plot-voronoi-labels>

ah perfect thanks <@U048X1JDDHN> - I'm finding `voronoiCentroids` a really good example of a custom initializer too, learned a lot from it

Thanks! If you play with it, you can modify it so that for example it doesn't move the labels too far (by setting a max radius or something). It would be nice too if it could decide on the label's textAnchor rather than just move the reference point. This is such a common task, and we always end up doing it by hand in Illustrator… it's frustrating.

Can you give an example of what you have in mind?

```Plot.groupX({
    y: "count",
    x: (d) =&gt; Math.ceil(d / 10) * 10
})```

Seems to work, although I’m still getting horizontal lines, it seems
<https://observablehq.com/d/53f900161ea1600c>

(graph below map)

<@U04DB6A0CEQ> this seems to work?
```Plot.plot({
  y: {
    grid: true
  },
  marks: [
    Plot.rectY(
      speeds,
      Plot.binX({
        y: "count",
        interval: 10,
      },{
        fill: (d) =&gt; getColor(Math.ceil(d / 10) * 10)
      })
    )
  ]
})```

Oh wow. I didn’t capture the interval part

All the <https://observablehq.com/@observablehq/plot-group?collection=@observablehq/plot|examples> are _categorical_, so it is hard to see what is possible with numerical axes…

Note that you can also apply the colors this way:
```Plot.plot({
  color: {
      domain: [100, -10],
      interpolate: d3.interpolateSpectral,
  },
  y: {
    grid: true
  },
  marks: [
    Plot.rectY(
      speeds,
      Plot.binX(
        {
          y: "count",
          interval: 10,
          fill: "x"
        }
      )
    )
  ]
})```

1. yes, it will automatically redirect. And if you decide to change your notebook's slug in the future, the old slug will redirect to the new one.
2. No. The history becomes accessible from the moment that you make your notebook public. But if you fork your notebook, the fork's history will start at the fork version.

1. That's great news, thanks!
2. I thought that might be the case. Good to know for the future - I guess this time everyone gets to see the sausage get made if they choose :sweat_smile: 

I was hesitant at first, but now I've come to embrace it. Pulling back the curtain this way lets everyone see that even experienced developers are only human, and that everyone struggles at times. :slightly_smiling_face:

If I fork another notebook, can I hide that one or will the forked notebook always link back to the original with full history?

Oh, maybe I should operate with a trunk private notebook, and fork individual public notebooks to share with experts?

a trick is to fork twice and trash the intermediate notebook

there is no option (yet?) for cutting the cord