this is the same issue as https://github.com/PostHog/feedback-app/issues/12

this is a follow up to https://github.com/PostHog/feedback-app/issues/6

using Hubspot I end up with the feedback window behind the chat button:

it would be helpful to get more control over the z-index or have the ability to position the form else where (f.e. next to the button that triggers the form).

Is your feature request related to a problem?

Right now it is only able to do insights on events (which can be related to persons/groups). However it would also be useful to be able to do an insight on persons/groups directly, f.e. to get the total number of accounts (potentially filtered by properties).

Describe the solution you'd like

Trend Insights get a new category "entity" where Persons/Groups can be selected.

Describe alternatives you've considered

Within the Persons & Groups it is possible to create graphs for persons/groups.

Additional context

https://posthogusers.slack.com/archives/C01GLBKHKQT/p1678722068741639

Thank you for your feature request – we love each and every one!

ie. setting * would just enable the plugin for evert event

I saw that there is a way to add/remove workers dynamically via https://docs.gunicorn.org/en/stable/faq.html#how-can-i-change-the-number-of-workers-dynamically

But I wonder if there is then a way to get the current worker count.

Also what is the best strategy to determine the $masterpid ?

Just use pgrep gunicorn and assume the lowest PID is the $masterpid ?

I guess for the main PID, there is the --pid file https://docs.gunicorn.org/en/stable/settings.html#pidfile

I have not test this out yet but I wonder if this PID file is automatically cleaned up due to the docs seemingly suggesting to clean up the file yourself: https://docs.gunicorn.org/en/stable/deploy.html#runit

At any rate, since my script will be triggered right after start of the server I will use pgrep gunicorn | head -n 1.

And to get the number of workers use:

$(pgrep gunicorn | wc -l | xargs)

though this I guess is always off by one due to the main worker.

Is your feature request related to a problem?

Using the popular PHP Laravel framework's HTTP client, array query parameters end up being encoded in a way that they end up being ignored by the PostHog API. I suspect it is because these parameters are expected to be encoded as JSON.

Describe the solution you'd like

PostHog API able to handle how Laravel encodes the query parameters.

Describe alternatives you've considered

PostHog SDK having an integrated HTTP client that properly encodes query parameters to how PostHog API expects them.

Additional context

This is for example how Laravel's HTTP client would encode the data:

date_from=2023-02-28&date_to=2023-02-28&entity_id=check&entity_type=events&entity_math=dau&offset=0&interval=day&events[0][id]=check&events[0][type]=events&events[0][order]=0&events[0][name]=check&events[0][math]=dau&properties[type]=AND&properties[values][0][key]=dashboard_id&properties[values][0][operator]=is_set&properties[values][0][type]=person&properties[values][0][value]=is_set

It seems like PostHog's API however expects some of these parameters to be encoded as JSON:

$query['events'] = '[{"id": "check", "type": "events", "order": 0, "name": "check", "custom_name": null, "math": "unique_group", "math_property": null, "math_group_type_index": 0, "properties": {"type": "AND", "values": [{"key": "$group_0", "operator": "is_not", "type": "event", "value": ""}]}}]';

Thank you for your feature request – we love each and every one!

only callable if the method is marked as static

I saw that there is a way to add/remove workers dynamically via https://docs.gunicorn.org/en/stable/faq.html#how-can-i-change-the-number-of-workers-dynamically

But I wonder if there is then a way to get the current worker count.

@tiina303 https://github.com/witty-works/posthog-property-filter-plugin/releases/tag/1.1.3

bump version

Merge branch 'release/1.1.3' into dev

Merge branch 'release/1.1.2' into dev

fix: don't add property values on missing top level filter part

Merge pull request #11 from tiina303/fix-property-filtering

fix: don't add property values on missing top level filter part

bump version

Merge branch 'release/1.1.3'

fix: don't add property values on missing top level filter part

Merge pull request #11 from tiina303/fix-property-filtering

fix: don't add property values on missing top level filter part

If we have a filter specified e.g. no-such.with-dot and that property doesn't exist on the event, then the code added no-such to properties with all the other properties from top level. See the updated test (which fails before code changes).

I opted to rewrite the function to be simpler and just remove the properties elements rather than assigning return values.

Internal details in this thread: https://posthog.slack.com/archives/C04BU5FS2Q7/p1677181224650189?thread_ts=1676581127.246339&cid=C04BU5FS2Q7

@tiina303 thank you!

@yakkomajuri merging and will release ASAP. so please update on US/EU cloud :)

Ideally the CPU should not spike to provide that the server is always responsive.

I have seen in the past that UseConcMarkSweepGC was at least implicitly recommended.

However, this GC was deprecated in favor of UseG1GC https://openjdk.org/jeps/291 https://openjdk.org/jeps/363

Would it be possible to add some information on https://dev.languagetool.org/http-server ?

Sentence Unsere Zubehörliste ist sehr lang with language de-CH results in a spelling mistake detected with Zubehör liste as an alternatives

curl -X POST --header 'Content-Type: application/x-www-form-urlencoded' --header 'Accept: application/json' -d 'text=Unsere%20Zubeh%C3%B6rliste%20ist%20sehr%20lang&language=de-CH&enabledOnly=false' 'https://api.languagetoolplus.com/v2/check'

"matches": [
    {
      "message": "Möglicher Tippfehler gefunden.",
      "shortMessage": "Rechtschreibfehler",
      "replacements": [
        {
          "value": "Zubehör liste"
        }
      ],
      "offset": 7,
      "length": 12,
      "context": {
        "text": "Unsere Zubehörliste ist sehr lang",
        "offset": 7,
        "length": 12
      },
      "sentence": "Unsere Zubehörliste ist sehr lang",
      "type": {
        "typeName": "UnknownWord"
      },
      "rule": {
        "id": "SWISS_GERMAN_SPELLER_RULE",
        "description": "Möglicher Rechtschreibfehler",
        "issueType": "misspelling",
        "category": {
          "id": "TYPOS",
          "name": "Mögliche Tippfehler"
        },
        "isPremium": false
      },
      "ignoreForIncompleteSentence": false,
      "contextForSureMatch": 0
    }
  ],

I ran into some issues with the german speller on the sentence “John Mearsheimer arbeitet in der Auslandberichterstattung.”

Note: “Mearsheimer” is in this case a correct name and “Auslandberichterstattung” contains a spelling mistake.

curl -X POST --header ‘Content-Type: application/x-www-form-urlencoded’ --header ‘Accept: application/json’ -d ‘text=John%20Mearsheimer%20arbeitet%20in%20der%20Auslandberichterstattung.&language=de-DE&enabledOnly=false’ ‘https://api.languagetoolplus.com/v2/check’

“Mearsheimer” has the following alternatives suggested “Merheimer” and “Marscheimer”. I think if the alternatices are also names the message should be more cautious and clarify that for this name there are different spellings.

For “Auslandberichterstattung” using de-DE the alternatives are “Auslandsberichterstattung” (correct), “Auslandsberichtserstattung” (imho incorrect in German), “Ausländerrichterstattung”/“Auslandsrichterstattung” (not even sure what these are supposed to mean).

Note with de-CH “Ausländerrichterstattung” is not included.

ok checking here https://forum.languagetool.org/t/community-languagetool-org-has-moved-to-a-new-server/3068/2?u=lsmith

thank you!

fixes https://github.com/laravel/framework/issues/45945#issuecomment-1419170465

@joelbutcher ok .. on what repo should this be documented? framework, jetstream or socialstream?