Embora você não possa evitar a mensagem do sistema, você pode gerenciar a sequência de feedback para o usuário no seu aplicativo:

1. Exibir uma mensagem clara imediatamente: Assim que você detectar o erro de "Certificado Revogado" (dentro do case .verified(let transaction) onde transaction.revocationDate != nil), você deve imediatamente atualizar a UI do seu aplicativo para informar o usuário sobre a falha e o motivo. Por exemplo, você pode exibir um alerta, uma mensagem de erro na tela ou desabilitar funcionalidades relacionadas à compra.

   Swift

   case .verified(let transaction):
       if transaction.productType == product.type, transaction.productID == productID {
           if transaction.revocationDate == nil, transaction.revocationReason == nil {
               purchaseState = true
               await transaction.finish()
           } else {
               // Certificado revogado ou transação revogada.
               // Aqui você exibe sua própria mensagem de erro para o usuário.
               self.error = .error(NSError(domain: "PurchaseError", code: 1, userInfo: [NSLocalizedDescriptionKey: "A compra não pôde ser concluída devido a um certificado revogado. Por favor, tente novamente mais tarde ou contate o suporte."]))
               // Não chame transaction.finish() para transações revogadas,
               // a menos que você tenha uma lógica específica para elas.
               // O StoreKit pode limpar essas transações automaticamente em alguns casos.
           }
       } else {
           // produto inválido
           self.error = .error(NSError(domain: "PurchaseError", code: 2, userInfo: [NSLocalizedDescriptionKey: "Produto da transação não corresponde."]))
       }
   
   

2. Gerenciar o estado purchaseState: Certifique-se de que purchaseState só seja definido como true se a compra for realmente válida e verificada. Em caso de certificado revogado, purchaseState deve permanecer false (ou ser redefinido) e você deve apresentar o erro.

Ao fazer isso, embora o usuário veja o "You're all set" momentaneamente, ele será rapidamente seguido pela mensagem de erro do seu aplicativo, o que comunica a falha de forma mais eficaz e reduz a confusão.

---

Em resumo, a arquitetura do StoreKit, especialmente com os arquivos de configuração, separa a confirmação inicial da compra da validação posterior. Concentre-se em fornecer um feedback claro ao usuário no seu aplicativo assim que você detectar a falha de verificação.

HOOMD-blue com GPU no WSL2: Analisando o Erro "invalid device ordinal"

O erro "invalid device ordinal" que você está enfrentando é peculiar, especialmente porque nvidia-smi e GPU.get_available_devices() detectam sua GPU corretamente.

Vamos abordar suas perguntas e possíveis soluções:

O HOOMD-blue é totalmente compatível com GPU no WSL2?

Sim, o HOOMD-blue (e o CUDA em geral) é projetado para ser compatível com GPU no WSL2. A NVIDIA tem feito esforços significativos para garantir que suas GPUs e o CUDA funcionem perfeitamente dentro do WSL2, incluindo o suporte para o subsistema de GPU virtual (vGPU) que o WSL2 utiliza. Muitos usuários executam com sucesso aplicações CUDA, incluindo outras bibliotecas de simulação e aprendizado de máquina, dentro do WSL2.

O fato de você conseguir ver a GPU com nvidia-smi e o HOOMD listar a GPU disponível indica que a camada de comunicação entre o WSL2, os drivers NVIDIA e o runtime CUDA está funcionando em um nível básico. O problema parece ser mais específico à forma como o HOOMD está tentando inicializar o dispositivo.

Como corrigir o erro "invalid device ordinal"?

O erro "invalid device ordinal" geralmente ocorre quando o programa tenta acessar uma GPU usando um índice que não corresponde a um dispositivo válido ou quando há um problema subjacente na inicialização do contexto CUDA para aquele dispositivo. Dado que você já tentou device_id=0 e CUDA_VISIBLE_DEVICES=0, e sua GPU é a única listada, o problema não é a seleção de um índice incorreto.

Aqui estão algumas abordagens e considerações para tentar resolver o problema:

1. Verifique a Versão do Driver NVIDIA:

   • Embora você mencione "drivers mais recentes", certifique-se de que são os drivers mais recentes para WSL2. A NVIDIA lança drivers específicos para o WSL2 que contêm otimizações e correções. Você pode baixá-los diretamente do site da NVIDIA (geralmente na seção de drivers para desenvolvedores ou drivers de notebook).

   • Pode ser útil tentar um driver ligeiramente mais antigo, mas ainda compatível com WSL2, para descartar uma regressão em uma versão muito recente.

2. Atualize o Kernel do WSL2:

   • Certifique-se de que o kernel do WSL2 está atualizado. Problemas de compatibilidade entre o kernel do Linux e os drivers da GPU podem causar erros estranhos.

   • Abra o PowerShell como administrador e execute:

     PowerShell

     wsl --update
     wsl --shutdown
     
     

   • Reinicie o Ubuntu no WSL2.

3. Verifique a Versão do CUDA Toolkit:

   • Você mencionou que testou o CUDA Toolkit 12.2 e 11.8. O HOOMD-blue 5.2.0 é compilado contra uma versão específica do CUDA. Embora o CUDA seja geralmente retrocompatível, pode haver pequenas incompatibilidades.

   • Sugestão: Verifique a documentação oficial do HOOMD-blue 5.2.0 ou as notas de versão para ver qual versão do CUDA Toolkit é a recomendada ou utilizada para a compilação oficial das builds do Conda/Micromamba. Tentar corresponder essa versão pode resolver o problema.

   • Às vezes, ter várias versões do CUDA Toolkit instaladas pode causar conflitos nas variáveis de ambiente. Certifique-se de que o PATH e outras variáveis estão apontando corretamente para a versão que você deseja usar.

4. Teste com um Exemplo CUDA Mais Simples:

   • Para isolar se o problema é do HOOMD ou do ambiente CUDA/WSL2, tente executar um exemplo CUDA simples (por exemplo, um programa C++ ou Python usando pycuda ou numba com CUDA) que apenas inicialize um dispositivo e faça uma operação trivial. Se esses exemplos funcionarem, o problema é mais provável que seja específico da interação do HOOMD. Se eles falharem, o problema é mais fundamental na sua configuração CUDA/WSL2.

5. Variáveis de Ambiente Adicionais do CUDA:

   • Embora CUDA_VISIBLE_DEVICES seja o mais comum, pode haver outras variáveis de ambiente do CUDA que afetam a inicialização do dispositivo.

   • Por exemplo, CUDA_LAUNCH_BLOCKING=1 pode ajudar a depurar, forçando as chamadas CUDA a serem síncronas, o que pode revelar o ponto exato da falha. Não é uma solução, mas pode ajudar na depuração.

6. Reinstalação Limpa do HOOMD (e dependências):

   • Às vezes, pacotes podem ser corrompidos. Tente criar um novo ambiente Micromamba/Conda do zero e instale apenas o HOOMD:

     Bash

     micromamba create -n hoomd_env python=3.10
     micromamba activate hoomd_env
     micromamba install hoomd=5.2.0
     
     

   • Isso garante que não há conflitos de pacotes ou arquivos residuais de instalações anteriores.

7. Verifique Logs de Erro do Sistema:

   • Procure por quaisquer mensagens de erro ou avisos relacionados a drivers de GPU ou CUDA nos logs do sistema do Ubuntu dentro do WSL2 (por exemplo, dmesg, /var/log/syslog).

Soluções Alternativas Conhecidas

Se as opções acima não funcionarem, e você precisar de aceleração de GPU, as principais alternativas (além de hoomd.device.CPU()) seriam:

• Linux Nativo: A maneira mais robusta de garantir compatibilidade e desempenho máximo para cargas de trabalho de GPU é executar o Linux nativamente (dual boot ou em uma máquina dedicada). Isso elimina a camada de virtualização do WSL2 e seus potenciais overheads ou incompatibilidades.

• Windows Nativo (se HOOMD oferecer suporte GPU nativo): Verifique se o HOOMD-blue oferece builds com suporte a GPU para Windows. Se sim, essa seria uma alternativa mais simples do que o Linux nativo, embora o desempenho possa não ser idêntico ao do Linux.

Informações de Usuários com Sucesso

Embora eu não tenha acesso a feedback de usuários em tempo real sobre o HOOMD-blue especificamente no WSL2, a comunidade CUDA em geral tem tido sucesso significativo. O erro "invalid device ordinal" é geralmente um sinal de que algo está errado na interação entre o runtime CUDA, os drivers e o ambiente (neste caso, o WSL2).

A chave para resolver problemas como este no WSL2 é garantir que:

1. Os drivers NVIDIA no Windows são os compatíveis com WSL2 e estão atualizados.

2. O kernel do WSL2 está atualizado.

3. A versão do CUDA Toolkit que você está usando no ambiente Micromamba/Conda é compatível com a versão do HOOMD-blue que você instalou.

Espero que estas sugestões o ajudem a resolver o problema! Se puder fornecer mais detalhes sobre a versão exata do seu driver NVIDIA para WSL2 e a output completa do erro, isso pode ajudar a refinar o diagnóstico.

In your sql query the input string type is NVARCHAR. This when encoding uses UTF-16.
In C# you used UTF8 encoding.
The byte representation differs in these 2 cases and hence you see diff SHA-256 hashes.

Both are valid. It depends on the encoding you prefer.
If you prefer to use UTF-8 in sql then use VARCHAR for input type.

I am not seeing any of this approaches work in multithreaded environments in PRODUCTION. it does not find exact match for class name or match for other flags. how to use all this code in multithreaded environment while different threads calls MessageBox window?

The redirect URI (where the response is returned to) must be registered in the APIs console, and the redirect_uri_mismatch error indicates that it's not registered or that it's been set up incorrectly. Ensure that the redirect URl you provide in your OAuth authorization request matches exactly the URl that you have added in the Google Developer Console. See related post that might be helpful to you.

Yes, you can set terminationGracePeriodSeconds to more than 15 seconds for non-system Pods on GKE Spot VMs, but it is honored only on a best-effort basis. When a Spot VM is preempted, it gets a 30 seconds termination notice. The kubelet prioritizes gracefully shutting down non-system Pods, usually allowing up to 15 seconds. If you set a longer period like 25 seconds kubelet may attempt to honor it, but system Pods also require time to shut down. If the total exceeds the 30 seconds window, your Pod may be forcefully killed.

Therefore, while values above 15 seconds are allowed, they may reduce the time available for critical system Pods, making 15 seconds or less the safest option for reliable shutdown behavior on Spot VMs. If your workloads require longer shutdown periods such as for draining connections or flushing logs, consider using standard VMs instead.

For best practice you can refer to this documentation for additional information.

Use Terminal to Launch the Executable

Instead of using open, you can directly invoke the executable file within the app bundle: bash

/Applications/MyApp.app/Contents/MacOS/MyApp

This ensures the application attaches to the terminal and can read user input. i hope*

It looks like Heroku just updated their Python classic buildpack. The error message you are seeing is addressed in the following change log:

https://devcenter.heroku.com/changelog-items/3324

Instagram’s API does not support programmatic login using a saved username and password due to strict security and privacy policies. OAuth 2.0 is the only official and supported method, which requires explicit user interaction for authentication. Attempting to bypass this flow by storing credentials and automating login violates Instagram's terms of service and may lead to API access being revoked.

Instead, you should authenticate the account once using the standard OAuth flow, obtain a long-lived access token, and securely store it (e.g., in a database). You can then use this token to programmatically fetch images without needing further logins. This method adheres to Instagram's API guidelines while allowing automated access to the account's media.

When creating an app, there is a new concept of "use cases". When asked to select a use case, select "Other" (at the bottom). It will then use the older layout and you are able to add Products.

Genius!

ps: if you work on a burger menu - change the "top" css attribute by -1 pixel -

This work for me:

Try to put the logic such as AuthOptions at the different file example

/src/auth.ts

export const { handlers, signIn, signOut, auth } = NextAuth({})

/src/app/api/auth/[...nextauth]/route.ts

import { handlers } from "@/auth"
export const { GET, POST } = handlers

You could try using JavaScript instead of VBScript:

<HTML>
  <HEAD>
    <TITLE>Fix</TITLE>
  </HEAD>
  <BODY>
    <FORM>
      <INPUT TYPE="Run" NAME="Button" VALUE="Click">

      <SCRIPT FOR="Button" EVENT="onClick" LANGUAGE="JavaScript">
        var WshShell = new ActiveXObject("WScript.Shell");
        WshShell.Run("cmd.exe '.\example.bat'");
      </SCRIPT>

    </FORM>
  </BODY>
</HTML>

const cors = Require [ 'cors '] app .usse [ cors ][]

This may not be the cleanest solution and I'll still leave this question as unsolved as it isn't fully understood in my opinion. Yet still I want to share my experience and how it worked out for me.

I modified it a little bit but I used the code from this post as minimal working example. I thought map_id has to be the same as activeMapType: map.supportedMapTypes[map_id] but this seems not to be the case. Using map_id = 1 and map.supportedMapTypes[0] resulted in the desired Street Map map type and using offline tiles.

I wrote an article on searching encrypted birthdays. As long as the data you are encrypting/searching isn't identifiable by its self, this method works well.

https://www.sqlservercentral.com/articles/searching-an-encrypted-column

Today Google will show the test it makes so you can confirm if Google is using the up to date validation as per Schema.ORG CSV or JSON file for allowed @type elements. Simply hover over the 'failed' test element and then compare to : https://github.com/schemaorg/schemaorg/blob/main/data/releases/14.0/schemaorg-all-https.jsonld

Line 19554 confirms.

Serilog–Sub loggers would be an adequate start. Almost your exact example.

To ensure the public/ directory is correctly exposed when deploying a PHP project on Vercel, you need to configure the routing so that all incoming requests are internally directed to that folder. This allows users to access the app without explicitly including public/ in the URL. Typically, this is done by setting up URL rewrites in the deployment configuration so that the public/ folder acts as the root for your application. Additionally, it’s important to place all your public-facing files inside that directory and avoid referencing the public/ folder in your paths or links directly. This setup helps maintain a clean and secure URL structure while aligning with Vercel's deployment model.

Thanks for help, I change a little bit inputs, reinstal python and lybraries to have path set automaticly and finnaly its giving me hand landmarks :)
First hand landmarks: [NormalizedLandmark(x=0.35780423879623413, y=0.005128130316734314, z=7.445843266395968e-07, visibility=0.0, presence=0.0),

What I finally have is:

import mediapipe as mp
from mediapipe.tasks import python
from mediapipe.tasks.python import vision
from mediapipe.tasks.python.core.base_options import BaseOptions

# Use an absolute path
task_path = Path(__file__).parent / "hand_landmarker.task"

print("Resolved model path:", task_path)

if not task_path.exists():
    raise FileNotFoundError(f"Model file not found at: {task_path}")

options = HandLandmarkerOptions(
    base_options=BaseOptions(model_asset_path=str(task_path.resolve())),
    running_mode=VisionRunningMode.IMAGE,
    num_hands=2
)

detector = vision.HandLandmarker.create_from_options(options)

# Hands
mp_image = mp.Image(image_format=mp.ImageFormat.SRGB, data=cv2.cvtColor(cv2_image, cv2.COLOR_BGR2RGB))
hands_results = detector.detect(mp_image)
results['hands'] = hands_results.hand_landmarks
hand_landmarks = results.get('hands')
first_hand = hand_landmarks[0] if hand_landmarks and len(hand_landmarks) > 0 else None
print("First hand landmarks:", first_hand)

You fail to call your mainLoop(). You initialize your app, but all that does is call the constructor. Add myApp.mainLoop() at the end.

You can do this using laravel eloquant. First create these modals

Country, PaymentMethod, CountryPaymentMethod, PaymentMethodConfiguration

And then create these relationships inside modals.

// Country.php
public function countryPaymentMethods()
{
    return $this->hasMany(CountryPaymentMethod::class);
}

// PaymentMethod.php
public function countryPaymentMethods()
{
    return $this->hasMany(CountryPaymentMethod::class);
}

// CountryPaymentMethod.php
public function configurations()
{
    return $this->hasMany(PaymentMethodConfiguration::class);
}

public function country()
{
    return $this->belongsTo(Country::class);
}

public function paymentMethod()
{
    return $this->belongsTo(PaymentMethod::class);
}

// PaymentMethodConfiguration.php
public function countryPaymentMethod()
{
    return $this->belongsTo(CountryPaymentMethod::class);
}

And you can query them like this

$country = Country::where('name', 'Ireland')->first();
$paymentMethod = PaymentMethod::where('name', 'Mobile Money')->first();

$configurations = PaymentMethodConfiguration::whereHas('countryPaymentMethod', function ($query) use ($country, $paymentMethod) {
    $query->where('country_id', $country->id)
          ->where('payment_method_id', $paymentMethod->id)
          ->where('is_active', true);
})->get();

Study about this more here

Laravel - Eloquent "Has", "With", "WhereHas" - What do they mean?

Thank you for the feedback. I am the author of grapa. It started out as more of a personal project where I could test out ideas, but evolved over time. There are a lot of capabilities in the language that I am finding with ChatGPT and Cursor are unique. With the help of Cursor, I've revamped the docs considerably, and also revised the CLI interface. The CLI piping is now there.

https://grapa-dev.github.io/grapa/about/

https://grapa-dev.github.io/grapa/cli_quickstart/

See the above for the new CLI options.

Just recently added is what appears to be the most feature rich and performant grep available.
https://grapa-dev.github.io/grapa/grep/

I will be focusing on completing the language over the next few months - it is pretty complete as it is and production ready and well tested and stable (with the exception of a ROW table type with the DB, but COL column store works very well and is quite a bit faster anyway).

I implemented the file system and DB using the same structure so you traverse the DB with similar commands you would a file system. And the DB supports multiple levels with GROUP.

https://grapa-dev.github.io/grapa/sys/file/

The docs were fully created by GenAI tools - and GenAI does not have the pattern of the grapa language, so it keeps generating grapa code with javascript or python syntax. I had it create the basic syntax doc in the docs to help constrain this, and that helps a great deal, but it doesn't always reference it. I still need to scrub all the docs to verify/fix things.

Your understanding is correct: volumeBindingMode belongs to StorageClass. The Helm chart's templating is likely using this variable to configure an associated StorageClass, and it's not being directly applied to the PersistentVolumeClaim in the final manifest.

If you find volumeBindingMode actually appearing in the rendered PVC YAML, then that would indicate an incorrect Helm chart definition, which should be reported to the chart maintainers.

https://www.parseviewerx.com/json/json_viewer
try this out

“I recently tried an online JSON viewer tool and found it amazing. I would like to implement something similar on my site to enhance usability.”

If I get it right, you can easily achieve this by using scipy.ndimage module. Like this:

import numpy as np
from scipy.ndimage import label, find_objects

DashBoard = np.zeros((10,10), dtype=int)
DashBoard[5,5] = 1
DashBoard[5,4] = 1
DashBoard[5,6] = 1
DashBoard[6,6] = 1
DashBoard[7,7] = 1
print(DashBoard)

initial Dashboard. note the "alien" at (7,7)

# define which neighbours count as adjacent (in this case, exclude diagonals)
s =  [[0,1,0],
     [1,1,1],
     [0,1,0]]
# label adjacent "islands"
labeled, nums = label(DashBoard, structure=s)
print(nums)
print(labeled)

labeled visualized:

loc = find_objects(labeled)[0]
res = labeled[loc]

res visualized:

print(res.T.shape)

will give you

(3, 2)

Only the data manipulation language (DML) changes are updated automatically during continuous migrations, whereas the data definition language (DDL) changes are the user's responsibility to ensure compatibility between the source and destination databases, and can be done in two ways—refer to this documentation.

It is also recommended to review the known limitations of your migration scenario before proceeding with the database migration. See here for the limitations specific to using a PostgreSQL database as the source.

But how will i run it on debug mode ?

@mass-dot-net:

Resolving the Cookies problem just requires creating your own implementation of HttpCookies that you can instantiate:

public class FakeHttpCookies : HttpCookies
{
  public override void Append(string name, string value)
  {
    throw new NotImplementedException();
  }
  
  public override void Append(IHttpCookie cookie)
  {
    throw new NotImplementedException();
  }
  
  public override IHttpCookie CreateNew()
  {
    throw new NotImplementedException();
  }
}

Now you can set the value in the MockHttpResponseData constructor:

public MockHttpResponseData(FunctionContext functionContext, HttpCookies? cookies = null) : base(functionContext)
{
  Cookies = cookies ?? new FakeHttpCookies();
}

Especially you want to create Indexed View (Materialized View) with UNION inside that view which is not possible: Cannot create index on view 'sql.dbo.vw_INDX' because it contains one or more UNION, INTERSECT, or EXCEPT operators. Consider creating a separate indexed view for each query that is an input to the UNION, INTERSECT, or EXCEPT operators of the original view.

But I need that view in another Query to join with.

PS: MS SQL Server 2014 Enterprise

I need help in adobe to create conditional font color for any value within a text field that is greater than 0 to be green, anything less than 0 to be red, and anything neutral to be black...? i am not a coder so any help to assist with custom java script in my pdf form would be helpful?

To limit Tailwind CSS styles to a specific React component, you can scope Tailwind classes using custom prefixes, context wrappers, or CSS Modules. Tailwind is global by default, so it doesn't naturally scope to one component

My observation is this. Schema.org content can be deployed in both the HEADER and BODY. The question is not cut and dry. The nuance is the size of the script and what tools you use to deploy.

Consider page load, time to paint implications too.

In a world of AI engines we need to tackle this problem further upstream and look to create AI endpoints with full JSON-LD Schema.org content files for real-time ingestion. Not depend on page by page loading.

There is an initiative, see git Schema.txt, to direct this traffic via robots.txt to a dedicated schema.txt file that contains links to json-ld files and catalog schema.org @IDs. AI crawlers can then access what they need to run semantic queries, that replace today's regular search. In turn this makes your Schema metadata even more precious to your future website.

When it comes down to numbers, then everybody is just "silent". Its a secret in data mining and huge conspiracy against end user ! Well i test a lot and i can only say, someone is lying big time ! All those tools aren't working the way are supposed to, results obtained in reality are clear evidence off that ! And all i hear in excuse of those tools used, "your data quality ins't good enough". What that sentence means in real, model cant learn from data (but it should due weights adjusted in model) !? Tools way to poor developed. Optuna in the wastes of HP space should find proper HP and model obtained should converge to that target. If u use the same file setup as in optuna with same classifier and target column u should get prety much the same results when test on rest file is used for that particular target column. Why those results differ so much its clue that i still figure it up. Its a math, that reproduces the same results then !? If not there is something wrong with math, weights aren't the same thus results are always worse. How to solve this i don't know. At the end i always manually tune models, find proper HP sometimes, that optuna newer finds, but it should. AI should do model HP search with all the needed metrics and evaluation plots and explanations. We don't have time to fool around with manualism's altogether. And i thought optuna will solve my problems too. Optuna at my experience newer delivers good models and i try out all the optionable metrics in optuna even custom balanced AUC original conditions formula as customizey by my ovn. Optuna simply isn't up the job. Issues for that can be very vast and all u hear is all philosophizing about it.

if you want to print hello world or just hello i know how to do that code using java

class main{

public static void main(String[]args){

System.out.println("Hello World")

}

}

if you want only hello then inside print function write ony hello

You're asking a few different things here. First, you already partly answered your question on ESM compatibility, and partly the const-style enums etc. The remaining question is how to share this between angular and nest.

Nothing in a DTO should be Angular or Nest specific. How you set it up is kinda up to you, and out of scope of at least the titular question. So I'll take the second part to try to answer.

---

What you want to do here is essentially share an typescript package, between two different typescript projects. How, depends on how your projects are set up. I can only suggest vague ideas and concepts, you'll have to either clarify the question further, or figure out the rest of the way on your own.

Do you have a monorepo set up? Something like nx.dev or turborepo? npm workspaces? They each have their own ways of setting up and sharing projects.

Otherwise, if you have separate repos, you could again do a few things differently. For one - you could install the DTO in, say, your NestJS package. You would have to export that library as an actual export. You could then add that entire package as a dependency for your Angular project. That would, of course, suck for various reasons.

The last option I'll propose is to make a third package, call it "validations" or something similar, and share that as a dependency for both other packages.

That's how you can get clean isolation, no duplication, even versioning support, support tree shaking properly (so that you actually import typescript files into your sources, before you transpile the resulting app code into javascript.

Seems to be working now in beta4.

Something like this? 

>>> p=r"([^()]*)" 
>>> re.sub(p," ","Hello (this is extra) world (more text)") 
'Hello world '

Maybe a little late for you, but the best way to do this is use useFetchClient

  const { get, post } = useFetchClient();   
  const test = await get('/users')

this uses internal admin routes with authentication

Not shared by default, but you can configure it
https://learn.microsoft.com/en-us/entra/msal/dotnet/how-to/token-cache-serialization?tabs=msal

That little “highlight” (called a tab attention indicator) is controlled by the browser itself, and you cannot manually trigger it via JavaScript. Browser only shoes this for specific built-in events like:
•alert(), confirm(), or prompt() dialogs.
•some types of audio playing.
•push notifications. It is a fundamental part of the UX of the browser, hence it can't be controlled whatsoever.

Use auto_lang_field package

I was working on a feature in a project and I want to catch the locale of the text inside the TextField to change the direction of the widget and TextStyle as well

I have build a flutter package to handle this.

Get the package

• pub.dev : auto_lang_field
• GitHub : AutoLangField
• flutter pub add auto_lang_field

For customization I also created this repo contains +81 languages data to customize the language detection if you want languages, also check the Language Detection Considerations in the package README section for more details

I really hope this package is helpful for the Flutter developers

Have you considered edge cases where the timezone shifts due to DST or other regional rules? Using Python’s zoneinfo (Python 3.9+) or pytz can help ensure the billing cycle consistently aligns with the 26th–25th range across different timezones.

Oh I got solution for this already, my problem wasn't AJV validation at all, my mistake was using a select option with a value = "", this will made the Select element required by Html.

<select>
<option value = ""> option 0 </option> <---- this will make select tag required.
<select>

Can I apply the required attribute to <select> fields in HTML?

Although this solution may not be an automation, but you can manually specify the folder name in which you want to clone your repository. Here's a way to do it:

git clone https://github.com/xyz/project1.git xyz/project1

This will clone your repository within xyz directory, in the project1 sub-directory.

have you found a solution to this problem? Thank you in advance for your response

The router address you're using (0xE592427A0AEce92De3Edee1F18E0157C05861564) is incorrect for Sepolia— that's the mainnet address for Uniswap V3's SwapRouter. On the Sepolia testnet, you should use the SwapRouter02 address at 0x3bFA4769FB09eefC5a80d6E87c3B9C650f7Ae48E instead.

In Remix, when the transaction simulates and fails, expand the error message—it often includes a specific revert reason (e.g., "INSUFFICIENT_LIQUIDITY" or "INVALID_POOL"). That can pinpoint if it's something beyond the router address.

Reverting to 1.17.5 seems to have fixed the problem for me.

I opened a gh issue so maybe it will be fixed in 1.18.2

The problem with the PIVOT function is my case I get insufficient privileges error message(s). Not sure which calls within the pivot function(s) are causing this error.

Metpy divergence computation uses the three-point method outlined in "Derivative formulae and errors for non-uniformly spaced points" Proceedings of the Royal Society A - May 2005 DOI:10.1098/rspa.2004.1430
Free download at https://www.researchgate.net/publication/228577212

Open the native tools command prompt for your platform:

Type whereis cl.exe:

SELECT
    *
FROM your_table
WHERE
    xmin = pg_current_xact_id()::xid
;

I asked this question on the Ansible forums as well and received a response.

Quick link: Error: 'ProxmoxNodeInfoAnsible' object has no attribute 'module' · Issue #114 · ansible-collections/community.proxmox · GitHub - i.e. check your proxmoxer version.

TLDR: ProxMox v.8.4 only comes with proxmoxer 1.2.0. Needs to be >=2.2. Best option suggested is to create a venv for python3 and run a special ansible user on the proxmox host.

It works for me, it is a save place ibm.biz don't worry.

to prevent potential disruptions in the main automation flow, especially in cases where the MySQL database is unavailable or encounters an error, all logging operations are executed asynchronously. This is implemented using the "Execute Workflow" node with the "Wait for Completion" option set to false. By offloading the logging process to an independent sub-workflow, the system ensures that logging tasks are executed in the background without impacting the execution or speed of the main workflow. The sub-workflow contains the logic to insert logs into the MySQL table and includes proper error handling (e.g., try-catch) to silently handle any database issues. This design pattern offers a reliable and fault-tolerant approach to logging, maintaining data traceability without compromising workflow continuity.

This code accesses the "ColumnTitle" (e.g., "Previous Value") from a SharePoint HTTP request output, extracting a list of user details with LookupId, LookupValue (name), and Email from a nested JSON structure.

DSC00005

DSC00009

DSC00016

DSC00052

DSC00064

DSC00110

DSC00111

DSC00115

DSC00122

DSC00125

DSC00127

DSC00144

DSC00158

DSC00161

DSC00170

DSC00174

DSC00179

DSC00188

DSC00198

DSC00203

DSC00213

DSC00238

DSC00253

DSC00262

DSC00273

DSC00291

DSC00307

DSC00311

DSC00320

DSC00324

DSC00340

DSC00366

DSC00393

DSC00404

DSC00423

DSC00439

DSC00453

DSC00466

DSC00471

DSC00477

DSC00479

DSC00481

DSC00485

DSC00488

DSC00493

DSC00495

DSC00496

DSC00504

DSC00505

DSC00514

DSC00518

DSC00523

DSC00527

DSC00534

DSC00538

DSC00545

DSC00559

DSC00560

DSC00562

DSC00563

DSC00569

DSC00570

DSC00576

DSC00578

DSC00581

DSC00589

DSC00592

DSC00595

DSC00611

DSC00635

DSC01622

DSC01626

DSC01628

DSC01640

DSC01641

DSC01645

DSC01648

DSC01655

DSC01663

DSC01668

DSC01671

DSC01672

DSC01676

DSC01689

DSC01708

DSC01710

DSC01712

DSC01718

DSC01725

DSC01727

DSC01734

DSC01739

DSC01741

DSC01756

DSC01759

DSC01762

DSC01767

DSC01769

DSC01774

DSC01778

DSC01784

DSC01790

DSC01800

DSC01803

DSC01808

DSC01815

DSC01816

DSC01824

DSC01826

DSC01834

DSC01838

DSC01844

DSC01845

DSC01849

DSC01851

DSC01855

DSC01858

DSC01860

DSC01880

DSC01896

DSC01904

DSC01909

DSC01912

DSC01917

DSC01922

DSC01930

DSC01932

DSC01937

DSC01943

DSC01945

DSC01948

DSC01954

DSC01957

DSC01960

DSC01964

DSC01970

DSC01972

DSC01973

DSC01979

DSC01980

DSC01981

DSC01986

DSC01991

DSC01992

DSC01998

DSC02009

DSC02013

DSC02014

DSC02017

DSC02025

DSC02027

DSC02029

DSC02038

DSC02040

DSC02042

DSC09511

DSC09534

DSC09544

DSC09548

DSC09550

DSC09559

DSC09567

DSC09579

DSC09588

DSC09592

DSC09634

DSC09640

DSC09662

DSC09666

DSC09672

DSC09680

DSC09702

DSC09707

DSC09713

DSC09716

DSC09721

DSC09746

DSC09777

DSC09845

DSC09871

DSC09945

DSC09949

DSC09955

DSC09989

IMG_1623

IMG_1628

IMG_1632

IMG_1636

IMG_1638

IMG_1640

IMG_1655

IMG_1661

IMG_1663

IMG_1666

IMG_1669

IMG_1670

IMG_1673

IMG_1674

IMG_1677

IMG_1679

IMG_1680

IMG_1683

IMG_1686

IMG_1691

IMG_1695

IMG_1697

IMG_1699

IMG_1701

IMG_1703

IMG_1704

IMG_1706

IMG_1709

IMG_1711

IMG_1713

IMG_1715

IMG_1718

IMG_1720

IMG_1721

IMG_1723

IMG_1728

IMG_1729

IMG_1732

IMG_1733

IMG_1735

IMG_1738

IMG_1740

IMG_1742

IMG_1743

IMG_1746

IMG_1748

IMG_1750

IMG_1755

IMG_1757

IMG_1760

IMG_1761

IMG_1765

IMG_1766

IMG_1769

IMG_1771

IMG_1773

IMG_1775

IMG_1777

IMG_1789

IMG_1791

IMG_1792

IMG_1797

IMG_1798

IMG_1801

IMG_1803

IMG_1805

IMG_1806

IMG_1809

IMG_1813

IMG_1814

IMG_1817

IMG_1821

IMG_1824

IMG_1827

IMG_1828

IMG_1833

IMG_1834

IMG_1838

IMG_1840

IMG_1841

IMG_1844

IMG_1845

IMG_1849

IMG_1850

IMG_1859

IMG_1861

IMG_1863

IMG_1870

IMG_1872

IMG_1875

IMG_1879

IMG_1882

IMG_1884

IMG_1887

IMG_1891

trying different things it looks like it could be the names of arguments in wnominate(). try adding "rcObject" argument:

res <- wnominate(
  rcObject = rc_samp,
  dims = 1,
  polarity = 4,  
  minvotes = 2
)

After 10 days of debugging, I found the root cause: I forgot to define the key and value in my Kubernetes ConfigMap in the Terraform config. This had several consequences:

• The application received an empty list for `allowedOrigins` in the CORS configuration

• The CORS filter couldn't match the request's origin against the empty allowed origins list

• For preflight requests, the filter simply passed the request down the chain

• Since I hadn't implemented any handler for OPTIONS requests in my application, it resulted in a 405 Method Not Allowed error

This is a network connection issue, mostly on the SAP part. If your on-prem SAP connects to GCP via Cloud VPN/interconnect, try to check the right firewall to make sure it is open for inbound connections to port 3300/33XX and also re-check your connection properties.

Hy, a bit late but i made a workaround :

I provide here the Javacode, the needed html been explained on the top.

// This Code needs a InputField with ID "Sequenz"
//  -  a div with ID "Feedback" to show keydown char
//  -  a div with ID "FeedbackU" to show input char
//  -  a div with ID "key" to show entered key 

function init(){
    document.addEventListener('keydown', function(event){keydown(event);})
    document.getElementById('Java').innerHTML="Keyress";

    //Use this Line to add Events to every Input Field, just call with the ID - Number
    AddEventtoInput("Sequenz");
}

var handled = false;
function keydown(event){    
    if(event.which == 229){
        return;
    }else{
        handled = true;
        document.getElementById('Feedback').innerHTML = event.key;
        if (keypress(event.key) == true){
            event.preventDefault();
        }
    }

    // ----- Demonstration of values. Please Delete.
    document.getElementById('Feedback').innerHTML = document.getElementById('Feedback').innerHTML + " - " + event.which;
    // ------------------------
}
function AddEventtoInput(div){
    document.getElementById(div).addEventListener("input", keyinput);
}
function keyinput(event){
    var key = event.target.value;
    key = key.slice(-1);
    if(key.indexOf('\t') != -1){key = "Tab";}
    if(handled == false){

        // ----- Demonstration of values. Please Delete.
        document.getElementById('FeedbackU').innerHTML = key;
        // ------------------------

        if (keypress(key) == true){
            event.target.value = event.target.value.slice(0,-1);
        }

    }else{
        handled = false;
    }
}
function keypress(key){
    prevent = false;
    //------------------------ Here Code to Check entered Char, Change prevent to true to prevent Char
    document.getElementById('key').innerHTML = key;
    //-----------------------
    return prevent;
}

I have written this code and it is free of use (Public Domain). Please help yourself to add it in your project.

M. Glaser

Munich Germany

Looks like this issue has been fixed in the last few versions of Angular.

Here is an example of Angular Material components working inside a component with encapsulation set to ShadowDom:

https://stackblitz.com/edit/angular-material-components-in-shadow-dom?file=src%2Fexample%2Fdate-range-picker-overview-example.ts

And if you need the contents of cdkOverlay, modals, etc to be inside the shadow dom, here is a proof of concept doing that:

https://stackblitz.com/edit/angular-overlay-inside-shadow-dom?file=src%2Fshadow-dom-with-overlay-override%2Fshadow-dom-with-overlay-override.ts

It seems like the n=n+1 is inside the for loop, which may be causing it to increment multiple times unexpectedly. You might be skipping some indexes or going out of range. Try moving the increment outside the for loop or consider using a nested loop instead of combining while and for.

I am writing a Python compatible syntax, but the concept should work for C++ too.

comboBox.setEditable(True)
comboBox.lineEdit().setAlignment(Qt.AlignCenter)
comboBox.lineEdit().setReadOnly(True)

Idea is, to make it editable, but setting lineEdit inside QComboBox as read-only later.

This code defines a function to find matches of a regex pattern that occur only outside paranthasis in a string by manually tracking nesting depth, ensuring precise, context-aware pattern matching.

One alternative is the Boxtin project, but as of now, it's not yet released. "Boxtin is a customizable Java security manager agent, intended to replace the original security manager"

You should add missing foreign keys to ensure data integrity and generate accurate database diagrams. However, don’t add all of them at once — do it in phases. First, check for invalid data that may violate the constraints. Adding too many FKs at once can cause locks, performance hits, or app errors if the data isn’t clean. Test changes in a dev environment, fix any orphaned rows, and roll out gradually. This improves design, avoids bad data, and helps tools like SSMS show correct relationships.

This happened to me, using the JUCE library with Projucer. Uninstalled the app from my phone, then ran it on Android Studio again.

I think this AI-assisted answer is the clearest:

=EOMONTH(A1,0)-MOD(WEEKDAY(EOMONTH(A1,0),2)-5,7)

A1 is a date value with any date in the subject month

-5 is for Friday. For other days:
-4 is for Thursday
etc for the sequence -1 (Monday) through -7 (Sunday)

Can someone create a zig zag indicator with ATR filter?

This is such a good game. This code implements a basic Tic-Tac-Toe game using Jetpack Compose in Kotlin, handling game logic, state management, UI rendering, and player turns with reactive updates and modal for end-game results like win or draw.

I was able to make this work by changing the 'print_info' argument to 'print_stats'. The documentation for the tg_louvain graph data science library function (https://docs.tigergraph.com/graph-ml/3.10/community-algorithms/louvain) incorrectly mentions the argument to print the result as 'print_info', I simply referred to the actual implementation here https://github.com/tigergraph/gsql-graph-algorithms and figured out what the correct argument name was.

Tried using the server’s own cert for verify, but it failed with an issuer error. Makes sense now: it’s just the leaf cert, not the whole trust chain.

This version has a conversational tone and reflects your own debugging experience, like how real developers talk when they’re troubleshooting quirks.

how's this?
I had to get rid of the sub picker each time the category changed... redundant code... but this worked...

struct ContentView2: View {
 enum Category: String, CaseIterable, Identifiable {
  case typeA , typeB , typeC
  var id: Self { self }
  var availableOptions: [ Option ] {
   switch self {
   case .typeA : return [ .a1 , .a2 ]
   case .typeB : return [ .b1 , .b2 ]
   case .typeC : return [ .c1 , .c2 ]
   }
  }
 }

 enum Option: String, CaseIterable, Identifiable {
  case a1 , a2 , b1 , b2 , c1 , c2
  var id: Self { self }
 }

 @State private var selectedCategory: Category = .typeA
 @State private var selectedOption: Option = .a1

 var body: some View {
  Form {
   Section ( header: Text ( "Selection" ) ) {
    HStack {
     Picker ( "" , selection: $selectedCategory ) {
      ForEach ( Category.allCases ) { category in
       Text ( category.rawValue.capitalized ) .tag ( category )
      }
     }
     Spacer()
     switch selectedCategory {
     case .typeA:
      Picker ( "" , selection: $selectedOption ) {
       ForEach ( self.selectedCategory.availableOptions ) { option in
        Text ( option.rawValue.uppercased() ) .tag ( option )
       }
      }
     case .typeB:
      Picker ( "" , selection: $selectedOption ) {
       ForEach ( self.selectedCategory.availableOptions ) { option in
        Text ( option.rawValue.uppercased() ) .tag ( option )
       }
      }
     case .typeC:
      Picker ( "" , selection: $selectedOption ) {
       ForEach ( self.selectedCategory.availableOptions ) { option in
        Text ( option.rawValue.uppercased() ) .tag ( option )
       }
      }
     }
    }
   }
   .labelsHidden()
   .pickerStyle ( .menu )
  }
 }
}

#Preview {
 ContentView2()
}

When I need to inspect multiple popup or dropdown elements that disappear on blur, I use this handy snippet:

1. Open DevTools → Console

2. Paste the code below and press Enter

    window.addEventListener('keydown', function (event) {
       if (event.key === 'F8') {
         debugger;
       }
     });
   

3. Press F8 (or change the key) anytime to trigger debugger; and pause execution

This makes it easy to freeze the page and inspect tricky UI elements before they vanish.

👋

You're asking a great question — and it's a very common challenge when bridging string normalization between Python and SQL-based systems like MariaDB.

---

✅ Answering your question directly:

Unfortunately, MariaDB collations such as utf8mb4_general_ci are not exactly equivalent to Python's str.lower() or str.casefold(). While utf8mb4_general_ci provides case-insensitive comparison, it does not handle Unicode normalization (like removing accents or special casing from some scripts), and it’s less aggressive than str.casefold() which is meant for caseless matching across different languages and scripts.

---

🧠 What's the difference?

• str.lower() only lowercases characters, but it's limited (e.g. doesn't handle German ß correctly).
• str.casefold() is a more aggressive, Unicode-aware version of lower(), intended for caseless string comparisons.
• utf8mb4_general_ci is a case-insensitive collation but doesn't support Unicode normalization like NFKC or NFKD.

---

💡 Recommendations

1. Use utf8mb4_unicode_ci or utf8mb4_0900_ai_ci (if available):

   • These are more Unicode-compliant than general_ci.
   • Still, they won’t match Python's str.casefold() completely.

   Example:

   CREATE TABLE example (
       name VARCHAR(255)
   ) CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci;
   

2. Normalize in Python before insert: If exact normalization (like casefold() or unicodedata.normalize()) is critical, consider pre-processing strings before storing them:

   import unicodedata
   
   def normalize(s):
       return unicodedata.normalize('NFKC', s.casefold())
   

3. Store a normalized column: Add a second column that stores the normalized value and index it for fast equality comparison.

   ALTER TABLE users ADD COLUMN name_normalized VARCHAR(255);
   CREATE INDEX idx_normalized_name ON users(name_normalized);
   

4. Use generated columns (MariaDB 10.2+): With a bit of trickery (though limited to SQL functions), you might offload normalization to the DB via generated columns — but it won't replicate Python's casefold/Unicode normalization fully.

---

🚫 TL;DR

There is no MariaDB collation that is fully equivalent to str.casefold(). Your best bet is to:

• Normalize in Python before insertion.
• Store and index a normalized version in the database.
• Use utf8mb4_unicode_ci for better Unicode-aware comparisons than general_ci.

---

Hope that helps — and if anyone found a closer match for casefold() in SQL, I’d love to hear it too!

Thank you for the outline on how to perform y-direction scrolling ! Thanks to the outline, I was able to perform x-direction scrolling...I thought that I would share it below.

<!DOCTYPE>
<html>
<head></head>
<body>
<div id="div0"></div>
<style>
#div0 { 
  position: absolute; 
  top: 0px; 
  display: flex;
  flex-direction: row;
  height: 100px; 
  max-width: 100px;
  overflow-x: scroll;
  scroll-snap-type: mandatory;
  scroll-snap-points-x: repeat(100px);
  scroll-snap-type: x mandatory;
} 
  
.div_simple_ { 
  position: relative;
  border: 1px solid gray; 
  background-color: seagreen;
  border-radius: 5px; 
  width: 100px;
  scroll-snap-align: start; 
  height: 100px;
  line-height: 100px;
  text-align: center; 
  font-size: 30px;
  flex: 0 0 auto;
}
</style>
<script>
var n = 3;
create_elements();
function create_elements() { 
  
  for (var i=0; i<n; i++) { 
    var divElement = document.createElement("div");
    divElement.setAttribute("id", "div_"+i);
    divElement.setAttribute("class", "div_simple_");
    
    divElement.innerText = i;
    document.getElementById("div0").appendChild(divElement);
  }

} 
</script>
</body>
</html>

You can go to: vendor/maatwebsite/excel/src/Sheet.php

Go to line 675 and change your chunk size from 1000 to (x)

I just noticed something very strange: when I drag the node to the top (as in the image below), the flow starts to work correctly. Now it executes as expected, does the insert (1) when passing the node, and executes the insert (2) at the end, all correctly... I dragged the node down and up several times to be sure, and this does indeed happen: when it's at the bottom, it waits for the last node, when it's at the top, it executes instantly. Is this an expected result?

Watch the video below:
https://drive.google.com/file/d/1HMM-hv1gjyZ6EUJ9dBu-pARknkRPN3_p/view?usp=sharing

And also, I would like to add to that sometimes when you have a project structure where your controllers need to be in a different package than your SpringBootProject class annotated, you can add how many path to be scanned by Spring boot, just like

@ComponentScan({"com.path1", "com.path2"})

There have been several posts elsewhere related to this topic. I experienced the same problem.

Some of the answers appear to me to be mumbo-jumbo -- the answers may indeed work around the issue but do not explain the cause of the problem (eg deleting old SDKs, using the command line tools infrastructure, reinstalling from HomeBrew..).

The only solution which worked for me is the one outline above: using

-isystem /Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/c++/v1

The -isystem command line argument is not described in the man pages on my ssystem (Sequoia 16.4) although it does seem to be documented at certain places online.

Surely this is a bug in Xcode's compile? How is it that the compiler cannot figure out, when compiling C++ code, that the C++ includes need to be queries before those from the standard C library. And how is that Apple's man page is deficient?

With running docker build and docker exec / run very often for the same dockerfile / compose i created a mess of unused images / data. After i purged all data and started over fresh the problem was solved.

What you observe is not by design, but a bug. I filed a ticket for it: https://issues.apache.org/jira/browse/KAFKA-19539

Are you a dumb nigger? Can't you uh comprehend how to format the fucking code without extra spaces before every line? FUCK OFF BRO

You are a clown.

Anyway here's the fix, you just idk DO PROPER TABULATION AND YOU ARE ALL SET. No syntax error, bro.

Off to gulag, bro, off to gulag.

They have a paper for the Mediapipe hand detection feature that talks about the general architecture and methods used. You can see the paper here: https://arxiv.org/abs/2006.10214

The exact weights and model architecture are still not available I am sure.

Alternatively, if you find CRON syntax hard to read, I maintain an open-source library called NaturalCron that lets you write human-readable recurrence rules like every 5 minutes between 1:00pm and 5:45pm

Very help full. This save alot of my time. I even tried AI like chatgpt and grok but not fruitful. this helped alot.

The package developer here, thanks for bringing this up.

You might be able to solve the issue by specifying "binomial" instead of binomial. If this does not help, please share full code and data (data need not be full data, a subset of rows is fine) so I can reproduce the issue and fix.

Apache Tika uses language profiles from Optimaize Language Detector, which are based on statistical n-gram models. If Farsi isn't recognised, it typically means that:

1. The text sample is too short or ambiguous, or

2. The language profile for Farsi (fa) is missing

You could try;

1. use a longer or more diverse sample in Farsi

2. make sure you are using the OptimizeLangDetector, and

3. ensure you're using a recent Tika version

What you mentioned is true. This is not IntelliSense , but IntelliCode. The settings to disable it may be under Text Editor > C# > IntelliCode .

Using Snowsight to get connection settings

I do think you can uninstall GitHub Copilot from VSCode to solve that, but probably your problem can be solved by the question below.
How do I disable partial line code suggestions in Visual Studio Code?

I just noticed something very strange: when I drag the node to the top (as in the image below), the flow starts to work correctly. Now it executes as expected, does the insert (1) when passing the node, and executes the insert (2) at the end, all correctly... I dragged the node down and up several times to be sure, and this does indeed happen: when it's at the bottom, it waits for the last node, when it's at the top, it executes instantly. Is this an expected result?

Check this shouldRevalidate

https://remix.run/docs/en/main/route/should-revalidate

We had issues as well with cart with MedusaJS + remix integration since everytime we added something to the cart it was making all the root calls.

When we added should revalidate on index route it would only revalidate the thing it wanted.

Did you manage to solve it? If so, how?

If your Python script in VS Code shows no output at all (not even print statements), it's likely a VS Code configuration issue rather than a problem in your code. First, confirm you're running the correct script by adding print("Script started") at the top. Try running the script manually in the terminal (python DBCon.py) to see if output appears—if it does, the issue is with VS Code's execution settings. Also, ensure the correct Python interpreter is selected (Ctrl+Shift+P → "Python: Select Interpreter") and that mysql-connector-python is installed. Avoid running in the interactive window by disabling related settings, and make sure to look at the TERMINAL, not the OUTPUT panel, for results.

On Linux it is possible to create a QPA plugin (platform plugin) that uses the Mesa library to do software rendering. The coding is not easy, but you only need to do it once, and it works for all your apps. I had to build Qt from source and write the QPA based on one of the existing plugins. It then renders to a memory buffer you can save as an image - very useful for testing.

I'd recommend you to implement lazy loading to reduce bundle size, also check if your files are gzipped and, if it's not, gzip them.

Also, are you sure all 150 NgModules are needed? Perhaps some could be eliminated without breaking your application.

Check this blog post.

To force execution order, you must connect Node 1 directly to Node 2 or create a dependency chain, like this:
Option 1: Direct Sequential Link Connect InsertMsgRecebida → InsertMsgEnviada, even if they are logically unrelated. This ensures n8n waits for the first insert to complete before continuing.

Option 2: Merge with Wait Use a Merge node in "Wait" mode to join both paths.
Flow: InsertMsgRecebida → Merge InsertMsgEnviada → Merge This ensures both writes must complete before the flow ends.

Option 3: Use a “Set” Dummy Chain If a direct link doesn’t make sense, use a dummy Set node to bridge: InsertMsgRecebida → Set → Merge → InsertMsgEnviada

Why "Waiting for end of stream" Happens in MySQL Node? This message is often seen when:
The MySQL node is waiting for the database connection to finish the current insert/update. This can also appear if there’s backpressure from async branches, especially when one flow is writing and another is blocked. Making sure writes are sequential and not parallel avoids this.

Final Flow Suggestion WhatsApp Trigger -> InsertMsgRecebida (Node 1) -> ... (rest of your flow) ↓-> InsertMsgEnviada (Node 2) Or use a Merge (Wait) node to enforce sync.

A thing I did was pickled the BM25Retriever.from_documents() at indexing time not run time. I'm not sure if it's the right decision, but let me know!