Just incase anyone finds this. PRAGMA journal_mode=DELETE doesn't turn off the journal. It just says that when the transaction is finished the journal file should be deleted rather than being truncated or overwritten. If you really want to turn off the journal, which you shouldn't, then you can use PRAGMA journal_mode=OFF.^**

See https://sqlite.org/pragma.html#pragma_journal_mode for more information.

** Or at least in theory. Right now PRAGMA journal_mode=OFF is not working for me, but that's another story.

Disable

Options -> IntelliCode -> General -> Promote lilkely items in IntelliSense completion lists

Use rdctl shell instead of wsl -d

Because Rancher Desktop creates a network namespace, and rdctl shell gets you inside that namespace

You need to use the same react and react-dom version for your packages/apps in your monorepo. Also put ^ before the version number.

For this specific case change your apps' A and B package.json dependencies to this:

    "react": "^19.1.0",
    "react-dom": "^19.1.0"

Source: There's a comment in Vercel's next repo that explains it in more detail: link

I have found to open gmail on android the url link (to inbox not compose)

await Linking.openURL('https://mail.google.com');

NPOI official tutorial for custom color:
https://github.com/nissl-lab/npoi-tutorial/tree/main/sample10

You will want to use OneToOneFields for the relationship, that way it guarantees there is at most one of each of the other table rows for the your table:

https://docs.djangoproject.com/en/5.1/topics/db/examples/one_to_one/#one-to-one-relationships

To do this all via forms, you will want to look into model formsets.

Verify that Collect_slow_queries: true is actually enabled.

Your YAML looks correct, but verify that it is actually being used with:

cat /etc/newrelic-infra/integrations.d/postgresql-config.yml

And then restart the agent:

sudo systemctl restart newrelic-infra

I will use these--and although the groups are not necessary, they make it clearer to me what I am doing.

^([0]+)\s([0]+)\s([0]+)$
^(0*[1-9]\d*)\s([0]+)\s([0]+)$
^(0*[1-9]\d*)\s(0*[1-9]\d*)\s([0]+)$
^(0*[1-9]\d*)\s(0*[1-9]\d*)\s(0*[1-9]\d*)$

Thank you, @sln

I'm revisiting an issue I encountered a long time ago. I now consistently use this method to handle UTF-8 encoding. It requires Windows, but it works reliably with C/C++. I'll provide the corrected source code that delivers the best results.

#include <stdio.h>
#include <stdbool.h>
#include <windows.h>

bool file_read(char name[]){
    FILE *files = NULL;
    files = fopen(name, "r");
    if(files == NULL){
        return false;
    }else{
        char Carack;
        while ((Carack = fgetc(files)) != EOF){
            printf("%c", Carack);
        }
    }
    printf("\n");
    fclose(files);
    return true;
}



int main(){

    SetConsoleOutputCP(CP_UTF8); // UTF 8 support

    file_read("d.txt");

}

I think I found a workaround to satisfy this question. I couldn't get @kindall 's answer to work for some reason. So I'm using this:

emails = [email[:email.find('@')] for email in user_emails] # return everything up to the @

This will return everything up to the @ symbol. If you want everything after the @ symbol, run the following:

emails = [email[email.find('@'):] for email in user_emails # returns everything after the @ symbol

That backs up the images, but not the container volumes. I'm still trying to find a way to do that as well.

Thank you to both Daniel Cruz and fdomn-m for your comments. Indeed the root cause of this confusing error was conflicting jQuery loads. Although my code did not load jQuery multiple times the template my page was using did. Once I remove the extra jQuery import everything worked as expected, regardless of where the .jstree() method call was located.

Cheers!

Check out the Django pagination class for some inspiration (or just use it directly :) ):

https://github.com/django/django/blob/main/django/core/paginator.py

You may find this article useful, it lists some more techniques such as Cosine and Jaro-Winkler:

https://medium.com/@appaloosastore/string-similarity-algorithms-compared-3f7b4d12f0ff

I tested it with MouseActivate (override of the function originally existing in TControl) and it seems to work correctly, even if the parent TForm is already active.

Another quick and simple solution for this is:

Open the Command Prompt (run with administrator), and

net stop winnat

and then

net start winnat

Then, start the service again.

This was extremely useful to me, thank you.

On my Mac commandline, this Just Worked, first time:

gs -sDEVICE=tiffsep -o separation%d.tif test.pdf

Flutter IOS, OneSignal app. I have the same problem.Although I am sure the phone, ios, gets the notification, the app never invoques or runs the app delegate function attached, so no action can be taken. Is notification handled somewhere else? Please help.

 override func application(
    _ application: UIApplication,
    didReceiveRemoteNotification userInfo: [AnyHashable: Any],
    fetchCompletionHandler completionHandler: @escaping (UIBackgroundFetchResult)

Whoever looking for a solution in 2025

First clear the gradle cache with this command:

rm -rf ~/.gradle/caches

And try enabling K2 mode from :

Settings -> Languages & Frameworks -> Kotlin

torch.multiprocessing.cpu_count()

by default, phpmyadmin will check foreign key, you must un-check, otherwise, you will get this error

When you write code in a programming language like C/C++, actually you are working with a compiler (I'm gonna talk about gcc, which is a C/C++ compiler) which takes your file (myfile.c), read it and "elaborate" what you wrote, doing optimizations when is possible.

Now, for the code you wrote, actually the compiler is working on int, so integers. Integers, in languages like C/C++, are just normal numbers, nothing more (just don't think that integers are a bunch of 1 and 0, just thinking at an high level) nothing less.

Actually, the code you wrote is saying "I want to create an integer n, and an integer index, where n has an x value and index has y value". Now, since you are working with integers (remember, just numbers), the compiler can actually evaluate the right-hand side of your expression (remember this term, also known as RHS) and assign the value to the LHS (left-hand side), where RHS is 54 and LHS is n. So, n will have a value, evaluated from the calculations of the RHS, and index will have another value, evaluated from the calculation of the RHS 2 * n. BUT, since n is an integer (just a number) there is no need to remember the fact that index is evaluated starting from n, because n = 54, so the compiler just substitute n with 54 and return the value 108.

You may asking "Why, why can't the compiler just think that like a math function". The answer is "C/C++ is not a mathematical language, but a programming language, so it's optimized to be a programming language". Actually, since you are interested in evaluating an int, the compiler just substitute the value of the variable with the actual value. For love of clarity, the compiler actually has to produce an assembly code. In assembly, to evaluate a variable you can only use basic math (addition, subtraction, multiplication and "division" (not exactly right, but fine for now)). If you just substitute the value of n with 54, as the compiler does, when evaluating index you just have to do a basic multiplication (one operation), meanwhile if you don't, you must use an instruction to recover the value in n (a load operation), then you have to store it in a new register, do the multiplication and store the result in the register of index, four operations.

Since C/C++ is not thinked to "work like math", compiler just does the substitution in order to optimize the code, so instead of four operations in assembly you do just one (if you are interested)

Notice that this is not an universal rule, if you use something like Python, where actually an int is not a number but a class, all what I said won't work, because actually Python is thinked not to create and optimize code, but for syntax clarity and a 'non-violent' approach for programming.

Since the compiler does the constant folding operation (the substitution), index won't be linked to n, so if you modify n this won't affect index. If you want to change this, I suggest you to study what a struct is and play with it

Greetings

For me, I needed to use https://github.com/softwareventures/resolve-typescript-plugin in order to get ts-loader to be able to handle importing TS files using the import ... from '.../*.js'; syntax. An alternative is upgrading to Webpack v5. See Webpack/ts-loader import with .js extension not resolving

This alternative seems to work:

createbucket:
  image: minio/mc:latest
  networks:
    - main-network
  entrypoint: >
    /bin/sh -c "
    mc alias set s3 http://minio:9000 <name> <password>;
    mc mb --ignore-existing s3/media;
    mc anonymous set download s3/media;
    "
  depends_on:
    - minio

I.e. I used mc alias set instead of mc config host add.

I don't have an answer, just posting to say that I also have this issue. This is in a Jupyter Notebook file in VS Code.

Example:

import duckdb
import pandas as pd

# Example data
data = {
    "ID": [1, 2, 3],  
    "Name": ["Alice", "Bob", "Charlie"],  
    "Score": [85.5, 90.0, 78.0] 
}

# Create the DataFrame
df = pd.DataFrame(data)

# Make DuckDB Connection, write, and try to close conn.
with duckdb.connect("test.duckdb") as conn:
    conn.execute("CREATE TABLE IF NOT EXISTS table_name AS SELECT * FROM df")
if conn: 
    print('still conn')
    conn.close()
conn

This prints "still conn" and also displays the whole connection string when "conn" is called on the last line. So neither the context manager nor the conn.close() appear to close the connection.

Restarting the kernel DOES work, but I think this is annoying enough that it may keep me from using this in its entirety. Or not use DuckDB in notebooks, I guess?

Another option is to use GetMethods and filter using linq:

returnType.GetMethods().Single(mi => mi.Name == "Parse"
    && mi.GetParameters().Count() == 1
    && mi.GetParameters()[0].ParameterType == typeof(string))

In that case you can also check for generic parameters if necessary.

All these approacehs mentioned above are correct...

Finally I chose this -

golden_final=golden.replace('_$', r'_\$')

Revise your toggle code to clearly include or exclude classes according to the existing state:

document.querySelectorAll('.toggle-password').forEach(icon => {
  icon.addEventListener('click', () => {
    const targetId = icon.getAttribute('data-target');
    const input = document.getElementById(targetId);
    if (!input) return;

    const isHidden = input.type === 'password';
    input.type = isHidden ? 'text' : 'password';

    // Update icon classes explicitly
    if (isHidden) {
      // Show as slashed eye
      icon.classList.remove('fa-eye');
      icon.classList.add('fa-eye-slash');
    } else {
      // Show as eye
      icon.classList.remove('fa-eye-slash');
      icon.classList.add('fa-eye');
    }
  });
});
                

Working codepen

For now, use the free key and cert provided in QZ Tray menu | Advanced | Site Manager | "+" | Create New.

Could you explain this to me step by step with js on the front and php on the back?

when i ran any flutter command, it suck without any output in flutter 3.16.0 Windows, i tried to kill all tasks with name of Git Windows.exe in Task Manager and it worked and command immediately began to work.

I was facing same issue since a long time

Here, I made a whole app about this - BharatGraph - https://bharatgraph.byvaibhav.com/

The public API documentation
https://github.com/mr-vaibh/BharatGraph?tab=readme-ov-file#-public-api

Merge the assemblies prior to obfuscation. You can use something like ILMerge to merge the assemblies into a single DLL (which is what Eazfuscator uses behind the scenes i believe?). After that is complete you can proceed with your obfuscation. This can be automated in your csproj as a build target as well.

If you want a really simple way to print a certain number of items in the fibonacci sequence you can do this:

num = input("How many numbers of the fibonacci sequence would you like to print?\n\n")
numbers = int(num)

a = 0
b = 1
print(a)
print(b)
for i in range(numbers):
    c = a + b
    print(c)
    a = b
    b = c

Since the rule of the fibonacci sequence is to have each item be the sum of the previous two items, you can create two variables, add them then set the value of the variables a and b to the new values of b and c and continue the loop.

The issue was that my app was being suspended by iOS in the background before it could prepare and play the sound. To fix this, I used a background task to request execution time from the system while preparing and playing the audio.

var bgTask: UIBackgroundTaskIdentifier = .invalid

func beginBackgroundTask() {
    bgTask = UIApplication.shared.beginBackgroundTask {
        UIApplication.shared.endBackgroundTask(self.bgTask)
        self.bgTask = .invalid
    }
}

Call beginBackgroundTask right before your code to play the audio.

beginBackgroundTask()
guard let url = Bundle.main.url(forResource: "TimerDone", withExtension: "wav") else { return }
do {
    audioPlayer = try AVAudioPlayer(contentsOf: url)
    audioPlayer?.play()
} catch {
    print("Audio playback error: \(error.localizedDescription)")
}

You need to add the complete configuration you have in GCP, I recommend adding screenshots of the flow.

This only works if your starting in the same directory as the file you want copied.

Example:

copy "%~dp0\wordpad.exe" "c:\Program Files\Windows NT\Accessories\"

How can I guarantee the correctly populated JSON objects on every server?

There's no guarantee, until you know which configuration is used on remote server.

How can I diagnose what is causing the difference between outputs?

There's an option for remote debugging if you attach sources to you application.

The Flutter SDK for Firestore always loads full documents. There is no option in it to load a partial document, or only document metadata.

If your use-case requires that you load a subset of the data, the common approach is to create an additional collection where you store documents with just the data you need. If you use the same document IDs in that collection, it's easy to map back-and-forth between them.

Also see:

• Firestore fetch document ids without fetching document data
• Get certain fields in document firestore flutter
• How to fetch a single Field from a Firestore document in a collection from Firebase?

The actual problem was due to no recomposition of when the state change. I was maintaining three states for different index but it can be maintained by single index which resolved the issue.
I have done lots of improvements like using key, content type and addition of intervals to support multiple calls of methods for LazyStackItemScope just like LazyRow /Column.
https://jmp.sh/k5NdTy0V

Essentially, in a relational data model, a unidirectional @OneToMany relationship is a type of relationship between two tables where one table has multiple related records in another table. Still, the second table doesn’t directly relate to the first table. This means that the relationship flows in only one direction.

Consider moving into the JPA, a unidirectional @OneToMany relationship can be established between two entities when one entity has a reference to the collection of related entities. Still, you cannot traverse back from the related entities to the first entity.

Generally, the entity containing the reference is called the parent entity, and the referenced entity is called the child entity.

The revalidate value needs to be statically analyzable. For example revalidate = 600 is valid, but revalidate = 60 * 10 is not.

NextJS revalidate docs

This was failing because my revalidate used 3600 * 24 for 24 hours. This is invalid config as the docs state.

The way I got past this was to blow away the venv entirely, and start fresh:

rm -rf /opt/netbox/venv
python3 -m venv venv
source /opt/netbox/venv/bin/activate
pip install -r requirements.txt
python netbox/manage.py runserver 0.0.0.0:8000 --insecure

I also made sure that the interpreter selected in VS Code was ./venv/bin/python once the new venv was created.

Still not sure why the execution context for both the debugger and the command line escaped out of the venv. That was super weird. If anyone has any idea, it would certainly help me sleep at night.

In the end, the answer was useBlocker but that required re-factoring the route handler to createBrowserRouter and directing the page content through an <Outlet /> in the RootLayout instead of having each page produce it's own child <RootLayout>

A bunch of work, but it stops exit before context is lost.

I had a similar one shot need, here is a strait forward logic hack to do the trick -- some one write a script to expand for any number bits, I wrote it out for 16 -- example for 4.

You pick what you want in case all bits are zero, I put -1 as a template.

Plus it takes a start bit position to start looking from.

 case (start_bit)
    0: bit = x[0] ? 0 : x[1] ? 1: x[2] ? 2: x[3] ? 3: -1;
    1: bit = x[1] ? 1: x[2] ? 2: x[3] ? 3 : -1;
    2: bit = x[2] ? 2: x[3] ? 3 : -1;
    3: bit = x[3] ? 3 : -1;
endcase 

There are unofficial bindings now: https://github.com/rcalixte/libqt6c

This will resolve your issue:

!pip install -U datasets

Resolved.

Turns out, I pasted a block that included the LSApplicationQueriesSchemes key with only snap defined, and that was scheme that got recognized in place of scheme what I send.

There is no need to use the .replace method.

This can be achieved by escaping the escape character which would print \

#/usr/bin/python3

x = "foo\\$"
print(x)
#outputs foo\$

When defining your materialized view, the columns in your SELECT are not matching the columns in the destination table. You are simply missing an AS mtm_sum clause in your view:

CREATE MATERIALIZED VIEW crypto_trade_mtm_aggs_view
TO crypto_trade_mtm_aggs_dest
AS SELECT 
    counterparty, 
    sumForEachState(mtm) AS mtm_sum
FROM crypto_trade_mtm
GROUP BY counterparty;

When you use JPA applications, you often encounter entities with parent-child relationships. Persisting these entities efficiently involves saving the parent and automatically persisting its associated child objects. In this answer I will give you details that explores how to achieve this automatic saving for both unidirectional and bidirectional relationships.

I was also struggling with that same error message [firebase_functions/unauthenticated] >UNAUTHENTICATED. In my case, there was a policy at the organization level that required all calls to Cloud Run functions to be authenticated by IAM and for users to belong to the organization, so I could not add the allUsers principal to the function. Instead, I had to turn off the IAM authentication for the function, which can be done by selecting the function and in the security tab, uncheck "Use IAM to authenticate incoming requests.

Note that during deployment of the function, you may still get an error message that the IAM could not be changed to public, this is fine, the function is deployed, you just need to apply that change afterwards.

Official Google documentation and thanks to Tom Elliott for pointing me towards the solution.

Uncheck IAM Authentication in the Security tab of the function

Did you manage to solve it? I'm having the same issue trying to create the connector profile fully automatic

I have an answer to my mystery.

The problem was in the definition of the module B's packaging. It was:

  <build>
    <plugins>
      <plugin>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-maven-plugin</artifactId>
      </plugin>
      <plugin>
        <groupId>org.apache.maven.plugins</groupId>
        <artifactId>maven-surefire-plugin</artifactId>
      </plugin>
    </plugins>
  </build>

So, the problem was that spring-boot-maven-plugin produces Fat Jar (executable) and basically has different structure that a plain library.

The package that I needed was unpacked to:

BOOT-INF/classes/my/needed/class/Clazz.class

instead of

my/needed/class/Clazz.class

at the root level as in a regular Jar.

Interestingly, IntelliJ was fine with both.

The solution would be to move the common classes to module C which acts like a library, which will sadly require quite a lot of effort, but at least my mystery is resolved now.

I wrote a short article on this topic—might be helpful for you: https://medium.com/@oryantechs/write-css-faster-with-this-alternative-to-tailwind-global-css-files-styled-components-644fd76486e8

With angular 19 and tailwind 4.1, I simply add @source in my main style.scss, and all tailwind CSS classes used by my lib are correctly exported

@use "tailwindcss";
@source "../node_modules/my-lib";

From documentation : https://tailwindcss.com/docs/functions-and-directives#source-directive

Use the @source directive to explicitly specify source files that aren't picked up by Tailwind's automatic content detection:

The ChessScanner app provides QR codes and URLs like https://chessqr.com/viewer?q=01GLEcQZ8IcFcjpjvG9BSfr3&w=MFH&b=N.N.

I have not yet been able to reverse-engineer the "q" value. It seems to start always with 01, I guess this is just a version specification. Thereafter, it seems that there's an urlsafe_base64 encoded string (using '-_' instead of '+/' w.r.t. standard base64), although in some rare cases it has length 4k+1 which can't be a valid base64 string... but always including the "01" or just the "1" doesn't work out better.

In any case I think that would be the way to go (use urlsafe_base64 encoding for the move list, see below for more) for your purpose.

The length of chessqr's q-string, without the initial "01", is about twice the number of moves (here "01"+ 22 characters for a game of 11 moves). Since one character represents 6 bits which could represent exactly one square of the 8x8 chess board, one might be tempted to think that it could be simply the UCI move specification, i.e., departure square + destination square (+ 2 (or 3 or 4 or 6) bits to specify the piece in case of promotion). However, it doesn't seem to be that... (I would be much interested if someone knows or would like to [maybe jointly] figure this out!)

Another way that has been considered for recording chess games in very few bits per move (cf. this discussion on rec.games.chess) is to create a list of moves, which will usually give much less than 64 x 64 possibilities: one can expect an average of about 40 moves which should usually fit in a single 6-bit number. I guess it could be possible to construct positions with more than 64 possible moves, but then you could use 0o77 (octal 63) as an "escape code" for, say, a move number with 2x6 bits.

OR, you use a variable bit size and use only as many bits for the next move as correspond to the maximum number of moves (i.e., as long as there are ≤ 32 moves possible, use 5 bits; as long as there are ≤ 64 moves possible, use 6 bits, etc.). I guess that might be very efficient under the assumption that very often, there might be even less moves available.

You might still want to use an "escape code" (maybe zero or the maximum value with all bits 1) to introduce a "special move", e.g. "end of game" (resign or draw...) or to allow for illegal moves (then probably the UCI notation will be required, to give both, starting and destination square).

For the metadata (PGN header), I think it should be OK to give the data in the form of "&variable=value" pairs, but if you really want to get it as short as possible, you could obviously also use zlib + urlsafe_base64 encoding of the compressed data.

After I tried some methods, I finally created a backup of the Android folder and then deleted the Android folder and rebuilt it again, problem solved and I moved changes back from android-copy to Android folder

Your platform version is not the same as your IPHONEOS_DEPLOYMENT_TARGET. They should be the same version number.

platform :ios, '12.0'

I found this blog post and was able to adapt my code and was able to pass the ajax url to the front end script. see below:

function add_my_script(){

$ajax_url = [
    'ajax_url' => admin_url( 'admin-ajax.php' ) 
];

$my_ajax_object = sprintf('window.%s = %s', 'my_ajax_object',json_encode($ajax_url));
    
wp_enqueue_script( 'modTest', get_stylesheet_directory_uri() . '/bsd-js.js', array ( 'jquery' ), false, true);
addModuleTag('modTest');
wp_add_inline_script('modTest',$my_ajax_object,'before');
wp_enqueue_script_module('modTest2',get_stylesheet_directory_uri() .'/Modules/testMod.js');
}

function addModuleTag($scriptHandle){
add_filter('script_loader_tag', 
function($tag, $handle,$src) use ($scriptHandle) {
    // Check if the script handle matches the one we want to modify
    if ($handle === $scriptHandle) {
        // Add the type="module" attribute to the script tag
            return str_replace(' src', ' type="module" src', $tag);
    }
    }, 10, 3);
}

The above code will result in being able to use my_ajax_object.ajax_url in the javascript module file(s). The addModuleTag function when the matching script handle is found will change the <script> tag type of the bsd-js.js file I enqueued to a module. wp_add_inline_script will send my ajax url to a script tag in the js file so it can be used. Next item to look into is how to get jquery also into the module

Little bit late, but I have somewhat of a similar problem when I publish my blazor webasembly app to github pages att https://mid-d-man.github.io/AirCode/ it loads up fine but when I try to go to another page or I get redirected back from my auth0 universal login I get a 404 error and it says something about csp content violation favicon.ico

Did you check if the data is correct in vector database? During chunking and vectorization time sometime Bedrock behave weird that could lead to incorrect output also.

StrictMode.VmPolicy.Builder builder = new StrictMode.VmPolicy.Builder();

StrictMode.setVmPolicy(builder.build());

You can't modify a non-managed, already named DynamoDB table. You can't even modify or rename a managed but explicitly-named DynamoDB table. So what you would have to do is that if the table exists, back it up to S3, delete the table, then create a new table with the same name but managed by CDK, and import your data from S3 to the now-CDK-managed table. Of course you would have to stop other apps accessing the table while deploying this to avoid errors.

n the dry and dusty backlands of Brazil, a humble clay house stands alone. Inside, a teenage girl lies on her bed, headphones on, lost in the sounds of her favorite K-pop group. Her small room is filled with posters of idols, a world away from her harsh surroundings.

Suddenly, her father barges in, rough and commanding.

“You better go clear the weeds,” he growls.

Annoyed and heartbroken, she gets up, grabbing the wrong tool, heading outside to “hoe the cat.” As she swings the tool, she mutters under her breath:

“Nobody loves me in this house… One day, I’m going to leave this place.”

After finishing her task, she walks back inside and collapses into bed. The next morning, something shifts inside her. She wakes up with a decision in her heart: she will go to New York.

She packs her few belongings onto a wooden cart, attaches it to an old donkey, and begins walking—days passing under the scorching sun. As she leaves, her father yells after her:

“Vai-te embora, maracaniça!”

Without turning back, she waves and shouts:

“Tchau, tchau, papai! Nunca mais irei voltar aqui!”

And for a while, that was true.

But one day, she does return. And nothing has changed.

“Go clear the weeds again,” her father barks.

Something breaks inside her. In silence, she walks away to the nearby river. There, she disappears beneath the water, ending the story that no one ever bothered to hear.

Can I read streams like this in a stored procedure?

Answer from ChatGPT: When samples are "strictly linearly separable", the (unregularized) cost function of logistic regression does not actually have a "finite global minimum".

If the data is linearly separable, then ||w|| (the norm of the parameter) can be continuously enlarged so that each sample is judged more and more "confidently" --- w.dot(X[i]) + b -> +Infinity for positive samples, w.dot(X[i]) + b -> -Infinity for negative samples, so that each sigmoid( w.dot(X[i]) + b ) approaches 1 (positive sample) or 0 (negative sample), and the cross entropy loss of each sample approaches 0.

So I customized two sets of data, they are all points on 2D plane, and the one is linear separable, the other is not linear separable. It turns out that the cost function of linear inseparable one indeed converges, while the separable one keeps descending and descending.

I found the error coming from Adobe extensions to Chrome. Removed the adobe extension and the error stopped!

I have faced this issue myself. Disabling spellchecker and other features didn't help. I was inserting a string of 70 000 rows of 17 symbols(vin numbers). After pasting a text of this size Chrome froze for roughly 20 seconds. Firefox had almost no issues with such amount of text.

This topic was hanging for a while, but I decided to resurrect it. I have recently built a library that solves a purpose of simple HTML to PDF conversion(under MIT license):
https://github.com/marisvigulis/iPDFGen

It has a pixel perfect HTML to PDF conversion and supports a several providers that performs a PDF creation.

Btw, I have a plan to resurrect a iTextSharp fork that was under LGPL license(the one that is free for commercial use).

JSON Pointer is weak for this very reason. JSON Path is much better for this use case.

JSON PATCH unfortunately leverages JSON Pointer, and that too must be updated to be able to use JSON PATH instead. I am sure the authors will have heartburn. But to preprocess the data and then determine the index value to construct the JSON Pointer statement renders it obsolete.

I want to prevent default for some elements, so i add css class .prevent-default to them:

$(".prevent-default").on("click", function (event) {
    event.preventDefault();
});

There are too many similar posts on this topic, but I thought this might be useful to someone.

I confirm. I uninstalled update KB5058499 and it's working again.

I had the same issue, and what solved this for me was installing the AWS EKS Pod Identity Agent as an Add-on in the cluster.

I basically followed this guide, and after installing it, the VPC CNI pod immediately started working: https://docs.aws.amazon.com/eks/latest/userguide/pod-id-agent-setup.html

Is there any chance you had this add-on in the pre-production cluster, but not in the production one?

It because your Window 11 security is blocking it.

Go to contolled folder access.

Click on Add allowed apps, then check recent blocked apps

You will see two apps here one is fat32format.exe and one more is rpi-imager.exe. Add/Allow both.

Now once it's added you can go and flash the image. It's should work smoothly.

Noting this down in case anyone runs into anything similar - I found the issue was related to an ErrorBoundary call on the POST request that resulted in the component being unmounted and not refetched again

I used the below syntax, and it seems to be working.

upper**(b.name1)** as EndUserName1**,**

I was asking this question to get a better idea of what the element values were, and using the getPageSource() method let me see what all the possible attributes are.

https://github.com/babashka/sci would be another option in this space

In your code, you make a requests.post() insted of requests.get()

1. Have you tried using any browser extensions or tools to improve how exceptions are displayed?

2. Does anyone know of plugins or upcoming features that could bring better IntelliJ integration for Kotlin Compose Web debugging?

This is due to strict mode in the development environment of React. When built for production, the API will only be called once.

You can simply select the same column twice with different aliases like this:

SELECT    client_code AS cl_code,   client_code AS acc_num FROM tb_accounts; 

This will return two columns, both populated with the values from client_code, but labeled differently as cl_code and acc_num.

MediaPicker the same way as FilePicker FullPath is returning a rather virtual path than the real path. Let me illustrate:

I have this file on my Android Emulator Downloads folder:

By using adb shell on my Android Adb Command Prompt I can see the physical path to my file is /storage/emulated/0/Download/ntk.p12.

Now let's try opening this file with FilePicker:

So the same file is showing in different paths.

But, what happens if I manually try to open the physical path I obtained via adb shell?

What about trying now with the FullPath obtained from FilePicker?

Or you can follow the documentation and just use OpenReadAsync:

I agree the documentation could be explained better. What they are trying to say is that you should be careful on your cross-platform solutions since FullPath has different behaviours.

Upon checking, your metadata.jsonl file format seems to be correct when validated using a JSON validator. Based on the error messages that you’ve received, it looks like it has something to do with the uri. In your .json file, you can try removing the documents with ID's 2-4 temporarily to help isolate the issue within your URIs. If there's no issue with id 1, then you can try id 2 next and so on.

Also, as mentioned in this documentation the maximum file size for Cloud Storage URIs is 2.5 MB for text-based formats, 200 MB for other formats which you can also check on your end.

As mentioned by @JaMiT, codespell just wanted the word in all-lowercase.

Thank you very much for the help, mate. You're a genius!

Afaik it's still not possible, recently did some research during implementation of "chat reactions". One of the ideas was to leverage preview providers for this and I couldn't make it work... Also I found this repo showcasing a private API for reactions, but at the end of the day had to implement context menu from scratch 🌚

I have the same problem. Did you find solution?

To build only the changed module and its dependents in a Maven multi-module project, use:

mvn -pl B -am clean install 

• -pl B builds module B.

• -am also builds modules that depend on B (D, E).

• Modules like A and C that didn’t change are skipped.

To skip tests in unchanged modules, add:

-DskipTests 

If you want to automate detecting changed modules, use Git diff in your CI to build only those modules dynamically.

This minimizes build time by compiling and testing only affected modules.

checkout : https://rkoots.github.io/styleguide/

I think it has something to do with the following settings:

1. Check your Thread Group setup, Same user on each iteration should be ticked

   

2. Check httpclient.reset_state_on_thread_group_iteration property value, if you want to re-use the same connection - you need to set it to false either in user.properties file or via -J command-line argument.

   More information:

   • SSL Configuration
   • Apache JMeter Properties Customization Guide

3. Check your operating system settings with regards to recycling/reusing TCP connections in TIME_WAIT state

I reduced the size by checking the checkbox "Combine inheritance margin with indicator margin".

Hey i am facing the same issue , did you find anything .If yes can you help me.

You're right to be confused — in C, a string like "0000000" is not empty, even though it looks like zeros.

🔹 Explanation:

If arr contains "0000000" (seven characters + a null terminator), then:

if (arr[0])

will return true, because arr[0] is '0' (ASCII 48), which is a non-zero value.

Only if arr[0] == '\0' (null terminator), the string is considered empty.

✅ To check if the string has any content:

if (arr[0] != '\0') {     // string is not empty }

But if you want to check for a string of only '0' characters, you need to loop or use:

if (strcmp(arr, "0000000") == 0) {     // arr contains exactly "0000000" }

Let me know if you're checking for zero characters or numeric zero.

checkout :https://rkoots.github.io/styleguide/

I ran into the same issue when integrating Clerk with Vite and react-router-dom under a custom base path like /admin. The key here is to ensure that all parts of the app — Vite, React Router, and Clerk — have consistent understanding of the base path, and avoid double-prefixing in redirects.

✅ The Root Cause

When using Google OAuth, Clerk redirects using the public base URL configured in your Clerk dashboard or environment. If VITE_CLERK_SIGN_IN_REDIRECT_URL is set to dashboard, Clerk assumes the full redirect should be /admin/dashboard. But if you already set your Vite base as /admin, and then also include /admin in the Clerk redirect, you’ll end up with /admin/admin/dashboard — which causes the 404.

🛠️ Working Solution

Update your .env with the relative path only — Clerk already knows to prepend the base:

env

VITE_CLERK_SIGN_IN_REDIRECT_URL=/dashboard VITE_CLERK_SIGN_IN_FALLBACK_REDIRECT_URL=/dashboard

Notice the leading /, but do not include /admin here.

📍Also Important

• In your vite.config.ts, make sure you have:

  ts
  

  base: '/admin/'

• And in your <BrowserRouter>:

  tsx
  

  <BrowserRouter basename="/admin">

• Finally, ensure your Clerk dashboard → Frontend Settings → Sign-In URL is consistent with /admin/sign-in (or whatever your sign-in route is under the base path).

Changing acroviewA25 to AcroviewA25 seemed to work for me on Win11, Adobe AcrobatReader (64-bit)

For the benefit of future readers:-

@Toto's answer may replace other numbers as well, so I am posting another PythonScript here to match <sup>, followed by a [ bracket and then the number sought as per the question. You can run a python script within the PythonScript plugin of Notepad++ for that.

If it is not yet installed, install it via Plugins Admin. To install the PythonScript plugin in Notepad++, navigate to Plugins > Plugins Admin. Check the box next to PythonScript and click Install. Notepad++ will restart, and after that, you can find the plugin in the Plugins menu.

• Create a script (Plugins >> PythonScript >> New Script)

• Copy this code and save the file (for example increment.py):

# encoding=utf-8
from Npp import *

def custom_replace_func(m):
    new_text = "<sup>\[{}".format(custom_replace_func.counter)
    custom_replace_func.counter += custom_replace_func.increment
    return new_text

custom_replace_func.counter = 1  # starting value
custom_replace_func.increment = 1
editor.rereplace(r'<sup>\[(\d+)', custom_replace_func)

• Open the file you want to modify

• Run the script (Plugins >> PythonScript >> Scripts >> increment)

• Then, in the result obtained after the above, put the cursor before the first <sup> and hit Ctrl+H and in the pop-up, type <sup>.*?\[ in the "Find" field and <sup>[ in the "Replace with" field and hit “Replace all” to get rid of the \ after every <sup>

Sorry for a necroposting, but where is this interface in a VSCode? Looks like MS shifted Logic App development from VS to VSC...

just update your compose version to 1.8.0+

My USB port doesn't work well so I use WiFi ADB app on my Android phone and I do this when this error occur:

• turn off the WiFi ADB
• disconnect from the WiFi
• turn off the Mobile Hotspot
• unplug Ethernet cable
• plug Ethernet cable
• turn on the Mobile Hotspot
• connect to the WiFi
• turn on the WiFi ADB
• turn off the WiFi ADB
• turn on the WiFi ADB (yes second time)
• run command: adb kill-server
• run command: adb connect 192.168.137.2

I encountered this issue. The issue clearly outlines that there is some missing resources. To be able to use the TextField composable and others, you have to add into your 'dependencies' section some material3 related implementations as documented officially here.