General and Desktop

@mpergand thank you

6.4.2 open source, mac os

@masa4 said in Deleted posts: This is a sad situation actually, why are you being a toxic guy at this point? Where am I being toxic?! I actually felt sorry to hear that. But these situations can be avoided very easily, when you describe your problem and post an example to reproduce it. Copy/pasting and posting code online, which is part of commercial software and is not owned by you, is never a good idea. Even if you're asking for help. It is what it is. If someone decided that he should not have shared that information, it was too much. Like OP wrote, he thought about it, and felt it wasn't that sensitive. I or "we" can not decided how sensitive it really was. I dont think so. You keep being a toxic while living in your imagination world. lol My "imagination" is, that if you work in a company and you share parts of hidden code and your boss or your client find it out, I guess, they wont like it. But that is, maybe, only my imagination, in my world :)

@mrjj Often the touch controller comes with a settings app that allows you to control if long press means right click and such things But actually, as this tutorial shows and i try it, even windows has it but it's very restrictive. I'm gonna try something like this. Your answer make me thinks maybe to install a macro. Could it be a solution?

@SPlatten I have PMed you.....

I found the issue, of course I had a digit wrong on the port... The connection seems to succeed now, I get a different error elsewhere but for the purpose of this thread it can be considered solved.

QList<T> requires that type T is an assignable data type. QTextDocument is a QObject and you cannot copy QObjects (by design). Store a pointer instead QList<QTextDocument *> list

Have a look at this. GCC-9 should work with the latest Qt 5.15-lts on Ubuntu 18.04 and 20.04. All other Linuxes on 5.15 prefer GCC-5, or GCC-10 and above. Qt 5.15.2, however, is not the latest LTS release. Isn't Qt 6 an option?

Although not really solving the issued, I created a workaround using a stored procedure at the server. DROP PROCEDURE IF EXISTS testdb.CreaUser; DELIMITER // CREATE PROCEDURE testdb.CreaUser(IN username TEXT, IN passwd TEXT) BEGIN PREPARE stmt FROM CONCAT("CREATE USER ", QUOTE(username), "@'%' IDENTIFIED BY ", QUOTE(passwd)); EXECUTE stmt ; END; // DELIMITER ; Having the procedure in place I can call it from Qt with bound values: QSqlQuery q(db); QString qu="CALL CreaUser ( :username , :passwd );"; if(q.prepare(qu)) qDebug()<<"query prepared"; else { qDebug()<<"q.prepare failed "<<q.lastError().text(); return; } q.bindValue(":username", "foo"); q.bindValue(":passwd", "; DROP TABLE x; --'"); if(q.exec()) qDebug()<<"query executed"; else { qDebug()<<"q.exec failed "<<q.lastError().text()<<q.lastQuery(); return; } db.close(); and get the user created without any loopholes for SQL injection.

@Bonnie Thank Sir, indeed it was caused by QT += testlib, I'm glad the problem is solved

@w-tkm said in How to show black-and-white image from binary data?: I was able to output from binary data. But, If the image size is set to 1000*1000, the output will be rectangular instead of square. Source pixels or screen pixels may not be square, or may differ in aspect ratio. Hard to say what applies to your case. There is no original format because the binary data is automatically generated by the program. So the source just sends you a million bits in random arrangement? Of course not... it is sending you data in some arrangement. Chances are this is left to right, top to bottom, 8 pixels per byte and yu can use something like this to make the QImage much faster: #include <QGuiApplication> #include <QByteArray> #include <QImage> #include <QDebug> int main(int argc, char *argv[]) { QGuiApplication a(argc, argv); // Build a test raw image. QByteArray data; data.reserve(1024*128); for (int row = 0; row < 1024; ++row) { // 1024 rows for (int col = 0; col < 128; ++col) { // 128 bytes = 1024 pixels data.append(0xaa); } } // Convert to QImage // Be sure to read about the lifetime requirements on the source buffer. QImage result(reinterpret_cast<uchar*>(data.data()), 1024, 1024, 128, QImage::Format_Mono); result.setColorCount(2); result.setColor(0, QColor("black").rgb()); result.setColor(1, QColor("white").rgb()); qDebug() << result; result.save("/tmp/test.png"); return 0; }

@Christian-Ehrlicher Oh, I don't know why I'm not using that. haha. I'll delete this post

Thanks for the answer, I'll try it and say what happened.

By the way easiest solution i found is: adding #undef signals before glib includes, and after it #define signals again.

@jsulm I have tried 3 times and it failed due to power outage here in Ukraine because of russian aggression :( Anyway, I overclocked the AMD FX8350 processor from 4 GHz to 4.60 GHz. Now, it compiles a bit faster. Thank you.

Thank you :)

@Sabrac there is a return value from blockSignals(true). Check it out. The return value is the previous value of signalsBlocked(). https://doc.qt.io/qt-5.15/qobject.html#blockSignals There is another way to avoid any unnecessary changes caused by a specific signal. disconnect( table, signal, this, slot); // ... Some other logics table->setItem(0, 0, value0); table->setItem(0, 1, value1); table->setItem(0, 2, value2); table->setItem(0, 3, value3); // ... Some other logics connect( table, signal, this, slot); This may work better sometimes since you may not want to block all signals.

Darn, this is painnnn. Thanks!