ia: Benvenite! In mi blog io scribe in interlingua, italiano e anglese.

it: Benvenuti! Nel mio blog scrivo in interlingua, italiano e inglese.

en: Welcome! In my blog I write in Interlingua, Italian and English.

Growing up, or leaving Facebook

I've finally decided to delete my Facebook account. I've been pondering this decision for several years now; my previous choice on the matter was to just use it as little as possible, in write-only mode, just to spread political propaganda and generally make it a horrible place for everybody else too. My reasoning was the following: if I had a huge following, then I'd definitely leave and make a big noise (and impact); but since I'm mostly connected with real acquaintances only, my only way to make a (albeit minimal) impact is to stay there and “save” my friends by convincing them of my opinions and/or by being so annoying that they'd decide to leave first.

There's life out there, too

However, the more time passes, the more I start doubting this position. What I really want is for people to leave Facebook; but as long as I'm there, as everybody else is there, no matter how grumpy or destructive we might be, we are still part of Facebook's business and not challenging the status quo in any meaningful way. We are not going to convince anyone to migrate to alternative platforms (or just leave all social media for good) while we continue to maintain a presence in there. Also, I know too well that it's nearly impossible to change someone's mind via social media, and my plan to make the place horrible for everybody is not going to work either: if people get annoyed, they'll just mute me and that's it.

On the other hand, if I leave Facebook for good, there's still a high chance that nobody will care, but at least I've quit. I'll no longer be a resource for Facebook. On the grand scheme, this is very little, a tiny drop in the ocean, but I believe it's the best I can do. Will I suffer it? Personally, not at all: I've not been consuming Facebook for a couple of years by now. But certainly, it's possible that the number of visits to my blog or YouTube channel (we'll talk about this other monster some other time) will decrease; not many people know how to use browser bookmarks (let alone an RSS feed) nowadays but, honestly, there's not much I can do about it.

The tech giants of the internet are certainly making our lives easier: it has become easier to stay in touch, to find information, to consolidate a mainstream opinion. It's so convenient, that it is making us not only lazier, but also dumber, to the point that we'd feel lost and powerless without their services. We are not talking only about a psychological addiction anymore: they are becoming a practical necessity. This, in turn, makes us weak, easy to manipulate and, ultimately, slaves.

I believe in the internet as the inter net: a distributed, interconnected place, with no owner. My hosting provider decides to censor me and shuts down my site? Fine, I open a new one in another hosting. If Facebook kicks you out, you immediately lose access to your friends, the groups you followed, the marketplace, local events, your daily news, your identity, your history.

And me, I want to live in a world where I'm the owner of my life.

How to be an awesome member of the Free Software community

No, this is not a tutorial, unfortunately. The main reason being that, while I strive to do my best, I don't consider myself to be an excellent member of the Free Software community, let alone be able to teach others about it. But this morning I got an email from the FSF about a campaign for St. Valentine's day, which reminded me of something I've been planning to do since a long time ago, but never got to it.

Love

I want to publicly send huge thanks to Robin Mills from the Exiv2 project, and not only because I've been fruitfully using his work in three (!) projects of mine (PhotoTeleport, Mappero Geotagger and Imaginario), but also, and especially, for being an extremely pleasant interlocutor. On the web, yes. Whereas most people tend to be more thorny and touchy in their interactions over the internet, Robin has always been friendly and coversational, trying to form a bond with some complete foreigner who just happened to report a bug on Exiv2.

Just adding some bits of information about one's personal events (such as travels) can make an enormous difference on how one's attitude is perceived. Mentioning that you visited the place that is familiar to the bug reporter almost makes one forget of being sitting in front of a computer on the internet, because your mind flies to that place. Considering that even on this personal blog of mine I'm kind of reticent about speaking of my private life, I cannot help appreciating the friendly attitude that Robin reserves for people writing on a bug tracker.

You are a wonderful netizen, Robin. A happy Valentine Day to you and your family. Thank you.

Invoking a C++ function from QML, asynchronously

In the Imaginario code I had written a C++ method to find all files present in a directory tree (for some reason this code is always encoding file paths as QUrl, but feel free to ignore that):

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
QList<QUrl> Utils::findFiles(const QUrl &dirUrl, bool recursive) const
{
    QList<QUrl> files;

    QDir dir(dirUrl.toLocalFile());
    if (Q_UNLIKELY(!dir.exists())) return files;

    auto list = dir.entryInfoList(QDir::NoDotAndDotDot |
                                  QDir::Dirs | QDir::Files,
                                  QDir::Name);
    for (const QFileInfo &info: list) {
        if (info.isDir()) {
            if (recursive) {
                files += findFiles(QUrl::fromLocalFile(info.filePath()), true);
            }
        } else {
            files.append(QUrl::fromLocalFile(info.filePath()));
        }
    }

    return files;
}

The Utils mentioned in this snipped is a QObject-derived class which is registered to QML as

1
2
3
4
5
6
7
8
static QObject *utilsProvider(QQmlEngine *, QJSEngine *)                                               
{                                                                                                      
    return new Utils;                                                                                  
}

...
// In some other part of the code, before entering the main loop:
qmlRegisterSingletonType<Utils>("Imaginario", 1, 0, "Utils", utilsProvider);

This allows me to call the C++ findFiles() method from QML, like this:

1
2
3
4
import Imaginario 1.0

...
    onClicked: importer.addFiles(Utils.findFiles(folder, true))

So far, so good. However, I couldn't help noticing that when the selected folder contains a large number of files, the whole UI freezes until the findUtils() method has returned. So, how can I invoke my C++ method without blocking the UI?

QML offers a WorkerScript element which seems to do exactly what we need, but unfortunately the possibility of invoking C++ code is only there since Qt 5.12 (before that version, worker script could not use import statements), and in any case the requirement to store the script into a separate Javascript file makes the code less readable.

A better option, in my opinion, is to write a C++ method that performs the lengthy task in a thread and invokes a Javascript callback once the job execution is completed. This sounds pretty complex at first, but fortunately Qt's nice APIs make this an almost trivial task. Without modifying our previous findFiles() method, we write a second one which will execute it in a thread, using QtConcurrent::run():

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
void Utils::findFiles(const QUrl &dirUrl,
                      bool recursive,
                      const QJSValue &callback) const
{
    auto *watcher = new QFutureWatcher<QList<QUrl>>(this);
    QObject::connect(watcher, &QFutureWatcher<QList<QUrl>>::finished,
                     this, [this,watcher,callback]() {
        QList<QUrl> files = watcher->result();
        QJSValue cbCopy(callback); // needed as callback is captured as const
        QJSEngine *engine = qjsEngine(this);
        cbCopy.call(QJSValueList { engine->toScriptValue(files) });
        watcher->deleteLater();
    }); 
    watcher->setFuture(QtConcurrent::run(this, &Utils::findFiles, 
                                         dirUrl, recursive));
}

You might be surprised, but this is all what is needed in the C++ side. The changes to the QML side are equally trivial:

1
2
3
4
5
6
7
8
import Imaginario 1.0

...
    onClicked: {
        Utils.findFiles(folder, true, function(files) {
            importer.addFiles(files)
        })
    }

That's it! Line 6 in this last snippet will be invoked once the findFiles() method has completed its execution, and you'll be glad to see that the UI will be responsive throughout the duration of the operation.

QtConcurrent alternatives

QtConcurrent::run() is very simple to use, but it's not without its shortcomings, the biggest of which is that it doesn't support cancelling. This is something I can live with in this particular case, but, if you can't, don't despair: there are other options. The second simplest one is probably the static QThread::create() method, which behaves similarly to QtConcurrent::run() but returns a QThread object which can be requested to terminate, for example with QThread::requestInterruption(). This method exists only since Qt 5.10, and that's why I didn't list it as my first choice; but if you are using such a recent version of Qt, then it's probably the best option, because one doesn't even need to subclass QThread (and there's no need to depend on QtConcurrent). Of course, subclassing QThread is also an option, but this involves a little more typing.

Beware of threading issues

The most tempting solution (and I confess, my first approach to the problem was exactly this) is to invoke the Javascript callback directly from the thread:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
void Utils::findFiles(const QUrl &dirUrl,
                      bool recursive,
                      const QJSValue &callback) const
{
    QtConcurrent::run([=]() {
        const QList<QUrl> files = findFiles(dirUrl, recursive);
        QJSValue cbCopy(callback);
        QJSEngine *engine = qjsEngine(this);
        cbCopy.call(QJSValueList { engine->toScriptValue(files) });
    });
}

However, this will work 95% of the times only, because QJSValue::call() is not thread-safe, and therefor your application will be victim of random crashes. So, we need to write a couple of lines more and invoke the callback from the main thread, like I'm doing before.