<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<p>Here, I copied the most interesting parts from the bug:<br>
</p>
<p>Basically, QNetworkAccessManager is sometimes stuck in
NotAccessible state after switching from mobile data to wifi.<br>
</p>
Among other things, it makes QNetworkAccessManager::get () to fail
with QDisabledNetworkReply which breaks XmlListModel as some people
have noted.<br>
<div class="moz-cite-prefix"><br>
Qt 5.2 had a similar problem but this helped:<br>
<br>
<a class="moz-txt-link-freetext" href="https://git.merproject.org/mer-core/qtbase/merge_requests/3">https://git.merproject.org/mer-core/qtbase/merge_requests/3</a><br>
<br>
Unfortunately, Qt connman bearer plugin has significantly changed
in Qt 5.6 and this trick stopped working. <br>
<br>
When the switch (between mobile and wifi) is happening, the action
occurs in parallel on two threads, Qt bearer thread and the main
thread, the signals are queued, although the main thread is
fetching the current state directly from the objects being
manipulated by the bearer thread. My theory was that
configurationChanged, onlineStateChanged and
networkSessionStateChanged (queued) signals from the past were
getting mixed with the current state (which was checked directly)
in such a way that QNetworkAccessManagerPrivate online state
remained false even though connman service was getting
successfully connected. So I came up with this:<br>
<br>
<a class="moz-txt-link-freetext" href="https://git.merproject.org/mer-core/qtbase/merge_requests/18">https://git.merproject.org/mer-core/qtbase/merge_requests/18</a><br>
<br>
but that didn't quite fix it. It does eliminate one scenario
(described in the merge request) but there must be something else
there. The whole connman bearer plugin thing looks generally prone
to race conditions, I would've done it differently but we have
what we have.<br>
<br>
Cheers,<br>
-Slava<br>
<br>
<br>
</div>
<blockquote
cite="mid:CAHCc5YXry1NwFdP6HrZDaSE1zU8_8re93mTR35vLEuVRQmnV_g@mail.gmail.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<div dir="ltr">
<div>
<div>Thanks for the quick reply;<br>
</div>
I was actually hoping to collect some more background intel on
this issue.<br>
</div>
<br>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Sun, Apr 30, 2017 at 10:56 PM, Slava
Monich <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:slava.monich@jolla.com" target="_blank">slava.monich@jolla.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"> Unfortunately, no,
it's an internal bug. If you would like to be notified via
email about a potential fix, you could watch qtbase pull
requests - those are quite rare, it shouldn't be too much
of a spam.<br>
<br>
Cheers,<br>
-Slava
<div>
<div class="h5"><br>
<br>
<div class="m_1251028547860576262moz-cite-prefix"><br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>
<div>Hi Slava;<br>
</div>
is there a MER bug for this?<br>
<br>
</div>
tortoisedoc<br>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Sun, Apr 23, 2017 at
11:33 AM, Slava Monich <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:slava.monich@jolla.com"
target="_blank">slava.monich@jolla.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">Network status
detection in Qt 5.6 is unreliable when the
system is switching between wifi and mobile
data. That results in QNetworkAccessManager
sometimes thinking that network access is
disabled when in fact the device is online. At
some point I attempted to fix it (<a
moz-do-not-send="true"
href="https://git.merproject.org/mer-core/qtbase/commit/8122f32a"
rel="noreferrer" target="_blank">https://git.merproject.org/me<wbr>r-core/qtbase/commit/8122f32a</a>)
however that only eliminated one particular
scenario of failure and reduced the
probability. But it's still happening. As I
understood it's a random thing (race
condition) triggered by two connman services
changing their status more or less
simultaneously.<br>
<br>
Jolla's Weather app also uses XmlListModel and
suffers from the same issue.<br>
<br>
Nobody at Jolla is actively working on it at
the moment, our resources are limited these
days. If somebody has an idea how to fix it,
feel free to submit a pull request against
mer-5.6 branch (<a moz-do-not-send="true"
href="https://git.merproject.org/mer-core/qtbase/commits/mer-5.6"
rel="noreferrer" target="_blank">https://git.merproject.org/me<wbr>r-core/qtbase/commits/mer-5.6</a>)<wbr>.<br>
<br>
Best regards,<br>
-Slava<span class="m_1251028547860576262im
m_1251028547860576262HOEnZb"><br>
<br>
<br>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;border-left:1px
#ccc solid;padding-left:1ex"> Hi,<br>
<br>
The same happens in my Rush hour app
available in OpenRepos. My only connection
to network is via QtQuick.XmlListModel
2.0. I am using Jolla C and the latest
Iijoki 2.1.0.11<br>
<br>
<a moz-do-not-send="true"
href="https://github.com/Rikujolla/trafficviewer/issues/2"
rel="noreferrer" target="_blank">https://github.com/Rikujolla/t<wbr>rafficviewer/issues/2</a><br>
<br>
Best regards, Riku<br>
<br>
Sebastian Wolf kirjoitti sunnuntai 23.
huhtikuuta 2017:<br>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;border-left:1px
#ccc solid;padding-left:1ex"> Hi
colleagues,<br>
some users of my application Piepmatz
are reporting network issues when their<br>
device network connection changes from
WiFi to Cellular and vice versa. I<br>
couldn't reproduce it so far with my FP2
on 2.0.5. As I'm only using standard Qt<br>
Network classes to access the internet
and no own network detection and<br>
especially that not only my application
is affected, it could possibly be<br>
something in 2.1 itself. For some
details see GitHub issue<br>
<a moz-do-not-send="true"
href="https://github.com/Wunderfitz/harbour-piepmatz/issues/11"
rel="noreferrer" target="_blank">https://github.com/Wunderfitz/<wbr>harbour-piepmatz/issues/11</a><br>
Didn't find anything with a quick
search on <a moz-do-not-send="true"
href="http://together.jolla.com"
rel="noreferrer" target="_blank">together.jolla.com</a>,
so does somebody<br>
of you know something about such an
issue?<br>
Thanks for any hints and best regards,
Sebastia<br>
</blockquote>
</blockquote>
<br>
</span>
<div class="m_1251028547860576262HOEnZb">
<div class="m_1251028547860576262h5">
______________________________<wbr>_________________<br>
SailfishOS.org Devel mailing list<br>
To unsubscribe, please send a mail to <a
moz-do-not-send="true"
href="mailto:devel-unsubscribe@lists.sailfishos.org"
target="_blank">devel-unsubscribe@lists.sailfi<wbr>shos.org</a></div>
</div>
</blockquote>
</div>
<br>
</div>
<br>
<fieldset
class="m_1251028547860576262mimeAttachmentHeader"></fieldset>
<br>
<pre>______________________________<wbr>_________________
SailfishOS.org Devel mailing list
To unsubscribe, please send a mail to <a moz-do-not-send="true" class="m_1251028547860576262moz-txt-link-abbreviated" href="mailto:devel-unsubscribe@lists.sailfishos.org" target="_blank">devel-unsubscribe@lists.<wbr>sailfishos.org</a></pre>
</blockquote>
</div></div></div>
______________________________<wbr>_________________
SailfishOS.org Devel mailing list
To unsubscribe, please send a mail to <a moz-do-not-send="true" href="mailto:devel-unsubscribe@lists.sailfishos.org">devel-unsubscribe@lists.<wbr>sailfishos.org</a>
</blockquote></div>
</div>
<fieldset class="mimeAttachmentHeader"></fieldset>
<pre wrap="">_______________________________________________
SailfishOS.org Devel mailing list
To unsubscribe, please send a mail to <a class="moz-txt-link-abbreviated" href="mailto:devel-unsubscribe@lists.sailfishos.org">devel-unsubscribe@lists.sailfishos.org</a></pre>
</blockquote>
</body></html>