question about AbstractFileManagerPluginPrivate::eventuallyReadFolder()

Tomaz Canabrava tcanabrava at kde.org
Fri Oct 6 12:51:03 UTC 2017


On Fri, Oct 6, 2017 at 2:48 PM, René J.V. Bertin <rjvbertin at gmail.com>
wrote:

> On Friday October 06 2017 14:07:09 Milian Wolff wrote:
>
> >Are you sure the signal is connected properly?
>
> All I did was specify the connection type explicitly. It's possible that
> means the signal was no longer connected properly, but as I understand the
> documentation that should not be the case. The watchDir signal I added to
> FileManagerListJob myself uses this type and works as expected. I haven't
> checked if removing the explicit connection type makes any difference there.
>
> >No runtime warnings, e.g. about
> >missing meta type registrations?
>
> No.
>
> >Is it recieved properly?
>
> In essence, no. Either it's not received at all, or not at the right time,
> or with the wrong context.
>
> >All in all this is standard debugging procedure which you'll have to do
> yourself.
>
> Standard but advanced and probably very annoying debugging...
>
> Let me ask the complementary question: why would it be safe here to
> connect a signal sent from a background thread to a slot in another QObject
> that will be executed on the main thread?
>

It's not safe unless you use QueuedConnection, basically making it a FIFO
call.


> R.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20171006/d99f7ad2/attachment.html>


More information about the KDevelop-devel mailing list