Hi,

I am using the qt only version of the marble widget as part of a larger application and every time I start marble I see these 2 warning lines printed out:

QInotifyFileSystemWatcherEngine::addPaths: inotify_add_watch failed: No such file or directory
QFileSystemWatcher: failed to add paths: /home/test/.local/share/marble/maps/earth/placemarks/

I greped through the code and the only occurance of the string 'maps/earth/placemarks' is in the file src/plugins/runner/local-osm-search/LocalOsmSearchPlugin.cpp. I was therefore able to prevent these warnings from showing by disabling the local-osm-search plugin. However, I would prefer not to need to disable any plugins when we ship our code and was therefore wondering if anyone knows if there is another way to prevent these warnings, or if perhaps there is some bug here.

Thanks!

ps. I'm using the master branch on linux

The same warnings will pop up up if you install a new map theme while Marble is running, which can be simulated by

Code: Select all

touch ~/.local/share/marble/maps/earth

Last time I looked into it I found that it was a problem in Qt, in combination with inotify, and expected to be fixed only in Qt5.

Earthwings wrote:The same warnings will pop up up if you install a new map theme while Marble is running, which can be simulated by

Code: Select all

touch ~/.local/share/marble/maps/earth

Last time I looked into it I found that it was a problem in Qt, in combination with inotify, and expected to be fixed only in Qt5.

Took another look, turns out that QFileSystemWatcher complains as well if you add a resource for watching more than once. This should be fixed by http://commits.kde.org/marble/865e5d7bb ... 746fa9bdc8 now. That works for me for the map themes, though I couldn't reproduce the warning you describe from LocalOsmSearchPlugin. The patch adds a check there as well though, can you test and reply back here whether it works?

Earthwings wrote:

Earthwings wrote:The same warnings will pop up up if you install a new map theme while Marble is running, which can be simulated by

Code: Select all

touch ~/.local/share/marble/maps/earth

Last time I looked into it I found that it was a problem in Qt, in combination with inotify, and expected to be fixed only in Qt5.

Took another look, turns out that QFileSystemWatcher complains as well if you add a resource for watching more than once. This should be fixed by http://commits.kde.org/marble/865e5d7bb ... 746fa9bdc8 now. That works for me for the map themes, though I couldn't reproduce the warning you describe from LocalOsmSearchPlugin. The patch adds a check there as well though, can you test and reply back here whether it works?

Thank you so much for looking into it, but alas the warning still appears after updating, unless I create the folder ~/.local/share/marble/maps/earth/placemarks/ before running Marble. Is it possible that Marble should perhaps first create any folders it watches to avoid such warnings?

hod wrote:

Earthwings wrote:

Earthwings wrote:The same warnings will pop up up if you install a new map theme while Marble is running, which can be simulated by

Code: Select all

touch ~/.local/share/marble/maps/earth

Last time I looked into it I found that it was a problem in Qt, in combination with inotify, and expected to be fixed only in Qt5.

Took another look, turns out that QFileSystemWatcher complains as well if you add a resource for watching more than once. This should be fixed by http://commits.kde.org/marble/865e5d7bb ... 746fa9bdc8 now. That works for me for the map themes, though I couldn't reproduce the warning you describe from LocalOsmSearchPlugin. The patch adds a check there as well though, can you test and reply back here whether it works?

Thank you so much for looking into it, but alas the warning still appears after updating, unless I create the folder ~/.local/share/marble/maps/earth/placemarks/ before running Marble. Is it possible that Marble should perhaps first create any folders it watches to avoid such warnings?

Right, I didn't read the error message carefully enough. The non-existing directory explains the behavior and I could reproduce it now. Should be really fixed in http://commits.kde.org/marble/0b9a7cc54 ... 7aaa2704a6. Thanks for testing and reporting.