mate-desktop / mate-user-share Goto Github PK

Orginal report at fedora bugzilla.
https://bugzilla.redhat.com/show_bug.cgi?id=1013385

see all details at fedora bugzilla.
In short, without setting a password to access the share folder, connecting to this folder fails.

no-manual-page-for-binary mate-file-share-properties
Each executable in standard binary directories should have a man page.

For some odd reason this package crashes the desktop when trying to access ~/Downloads folder or a symlink with the name Downloads that point to some directory.
The desktop or file manager to be more precise (caja) restarts but every window that was opened in caja does not reopen.
No other folder or symlink do that, or if I rename "Downloads" to "Download" or "downloads" the file manager doesn't crash upon opening the folder, it has to do something with the name "Downloads"...
$HOME/Downloads is not a bookmark BTW.
If I open ~/Downloads with thunar on mate-desktop nothing happens...
I even created a new user and tested it to be sure that is not related to some odd config...
~/.xsession-errors doesn't say anything...
The permissions are fine, I even created the folder manually and after with xdg-user-dirs and logout - login and still the same.
No matter how I open the "Downloads" folder/symlink to a directory, be it double click, context menu > Open in New Tab or Open in New Window...
I've managed to reproduce the bug on virtualbox with archlinux but I can't be sure if it's present on other distros (with the latest packages).
I first installed (on arch-virtualbox) the mate group "pacman -S mate" then saw that everything was fine and caja didn't crashed, then installed the extra packages with "pacman -S mate-extra" and caja crashed when opening ~/Downloads.
So I knew that some package in "mate-extra" was to blame so I removed them one by one...
After removing mate-user-share the desktop crashes once maybe because it was still loaded in memory but after that everything was ok, no more crashes.
Culprit found :D
Xmas morning and I'm writing novels ... I mean reporting bugs... haha.

At the moment, files via bluetooth recieving to the hard-coded folder ~/Downloads. It's will be good to take location of folder via xdg, because in various localizations it is called by the different names.

In short, mate-user-share doesn't work on port 80, which everybody would expected.
It doesn't use a any pre-defined port but a random port (from range 32768-65535).
This behaviour makes it impossible to handle by a firewall.
see https://bugzilla.redhat.com/show_bug.cgi?id=1018204
https://bugzilla.redhat.com/show_bug.cgi?id=1013385#c17
Same with old versions of gnome-user-share, see
https://bugzilla.redhat.com/show_bug.cgi?id=179187#c23
https://bugzilla.gnome.org/show_bug.cgi?id=336201#c5
It seems that gnome found a solution, but i didn't really understand how this work.
For me using a system without a firewall is a huge security risk nowadays,
which prevent me to bring in the package to rhel7.

Debian has raised the following bug:

• mate-user-share - https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=827231

This MATE component needs migrating to GtkApplication:

• https://developer.gnome.org/gtk3/3.20/gtk-migrating-unique-GtkApplication.html

Selecting only "Receive files over bluetooth" doesn't work.
It needs the first option (share files over network)

If i install/uninstall this new package i get an minor rpm scriptlet output error.
In /etc/xdg/autostart/mate-user-share.desktop are two entries with gconf instead of expected mateconf.

Changing gconf to mateconf in mate-user-share/data/desktop_mate_file_sharing.schemas.in fix the error.

I: mate-user-share: desktop-entry-lacks-keywords-entry usr/share/applications/mate-user-share-properties.desktop
N:
N: This .desktop file does either not contain a "Keywords" entry or it does
N: not contain any keywords not already present in the "Name" or
N: "GenericName" entries.
N:
N: .desktop files are organized in key/value pairs (similar to .ini files).
N: "Keywords" is the name of the entry/key in the .desktop file containing
N: keywords relevant for this .desktop file.
N:
N: The desktop-file-validate tool in the desktop-file-utils package is
N: useful for checking the syntax of desktop entries.
N:
N: Refer to
N: http://standards.freedesktop.org/desktop-entry-spec/latest/ar01s05.html,
N: http://bugs.debian.org/693918, and
N: https://wiki.gnome.org/Initiatives/GnomeGoals/DesktopFileKeywords for
N: details.
N:
N: Severity: wishlist, Certainty: certain
N:
N: Check: menu-format, Type: binary
N:

Expected behaviour

Intact *.desktop files in /etc/xdg/autostart/

Actual behaviour

OnlyShowIn= key gets appended to Categories= key and thus "disables" OnlyShowIn= field. Consequently, mate-user-share-*.desktop application get launched in every other desktop env installed on the system

Steps to reproduce the behaviour

Launch non-MATE desktop env, inspect /etc/xdg/autostart/mate-user-share*.desktop

MATE general version

1.24

Package version

1.24.0-1

Linux Distribution

Debian

Link to downstream report of your Distribution

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=952401

After last update from branch, bluetooth file receiving was broken. I haven't add any log files, because doesn't know where it are. I'll show it to you if you say where i can get it.

mate-bluetooth version is 1.4.0-1

Hi,

This bit here:
https://github.com/mate-desktop/mate-user-share/blob/master/src/user_share.c#L376
is very weird and won't make GCC very happy either:

user_share.c: In function 'main':
user_share.c:376:16: warning: type defaults to 'int' in declaration of 'x_fd' [-Wimplicit-int]
G_GNUC_UNUSED x_fd;
^
I don't know what the point of an unused local variable is, but an unused local variable without a type is certainly worse.

Best,
Michael

Like it's done in the caja-extensions repo.

@stefano-k: since the fix for GLib issue needs to make it to 1.8.x releases, please create 1.8 branch in this repo from 191f78c.
After that, #29 should be merged there. It's release critical for a new 1.8 tarball as agreed with @flexiondotorg.

Hi,

in the debian/ folder obtained from mate-desktop/debian-packages there was a patch file provided originally authored by Steve Zesch [1].

I have kept this patch and also spread it out to the Apache2.4 config file [2].

Please considering, applying that patch upstream (or tell me to drop the patch from the mate-user-share Debian package).

Thanks!
Mike

[1] https://github.com/mate-desktop/debian-packages/commits/master/mate-user-share/debian/patches/01_apache-config.patch

Arch Linux current (not testing). MATE from official repo.

:::/usr/lib/mate-applets/mate-user-share  

** (mate-user-share:19168): WARNING **: Couldn't request the name: Unable to lookup session information for process '19168'

(mate-user-share:19168): GLib-GIO-CRITICAL **: g_settings_get_value: assertion `G_IS_SETTINGS (settings)' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_get_type: assertion `value != NULL' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_type_is_subtype_of: assertion `g_variant_type_check (type)' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_get_boolean: assertion `g_variant_is_of_type (value, G_VARIANT_TYPE_BOOLEAN)' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_unref: assertion `value != NULL' failed

(mate-user-share:19168): GLib-GIO-CRITICAL **: g_settings_get_value: assertion `G_IS_SETTINGS (settings)' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_get_type: assertion `value != NULL' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_type_is_subtype_of: assertion `g_variant_type_check (type)' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_get_boolean: assertion `g_variant_is_of_type (value, G_VARIANT_TYPE_BOOLEAN)' failed

(mate-user-share:19168): GLib-CRITICAL **: g_variant_unref: assertion `value != NULL' failed

Version 1.6.0-1

Could this be replaced with a dependency on blueman ?