| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
This allows us to check further properties such as UUIDs.
|
|
|
|
|
|
|
|
| |
Having only one iso9660 test image available for all Drive tests is too
inflexible, as that is not writable, and has no UUID. Ship an ISO9660 and a
VFAT image in test/files (they compress to < 1 kB) and add a load_image()
method so that every test can pick which kind of data it wants to see on the
test device.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
The content_type argument is required to be set for soup_message_set_request()
if passing data in. Using "application/octet-stream" makes it work,
though I'm not sure whether it's accepted by all dav servers.
https://bugzilla.gnome.org/show_bug.cgi?id=687757
|
|
|
|
|
|
| |
Apparently SoupRequester needs this to be set TRUE.
https://bugzilla.gnome.org/show_bug.cgi?id=687757
|
|
|
|
|
|
|
|
|
|
|
| |
SoupOutputStream was never particularly useful, since we ended up not
doing chunked requests (since server support for them is mostly
nonexistent). So kill SoupOutputStream off and just use a
GMemoryOutputStream instead.
https://bugzilla.gnome.org/show_bug.cgi?id=687757
Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
gvfsbackendhttp defined g_vfs_job_failed_from_http_status(), but
didn't export this, so gvfsbackenddav was sort of forced to
reimplement it. Fix that by exporting it as http_job_failed().
https://bugzilla.gnome.org/show_bug.cgi?id=687757
Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Replace the hacky SoupInputStream with a new GVfsHttpInputStream that
is a wrapper around SoupRequest. (We need a wrapper stream rather than
just using SoupRequest directly because we want the stream here to be
seekable, which requires cancelling and re-sending the HTTP request
and getting a new underlying stream.)
The http and dav backends still use both a sync and an async
SoupSession, even though this is no longer necessary, since changing
this would require a lot of rewriting of code that currently works.
https://bugzilla.gnome.org/show_bug.cgi?id=687757
Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>
|
| |
|
| |
|
|
|
|
|
|
|
| |
If the drive is using removable media, we want the shell and file
manager to eject the media, not turn the drive off.
Signed-off-by: David Zeuthen <zeuthen@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With this change, USB-connected drives that appear *during* the user
session will now be powered off when the user clicks the "Eject" icon
in the GUI. This involves powering off the USB port typically
resulting in LEDs and other user-visible features turning off. In
particular, bus-powered HDDs will completely turn off by virtue of
being, well, bus-powered.
This is generally what users expect and should make them feel all
cozy, warm inside and, most importantly, more confident removing the
device now that the LED is off. Or something.
Since we only do this for devices detected *during* the user session,
we should not run into problems with internal USB devices such as
laptop SD card readers or the various 4-in-1 flash card readers in
many workstations. This is actually important because removing the
power to the USB port of an "internal" device is generally a bad idea
as it requires a system restart to make the laptop BIOS power the port
up again (or the user to crack open the chassis).
(No, unfortunately there is no *reliable* way to determine if a USB
device is "internal" (e.g. reachable and/or servicable by the user) or
not. There are a couple of mechanisms - one is to look at various USB
device descriptors - but none of them seem very reliable. So instead
we simply look at whether the device is hotplugged.)
See also this udisks commit
http://cgit.freedesktop.org/udisks/commit/?id=81dcb6eeaeceb6c6faae1a40a5b34a65cd5af653
introducing the Drive:CanPowerOff property and the Drive.PowerOff()
method.
Signed-off-by: David Zeuthen <zeuthen@gmail.com>
|
|
|
|
|
|
|
| |
We need to set initial value to local variables we're about to use
later in the code.
Found the volume monitor segfaulting on stable udisks-2.0.0
|
|
|
|
|
| |
A matter of developer's taste, but since the function has been simplified
a lot, no need to keep the goto around.
|
|
|
|
|
|
|
|
|
| |
The the_volume_monitors hash table is a shared global instance, make
sure to only initialize it once.
Based on a patch by Tim Lunn <tim@feathertop.org>
https://bugzilla.gnome.org/show_bug.cgi?id=689946
|
|
|
|
|
|
| |
Found by reviewing usage of the global lock, this place is not obvious
whether is inside the lock or not. Let's add a simple comment to make it
clear.
|
|
|
|
| |
Found by Coverity Scan analysis
|
|
|
|
|
| |
Don't try to seek if there was an error during padding. Also report
seek errors.
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=625741
Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>
|
|
|
|
|
|
|
| |
... and also the new Block:HintSymbolicIcon icon just added to udisks
master.
Signed-off-by: David Zeuthen <zeuthen@gmail.com>
|
| |
|
|
|
|
| |
Signed-off-by: David Zeuthen <zeuthen@gmail.com>
|
|
|
| |
Signed-off-by: David Zeuthen <zeuthen@gmail.com>
|
|
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=684328
Signed-off-by: Tomas Bzatek <tbzatek@redhat.com>
|
|
|
|
|
|
| |
We always reported success despite storing the smb call result.
https://bugzilla.gnome.org/show_bug.cgi?id=687778
|
|
|
|
|
|
|
| |
Certain operations may take a lot of time to complete, don't timeout
on that. Use cancellable if you need to control the flow.
https://bugzilla.gnome.org/show_bug.cgi?id=687534
|
|
|
|
|
| |
This covers trashing a file from $HOME (both API and CLI), trying to trash
a file from /tmp/, and handling trashing files with the same path.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For adding further tests such as for trash://, we want to be able to use a
temporary private $XDG_DATA_HOME, and also ensure that we do not touch the
user's configuration in any way by setting a temporary $XDG_CONFIG_HOME as
well. (The latter needs to be in the actual $HOME until
https://bugzilla.gnome.org/show_bug.cgi?id=142568 gets fixed)
To achieve this, set up the temporary session D-BUS in gvfs-test itself (so
that it sees our new XDG environment variables), and stop setting up a session
D-BUS in run-in-tree.sh when running under make. Retain the funtionality in
run-in-tree.sh for calling the script manually.
This also allows us to capture the output of the D-BUS daemon and its attached
processes. Show its stdout and stderr on test case failures for easier
debugging, and enable general GLib and gvfs debug messages.
In addition this simplifies the handling of $LIBSMB_PROG, as we can now set it
in the test suite and can stop setting it in the Makefile.
|
|
|
|
|
|
|
|
|
| |
When running under the normal user account, the StrictHostKeyChecking option is
client-configurable behaviour which cannot be temporarily changed. Skip the
test entirely in this case instead of exfailing it, as it does not test
anything meaningful then.
Drop the exfail so that the test must succeed when running under gvfs-testbed.
|
|
|
|
|
|
|
| |
We were returning folder-remote-symbolic which is incorrect since that
should only be used for network filesystems.
Signed-off-by: David Zeuthen <zeuthen@gmail.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
In SSH_FXP_NAME messages, the name field is preceded by a uint32
count field. query_info_reply() in gvfsbackendsftp.c did not take
the existence of the count field into account, and so set symlink
targets incorrectly.
For a simple demonstration, use gvfs-info to stat a symlink on an
ssh mount.
https://bugzilla.gnome.org/show_bug.cgi?id=652535
|
| |
|
|
|
|
|
| |
Skip the Dav tests instead of failing when running against the build tree and
http support is disabled.
|
|
|
|
|
|
| |
This tries to spawn gnome-keyring which can potentially mess up user's real
keyrings. It also causes long timeouts during D-BUS activation which make the
testsuite painfully slow.
|
|
|
|
|
|
| |
So far we were only using the command line programs. This adds some tests that
exercise the Gio API through gobject-introspection, covering archive mounts,
and anonymous and authenticated FTP.
|