| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
It is racy and it is unnecessary as we've replaced all its instances
with g_file_set_contents().
|
|
|
|
|
|
|
|
| |
ephy_string_get_host_name()
And return NULL like for file:// URIs.
https://bugzilla.gnome.org/show_bug.cgi?id=691416
|
|
|
|
|
|
|
| |
Add ephy_session_open_uris() instead and use it from EphyShell instead
of scheduling a command.
https://bugzilla.gnome.org/show_bug.cgi?id=641739
|
|
|
|
|
|
| |
resume command
https://bugzilla.gnome.org/show_bug.cgi?id=641739
|
|
|
|
|
|
|
|
|
| |
Make ephy_session_load() asynchronous and add
ephy_session_load_from_stream() to replace
ephy_session_load_from_string(). Use a xml sax parser now that is fed
from a GInputStream.
https://bugzilla.gnome.org/show_bug.cgi?id=681782
|
|
|
|
|
|
|
|
| |
It's always set right after the download object is created and it
doesn't have any effect when it's set after the download has been added
to the embed shell. This way we can make sure that when the download is
added to the embed shell the parent window has already been set if
there's any.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Check whether the app that will launch a given download is actually
the browser itself, and do nothing *before* going ahead. Seems better
than actually launching and then aborting on startup through UUID
hacks.
|
|
|
|
|
|
|
|
|
| |
GNOME Shell tries to open new windows by calling the application with no
parameters. This strategy is also used by the 'Launch web browser' keybinding,
that can be interpreted as the user requesting a new window, so open a new
window when called with no URIs, unless --new-tag is given explicitly.
Partially fixes https://bugzilla.gnome.org/show_bug.cgi?id=685976
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=683614
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=683614
|
| |
|
|
|
|
|
| |
When called to open URIs, the shell will still queue a resume session command,
this test ensures that command is not loading the session and creating windows.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=687744
|
|
|
|
|
|
|
|
|
| |
New warning flags have been added to gnome-common recently, and we
were getting this a lot. Turns out in C 'foo ()' is not the same than
'foo (void)'; the first just means that no information is given about
the number of arguments, the second means the function has exactly
zero arguments, so add the 'void' thing all over the place when
needed.
|
|
|
|
| |
That's what it's for!
|
|
|
|
|
|
|
|
|
|
|
|
| |
EphyShell is now a GtkApplication that already does window handling,
however we want to maintain only the of browser windows. Instead of
keeping the window lists in ephy-session, it's easier to override
window_added and window_removed virtual methods of GtkApplication and
update the browser window lists. Most of the cases where this API was
used, had to get the session object from the shell first, now we save
this step keeping the code simpler.
https://bugzilla.gnome.org/show_bug.cgi?id=641734
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
ephy_snapshot_service_get_snapshot_async() receives an option web view
parameter, that it's only used in case the snapshot is not the in the
thumbnails cache. We can split the method into
ephy_snapshot_service_get_snapshot_async() to get a snapshot from a web
view and ephy_snapshot_service_get_snapshot_for_url_async() to get a
snapshot from the cache. The former uses the latter to try first if the
web view URI is in the cache.
Patch includes other cleanups and fixes:
- Add missing ephy_snapshot_service_save_snapshot_finish()
- Add EphySnapshotServiceError to handle errors
- Use GSimpleAsyncResult API instead of using GIOScheduler directly
- Use different async data structs for every async operation
https://bugzilla.gnome.org/show_bug.cgi?id=683327
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Needs a fairly recent gnome-common. To use do:
- Run configure with --enable-code-coverage
- Type 'make check-code-coverage'
- Open the HTML results
https://bugzilla.gnome.org/show_bug.cgi?id=683297
|
|
|
|
|
|
|
|
|
| |
ephy_profile_utils_do_migration()
This is passed further on to the migrator process. Right now this is
unused.
https://bugzilla.gnome.org/show_bug.cgi?id=681679
|
|
|
|
|
| |
Since this is also used for persistency of temporary directories, but
also for --profile.
|
|
|
|
|
|
| |
Unfortunately the state propagation of the overview loading needs the
mainloop to run, since we use the load status signals from WebKit. Do
so and update things for about:blank -> about:overview.
|
|
|
|
| |
This fixes one test, there's another one broken in this file.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This usually happens in _run, it seems, but the tests don't do that,
so we need to call this manually. Otherwise the test will fail with:
(/home/xan/git/epiphany/build/tests/test-ephy-session:9865): GLib-GIO-CRITICAL
**: g_application_list_actions: assertion `application->priv->is_registered'
failed
No idea why this has started happening only now.
https://bugzilla.gnome.org/show_bug.cgi?id=682653
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=668578
|
|
|
|
| |
And add unit tests for it.
|
|
|
|
|
|
| |
SoupTLD is not happy if we give it something like ".com/blah/blah".
Add a unit test for this case too.
|
|
|
|
|
|
|
|
|
| |
Otherwise we end up with dummy URIs in the history, overwritten titles
("Oops, could not load..."), etc.
Add a unit test to make sure we do not regress.
https://bugzilla.gnome.org/show_bug.cgi?id=655619
|
|
|
|
|
|
|
| |
webkit_web_view_load_string, which was used previously, does not add
the error page to the back forward list. This means that when you
go back from an error page (n) you will skip directly to an older page
(n-2) or not be able to go back if the error page was the first loaded.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Give SoupTLD nothing but the TLD when trying to figure out whether a
URI actually has a TLD, otherwise it won't be happy. In particular:
'.gnome.org' -> Not OK
'.org' -> OK
Add a test for this case.
|
|
|
|
|
|
|
|
|
| |
Using the new SoupTLD methods. Had to split the non-search regexp in
two so that we can reuse its 'is this a domain?' bits. Ugly as hell,
but not worse than before... At least we have unit tests to catch
regressions.
https://bugzilla.gnome.org/show_bug.cgi?id=681022
|
|
|
|
|
| |
We are going to modify it a bit, and we really want to test it. Add
some initial tests for the existing functionality.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=680590
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=680905
|
|
|
|
| |
They are derived from 'title', no need to pass them as parameters.
|
|
|
|
|
| |
We should make the test-elided and collation-key properties be
computed automatically from the other parameters.
|
| |
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=678532
|
|
|
|
| |
It builds and basic functionality works.
|
|
|
|
| |
Instead of relying on webkit header to include it.
|
| |
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=673683
|
|
|
|
|
|
|
| |
Tests create, exists, delete, list API.
No network access required.
https://bugzilla.gnome.org/show_bug.cgi?id=673271
|
|
|
|
|
|
|
|
|
| |
For now a simple test that verifies that loading a session works.
We need to link with the resource files, which is unfortunate, but I'm
not sure there's a way around it since this test will actually spawn a
full EphyWindow (ie, it's not only testing the session or shell code
but the entire browser as a side effect).
|
|
|
|
| |
Otherwise tests using its contents will fail on distcheck.
|
|
|
|
|
|
|
|
|
|
| |
Test API for file switch, tmp file creation, downloads, desktop
directories.
Take two of this test. Updated for ephy-file-helpers API changes.
Previously reverted on 185ab751c20a825f832efd89a7cfbd75fa09e20c.
https://bugzilla.gnome.org/show_bug.cgi?id=673337
|
| |
|
|
|
|
|
| |
Trivial first test, we just check the migration binary runs and does
not catastrophically.
|
|
|
|
|
| |
Perhaps PRIVATE_PROFILE should implicitly add ENSURE_EXISTS, seems
logical. Do this for new.
|
|
|
|
|
|
|
|
|
|
| |
Since we are about to migrate our profile dir, allow file helpers init
to not ensure the profile dir exists (it was hardcoded until now). For
this we get rid of the ugly boolean parameters and add a flags
parameter, which preserves the old behaviors and allows for this new
option.
We update all the callers in the tree.
|
|
|
|
|
|
| |
This reverts commit d1ba46b3f4a4e2afbffe5a0ab2543462caea0bbe.
Does not pass distcheck and I need to release.
|
|
|
|
|
|
|
| |
Test API for file switch, tmp file creation, downloads, desktop
directories.
https://bugzilla.gnome.org/show_bug.cgi?id=673337
|
|
|
|
| |
These are clearly not meant to be used outside of Epiphany.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=673684
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=673273
|
|
|
|
|
|
|
|
| |
Do not use adhoc EphyDownload objects that have no proper destination
set, or your current user's downloads directory will be filled with test
files.
https://bugzilla.gnome.org/show_bug.cgi?id=673274
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=671792
|
|
|
|
| |
We are not using it. Easy enough to bring it back if needed.
|
|
|
|
|
|
|
|
|
|
|
| |
The main purpose of the cancellable API in ephy-history-service
is to let the user notify when the results of the operation
are no longer needed and the callback call can be omitted. Since
performing a read operation, in such cases, makes no sense, we
cancel them altogether. However, given the nature of the service,
we don't cancel write operations.
https://bugzilla.gnome.org/show_bug.cgi?id=671901
|
|
|
|
|
|
| |
Testing ephy_web_view_load_url and the internal GRegex of EphyWebView.
https://bugzilla.gnome.org/show_bug.cgi?id=669548
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Zoom levels have always been stored per host in Epiphany, not per URL
(otherwise you'd have to re-apply the zoom level again and again when
visiting, say, every news entry in your favorite newspaper). Change
things to work like that.
Note: this changes the SQL table format for the history, so you'll
need to re-migrate your history.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
EphyHistoryService provides a high-level API to store history
information. It will processed by a worker thread using SQLite to
provide a fast, responsive service to the main UI.
Based on the code by Martin Robinson (mrobinson@igalia.com) and
Claudio Saavedra (csaavedra@igalia.com).
|
|
|
|
|
|
|
| |
We'll use them to implement the new history/bookmarks storage backend.
Code by Martin Robinson (mrobinson@igalia.com) and Claudio Saavedra
(csaavedra@igalia.com)
|
|
|
|
|
| |
It's actually impossible to catch this through normalization, so it
will need to be moved to an EphyWebView test.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=669461
|
| |
|
|
|
|
| |
They were testing the same URI again and again...
|
|
|
|
|
|
| |
To avoid trivial line changes, ironically.
https://bugzilla.gnome.org/show_bug.cgi?id=669766
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=669766
|
|
|
|
|
|
|
| |
Append -test to .c files in tests/ to avoid duplicating filenames in the
repository.
https://bugzilla.gnome.org/show_bug.cgi?id=669766
|
|
|
|
|
|
|
| |
Specifically ephy_embed_utils_address_has_web_scheme.
Based on a patch by Alexandre Martani.
https://bugzilla.gnome.org/show_bug.cgi?id=613756
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=668758
|
| |
|
|
|
|
|
|
| |
In 99% of the cases we were using it for gtk_window_set_icon_name,
which does nothing in GNOME 3. In the couple of valid use cases, just
use "web-browser" directly.
|
|
|
|
| |
I'm sure Diego will fix them.
|
|
|
|
| |
Which tells us that test-ephy-download is broken.
|
|
|
|
|
|
|
|
| |
Instead of a GtkToolItem. A future commit will move the code in
EphyLocationAction not to be a GtkAction, which is why this was a
GtkToolItem in the first place.
https://bugzilla.gnome.org/show_bug.cgi?id=668206
|
| |
|
|
|
|
| |
They are not needed anymore.
|
|
|
|
|
|
|
|
|
|
|
| |
We want to use EphyApplication from embed/ too, so instead of
following the EphyEmbedShell/EphyShell pattern just fold the code into
the Shell itself. This makes sense since both classes have basically
the same function, and we can easily leave the non-embed code in
EphyShell and move to EphyEmbedShell the bits we want to be global.
The better name would be EphyCoreApplication/EphyApplication for the
resulting object, but we can do this later.
|
|
|
|
| |
We'll want to create our own in main(), this prepares us for that.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
EphyDownload is a wrapper object around WebKitDownload that handles common
behavior in downloads: auto-destination, default action for the MIME type.
It can be used to wrap a WebKitDownload coming from a WebKitView or to download
a url: ephy_download_new_for_uri and ephy_download_new_for_download are
provided.
Its lifetime is not automagic like EphyEmbedPersist, so you have to unref it
when you no longer need it.
This new object replaces EphyEmbedPersist and enables us to use a single
codepath for downloads in all Epiphany.
Bug #618443
|
|
|
|
| |
Make filenames readable and tab-completion friendly.
|
|
|
|
|
|
|
|
|
|
|
| |
The migration code is now in the ephy-profile-migrator binary, this means:
- epiphany is not linking to NSS anymore
- lib/ephy-profile-migration.c was split into:
+ lib/ephy-profile-migrator.c: the new ephy-profile-migrator binary
+ lib/ephy-profile-utils.c: convenience _ephy_profile_*_form_auth_data functions.
- testing for migration is done without running ephy-profile-migrator
Bug #636685
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
This flag determines if a download should not appear in the DownloaderView
window. The default is to appear, like in gecko times.
Bug #611635
|
|
|
|
| |
Bug #600987
|
|
|
|
| |
Don't cherry pick internal libraries for tests.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
EphyLocationEntry, EphyLocationAction and EphyToolbar take an @adress and
@typed_address parameter, while both are useful we only end up using one so we
can easily decide which one to use in ephy-window.c instead of carrying both
around until ephy-location-entry.c
Bonus: make ephy_location_entry_set_location accept NULL as @address safely.
Bug #603651
|
|
|
|
| |
svn path=/trunk/; revision=8655
|
|
|
|
|
|
| |
There's no 'internal entry' now.
svn path=/trunk/; revision=8654
|
|
|
|
| |
svn path=/trunk/; revision=8621
|
|
|
|
| |
svn path=/trunk/; revision=8581
|
|
|
|
| |
svn path=/trunk/; revision=8569
|
|
|
|
| |
svn path=/trunk/; revision=8554
|
|
|
|
|
|
| |
Add vim modelines to the files.
svn path=/trunk/; revision=8553
|
|
|
|
|
|
| |
See the FIXME about the mystery crasher, it's probably some kind of leak.
svn path=/trunk/; revision=8552
|
|
|
|
|
|
| |
This reverts commit 58afaf81fbf64dfcb779ef2e7531afb4c11693e1.
svn path=/trunk/; revision=8551
|
|
|
|
| |
svn path=/trunk/; revision=8439
|
|
|
|
|
|
| |
Part of #544642.
svn path=/trunk/; revision=8385
|
|
Include a test for ephysearchentry, need to make them optional too.
Part of #544642.
svn path=/trunk/; revision=8384
|