The preference determining where to open those windows
is not currently honored, so everything goes in a new tab.
Incidentally compile time checks guarding the web inspector
are replaced with runtime checks here, and the same
approach is used for the create-web-view signal. The
features are used if available.
Also three strings are marked untranslatable since they never
appear in the user interface and 'Unhandled settings value'
is reworded to 'Invalid configuration value'.
intltool automatically embeds comments in the po template
if they are directly before a translatable string. From
now on we will try to avoid leaving accidental comments
and instead put descriptive comments for translators
before strings, starting with 'i18n'.
KatzeNet can from now on be used where
URIs are handled and one would have used
libSoup and/ or local file handling as
appropriate, whereas KatzeNet does both.
Since we are displaying icons in several
places KatzeNet also provides an icon
loader that saves us from doublicating
even more code.
All bookmarks and also history items have
icons now, since KatzeNet makes that
incredibly easy.
Search engines are also using favicons
now, and right now custom icons don't
work, we still need to fix that.
Note that only icons are cached, nothing
else and the code is known to have a
hidden, hard to reproduce crasher which
may appear when an icon is newly loaded.
The previous code relied on GIO implementing HTTP
with libsoup, so this is effectively not a new
dependency. And since many people don't have GVfs
and even if they do, somehow GIO doesn't work as
good as one should expect, direct use of libsoup
is actually an improvement. Plus the new code
caches icons on disk which we didn't do before.
Property name localization is only useful if the
strings are meant to be used in a user interface,
such as is the case with WebSettings, or for
graphical interface builders, which are not of
our concern, since we don't use Glade and friends
and nobody else is going to use our widgets.
In short, make translation work much easier.
We are not literally undoing in the sense of
reverting, this is actually about undoing
the individual parts of the last refactoring
that implemented the socket/ plug logic.
As turned out, the idea was nice but the
implementation was absolutely not, embedding
of processes in a graphical way is not
at all reliable enough for complex use cases.
Naturally this change should solve all sorts
of peculiar issues, including actual
regressions in functionality. Relieving.
