The "ready to write" notifier we set up in generic socket connection is
really only meant to detect a successful connection. Once we have a TCP
connection, for example, it will fire on every event loop iteration.
This was causing IRC Client to max out the CPU by getting this no-op
notifier callback over and over.
Since this was only used by TLSv12, I changed that code to create its
own notifier instead. It might be possible to improve TLS performance
by only processing writes when actually needed, but I didn't look very
closely at that for this patch. :^)
This fixes an issue where continuously posting new events to the queue
would keep the event loop saturated, causing it to ignore notifiers.
Since notifiers are part of the big select(), we always have to call
wait_for_events() even if there are pending events. We're already smart
enough to select() without a timeout if we already have pending events.
Moves DirectoryServices out of LibCore (because we need to link with
LibIPC), renames it Desktop::Launcher (because Desktop::DesktopServices
doesn't scan right) and ports it to use the LaunchServer which is now
responsible for starting programs for a file.
Prior to this commit, we would (re-)allocate the output buffer aligned
to 1024 bytes, but never trim it down to size, which caused
Gzip::decompress to return uninitialised data.
`read_bool_entry()` can now interpret both integers (1 or 0) and
Boolean strings ("true" or "false") in configuration files.
All values other than "1" or "true" are considered false.
It was impractical to return a RefPtr<File> since that left us no way
to extract the error string. This is usually needed for the UI, so the
old static open() got basically no use.
The event that triggered the exit from an inner event loop would always
get re-delivered in the outer event loop due to a silly off-by-one
mistake when transferring pending events between loops.
This helper opens a file with a given name, mode and permissions and
returns it in a RefPtr<File>. I think this will be a bit nicer to use
than having to go through Core::File::construct() every time.
The kernel was already using the UDP prefix, and the TCP LibCore classes
are also uppercased. Let's rename for consistency.
Also order the LibCore Makefile alphabetically, because everywhere else
seems to be doing that :)
Add a SetInspectedObject call that tells us which Core::Object a remote
client is currently looking it. Objects get notified when they gain
their first inspector, and when they lose their last one.
Since the returned object is now owned by the callee object, we can
simply vend a ChildType&. This allows us to use "." instead of "->"
at the call site, which is quite nice. :^)
Previously, this function was using `AK::String::to_uint()`, which is
wrong considering the function returns type `int`. This also means that
configuration files would revert to the default value on negative
values.
Consider the old pattern for creating a Core::Object parent and child:
auto parent = Core::Object::construct(...);
auto child = Core::Object::construct(..., parent);
The above was an artifact of the pre-reference-counting Object era.
Now that objects have less esoteric lifetime management, we can replace
the old pattern with something more expressive:
auto parent = Core::Object::construct(...);
auto child = parent->add<Core::Object>(...);
This reads a lot more naturally, and it also means we can get rid of
all the parent pointer arguments to Core::Object subclass constructors.
