There's no use case for stateful comparators, so they can be just plain
function pointers.
This is used in some hot places like CSS selector matching.
Differential Revision: https://phabricator.services.mozilla.com/D77084
This is so that we can avoid needing mozSystemGroup (to get keypress for
non-printable keys), which in turn prevents racing with the native key event
listeners, see the second patch in bug 1624657.
This turned out to be a bit tricky, because we need to guarantee the ordering of
the search one-offs handling in the searchbar with the usual autocomplete-input
handling. We could try to move this to the popup subclass but this is already a
bigger patch than what I'd like.
We can also revert the customElements.js change I did in bug 1624657, as the bug
is no longer relevant.
Differential Revision: https://phabricator.services.mozilla.com/D69743
This is so that we can avoid needing mozSystemGroup (to get keypress for
non-printable keys), which in turn prevents racing with the native key event
listeners, see the second patch in bug 1624657.
This turned out to be a bit tricky, because we need to guarantee the ordering of
the search one-offs handling in the searchbar with the usual autocomplete-input
handling. We could try to move this to the popup subclass but this is already a
bigger patch than what I'd like.
We can also revert the customElements.js change I did in bug 1624657, as the bug
is no longer relevant.
Differential Revision: https://phabricator.services.mozilla.com/D69743
The inclusions were removed with the following very crude script and the
resulting breakage was fixed up by hand. The manual fixups did either
revert the changes done by the script, replace a generic header with a more
specific one or replace a header with a forward declaration.
find . -name "*.idl" | grep -v web-platform | grep -v third_party | while read path; do
interfaces=$(grep "^\(class\|interface\).*:.*" "$path" | cut -d' ' -f2)
if [ -n "$interfaces" ]; then
if [[ "$interfaces" == *$'\n'* ]]; then
regexp="\("
for i in $interfaces; do regexp="$regexp$i\|"; done
regexp="${regexp%%\\\|}\)"
else
regexp="$interfaces"
fi
interface=$(basename "$path")
rg -l "#include.*${interface%%.idl}.h" . | while read path2; do
hits=$(grep -v "#include.*${interface%%.idl}.h" "$path2" | grep -c "$regexp" )
if [ $hits -eq 0 ]; then
echo "Removing ${interface} from ${path2}"
grep -v "#include.*${interface%%.idl}.h" "$path2" > "$path2".tmp
mv -f "$path2".tmp "$path2"
fi
done
fi
done
Differential Revision: https://phabricator.services.mozilla.com/D55444
This patch marks `ScrollContentIntoView()` as `MOZ_CAN_RUN_SCRIPT` and changing
some callers of them to guarantee thar their parent callers are also safe.
Additionally, this patch moves it from `nsIPresShell` to `PresShell` because
all callers can refer `PresShell` directly.
Unfortunately, this patch changes a lot of methods in autocomplete and satchel
since this patch needs to mark some interface methods as `can_run_script` and
they are called each other. This means that autocomplete module is really
sensitive like editor module. Perhaps, autocomplete and satchel should do
scroll asynchronously and unmark some of them as `MOZ_CAN_RUN_SCRIPT` again.
Differential Revision: https://phabricator.services.mozilla.com/D28320
Normally autocomplete results are cached based upon the search string but to get the desired behaviour we want two different sets of results for the same search string depending on how the autocomplete search was started:
a) Via automatically focusing a password field.
b) Every other method of starting an autocomplete search.
In order to not have cached results used, the result code for case (a) [an empty result] will be `RESULT_FAILURE` and I've updated the autocomplete code to not re-use an error result.
In the coming months we may be rewriting our content autocomplete code but that would be too risky to uplift to 67 so for now I'm tracking when satchel automatically opens the popup upon focus and then using that state in the autocomplete result creation code to know whether to include the footer.
Differential Revision: https://phabricator.services.mozilla.com/D25173
Some pages, like the Google login form, submit information on keydown, that
causes us to not autocomplete, because we handle keypress instead.
The patch changes autocomplete to happen on keydown.
Unfortunately formautofill also uses keydown and tries to access popup data too
late, thus it needs some hacks to work properly.
In general the formautofill code has too many indirections due to e10s, and that
makes the fix more fragile than we'd want. Ideally content autocomplete should
have its own codebase, rather than sharing the same controller as chrome code.
MozReview-Commit-ID: oAyASmDFm1
The trigger is missing a WHERE clause, and as such it ends up doing a lot more work than necessary.
The nsAutoCompleteController changes cover a bug where VK_RIGHT autofills when it's not necessary,
and by doing that it will move the caret to an unexpected position (far right) even if it didn't
actually complete some preselected text.
MozReview-Commit-ID: 1mVbxCdqVSr
When Escape key is pressed, nsAutoCompleteController needs to restore last
string which was default value of the input or typed by the user. However,
somebody may change the value, e.g., an event listener which handles
Escape key. In this case, nsAutoCompleteController shouldn't restore the
last string.
Unfortunately, when JS sets input value, DOM "input" event won't be fired.
Therefore, nsAutoCompleteController doesn't have a chance to modify
mSearchString in this case. Therefore, nsAutoCompleteController needs to
store expected input string for checking if somebody modified the input value.
For solving this issue, this patch adds a new member, mSetValue which is
modified when the input value is modified by nsAutoCompleteController itself
or mSearchString is modified.
Even with this patch, if user temporarily selects an item of the popup and
JS sets same value as the selected item from JS, nsAutoCompleteController
restores the input value with mSearchString. However, this must be rare
case and I don't have idea to fix this issue with simple patches.
MozReview-Commit-ID: lig8c7xvD7
Fixes a timing bug where in certain moments matchCount may not be in sync with the current search status,
due to previous results not being cleared immediately. Still delays tree updates to avoid UI flickering.
Fixes a theorical timing issue in unifiedComplete where a stopped search could notify a result.
Removes the no more used OnUpdateSearchResult API.
MozReview-Commit-ID: COoIN4oQT4v
Due to recent changes to tabbrowser focus behavior, now the "focus" event to the location bar
happens before the "TabSelect" event. On "focus" we would like to open the location bar popup,
but detaching the controller would immediately close it. Thus we don't want "TabSelect" to
detach the controller just to reset its internal state. Moreover, this should be cheaper.
MozReview-Commit-ID: 5NZ1TTI9NFW
