This makes sure that the first option gets focused
after the select is open and also if the options are changes
in cases where the options are being loaded and passed
to the select at a later point.
This commit addresses the issue where the overlay in components does not hide
when clicking outside the component, if the click event is stopped using
stopPropagation on an element or its parents.
The solution changes the event listener for detecting outside clicks to be
triggered in the capture phase instead of the bubbling phase.
This ensures that the outside click is correctly detected even if
stopPropagation is called.
On a datatable that has the paginator enabled, with lazy=true, and selectionMode="multiple", then on the first page, shift-clicking multiple rows functions properly as expected by selecting all shift-clicked rows. However, when navigating to any page other than the first page, shift-clicking does NOT work, and instead cancels out any selection.
What's happening is that inside the [`selectRange()`](https://github.com/primefaces/primevue/blob/master/packages/primevue/src/datatable/DataTable.vue#L1190) method of datatable, `this.processedData` contains an array of data that is rendered on the current page of the table, and we use the indexes of rows that the user has shift+clicked in order to access the array by those indexes, which we call [`rangeStart` and `rangeEnd`](https://github.com/primefaces/primevue/blob/master/packages/primevue/src/datatable/DataTable.vue#L1190). For example, consider a collection of data containing 50 total items. On a datatable displaying 10 rows at a time, `this.processedData` will contain an array of length 10, and when the user shift+clicks the first three rows, we will assign a `rangeStart` of 0 and a `rangeEnd` of 2 for selection.
The problem is that when we navigate to the other pages other than the first, `this.processedData` still only contains a subset of the total dataset, but our `rangeStart` and `rangeEnd` indexes contain the indexes relative to the entire total dataset. Continuing on with the previous example, when we navigate to page 2 and try to shift+click the first three rows, `this.processedData` still only contains an array of length 10, since we are only rendering 10 items at a time, but `rangeStart` will be 10 and `rangeEnd` will be 12, which are both outside the bounds of the `this.processedData` array so no rows end up being selected.
Currently there is [code that meant to account for this](https://github.com/primefaces/primevue/blob/master/packages/primevue/src/datatable/DataTable.vue#L1182C13-L1182C45), by subtracting the number rows from previous pages so that we get the properly offset indexes relative to the `this.processedData` subset, not the total. Specifically, it first determines `rangeStart` and `rangeEnd` relative to the total dataset, and then subtracts `this.first` from both values, where `this.first` appears as though it's meant to be the number of items that are contained in the pages previous to the current page (a.k.a. the index of the _first_ item on the _current_ page of the subset of data from the total dataset).
The bug is that `this.first` does not actually contain the expected value here to properly perform the offset. Continuing from the above example, when shift+clicking the first three rows on the second page of results, `rangeStart` and `rangeEnd` will be 10 and 12 respectively, and we would expect that we subtract 10 from the first page of results to get correct values of 0 and 2. However, `this.first` is actually 0, and so we end up with `rangeStart` and `rangeEnd` of 10 and 12, which are outside the bounds of `this.processedData` which cannot be processed on subsequent lines.
This PR attempts to fix this bug by adjusting the subtraction to use `this.d_first` instead of `this.first`, which has been confirmed to contain the desired value of the total number of rows in previous pages. In the above example, it contains 10 for the second page, 20 for the third page, 30 for the fourth page, so on, which causes `rangeStart` and `rangeEnd` to contain the proper expected values for accessing `this.processedData` correctly.
Alternatively, we could've instead updated the assignment of `this.first` to contain `this.d_first`, so that the current code here would function correctly. However, I'm not sure what other undesired effects that might cause, so am requesting more knowledgeable eyes to review the above proposal. Many thanks!! 🙏❤️
tugcekucukoglu