Today I discovered a rather strange anomaly in JavaScript, or more specifically in the DOM.
The Background - you can skip this if you'd like and read from the next bolded label.
I was recently assigned the task of making a complicated table be sortable by any column. The table is complicated because each row in the table contains, in its first cell, a "+" icon that can be clicked to expand additional data about that row. This extra data is really just the following <tr> row, which was set to "display:none" initially.
So every second row in this table is hidden by default until the user clicks a + icon. So, existing table sorting scripts wouldn't work, because they'd try to also take into account the contents of these hidden rows when figuring out how to re-order the rows in the table, which would either cause JS errors or, at the very least, totally mix up the table so it doesn't make any sense anymore to the human.
Rather than haxxor up an existing script until it works, I decided it would be better if I wrote one from scratch that has been built from the ground up specifically to handle complicated tables like this.
The Weirdness
The logical approach I was thinking of, for how to re-arrange the rows in the table, was to delete all the rows first and then reinsert them in the correct order.
This approach didn't work. The new rows seemed out-of-place; the cells in the new rows no longer lined up with the cells of the header row. For my approach to work I would have had to copy all attributes from all cells, and any style information, and then reapply all of these attributes and styles for each row I insert back. This isn't acceptable.
I was looking at other codes for sortable tables while I was coding mine, and the one I was looking at had a really strange way of re-ordering the rows:
Instead of deleting rows and re-adding them, it instead simply re-appended the already existing rows.
Basically, it did this to copy the row objects into an array:
Then sorted newRows, and then it re-inserted them into the table by simply doing this:
When I looked at the code, I logically assumed that this would duplicate all the rows in the table, and you'd wind up with a table twice as long as you started with. But in fact, when you do appendChild to insert a row that already exists in the table, it simply moves that row to the end of the table.
Can anyone explain why this is? I wondered if it was just a Firefox bug (the code I was referencing was used on our intranet, and everyone here uses Firefox so I was wondering if it's simply gone unnoticed all this time - but on IE 8 the code still works).
For the full code for my custom table sorter, look here: (its a temporary link and within 6 months it'll be gone for sure).
Kirsle.net | My personal homepage
The Background - you can skip this if you'd like and read from the next bolded label.
I was recently assigned the task of making a complicated table be sortable by any column. The table is complicated because each row in the table contains, in its first cell, a "+" icon that can be clicked to expand additional data about that row. This extra data is really just the following <tr> row, which was set to "display:none" initially.
So every second row in this table is hidden by default until the user clicks a + icon. So, existing table sorting scripts wouldn't work, because they'd try to also take into account the contents of these hidden rows when figuring out how to re-order the rows in the table, which would either cause JS errors or, at the very least, totally mix up the table so it doesn't make any sense anymore to the human.
Rather than haxxor up an existing script until it works, I decided it would be better if I wrote one from scratch that has been built from the ground up specifically to handle complicated tables like this.
The Weirdness
The logical approach I was thinking of, for how to re-arrange the rows in the table, was to delete all the rows first and then reinsert them in the correct order.
This approach didn't work. The new rows seemed out-of-place; the cells in the new rows no longer lined up with the cells of the header row. For my approach to work I would have had to copy all attributes from all cells, and any style information, and then reapply all of these attributes and styles for each row I insert back. This isn't acceptable.
I was looking at other codes for sortable tables while I was coding mine, and the one I was looking at had a really strange way of re-ordering the rows:
Instead of deleting rows and re-adding them, it instead simply re-appended the already existing rows.
Basically, it did this to copy the row objects into an array:
Code:
newRows[ i ] = table.rows[ i ];Then sorted newRows, and then it re-inserted them into the table by simply doing this:
Code:
table.tBody[0].appendChild (newRows[i]);When I looked at the code, I logically assumed that this would duplicate all the rows in the table, and you'd wind up with a table twice as long as you started with. But in fact, when you do appendChild to insert a row that already exists in the table, it simply moves that row to the end of the table.
Can anyone explain why this is? I wondered if it was just a Firefox bug (the code I was referencing was used on our intranet, and everyone here uses Firefox so I was wondering if it's simply gone unnoticed all this time - but on IE 8 the code still works).
For the full code for my custom table sorter, look here: (its a temporary link and within 6 months it'll be gone for sure).
Kirsle.net | My personal homepage
Code:
perl -e '$|=$i=1;print" oo\n<|>\n_|_";x:sleep$|;print"\b",$i++%2?"/":"_";goto x;'
