This reverts commit 3eacfa88f2.
Apart from the patch not being correct, we don't want to expose private
structures in header files if we can avoid it.
And this type-checking overhead is not an optimization that is even
measurable.
https://bugzilla.gnome.org/show_bug.cgi?id=754932
Statistics for the gtk3-demo listbox example show that the
vast majority of calls to _gtk_allocated_bitmask_resize go
from a size of 2 to 2. Don't needlessly call realloc() in
this case.
The previous code was crashing when used as the returned classes array
would have been invalid after the first deletion. So if a 2nd class
would be deleted, invalid memory might have been referenced.
In order to retrieve the user options for a printer, the respective
printer name is used.
This fixes the comparison of printer names to avoid that the options of
another printer are accidently read whose name starts with the same
letters, but is longer (e.g. "myprinterlongername" instead of
"myprinter").
This fixes Bug 753628.
This function is getting called a lot. Statistics for the gtk3-demo
listbox example show most calls with 0-4 classes. Unrolling the cases
a bit brings the instruction count in callgrind from 93M to 52M.
The same trick that was applied to _gtk_css_change_for_child in
the previous commit can be applied to _gtk_css_change_for_sibling
as well, and that is what this commit does.
With both functions converted, gtk_css_change_translate is no
longer needed and gets dropped.
The function to translate GtkCssChange enum values to the PARENT
ones is called very frequently. This patch speeds it up tremendously.
The callgrind instruction count for this function in the listbox
demo goes from 108M to 7M.
My statistics show that more than half of all calls end up
with 0 matches, so we can avoid some overhead by not allocating
an array at all in this case.
We are dealing with really short lists here.
95% are < 10 matches, and the longest I've been able to record was 19.
So just do away with the hash table and do sorted insertion in
the array directly.
Otherwise the outer loop control variable is messed up, and we end
up with uninitialized axes if there were any more valuators after
the XIKeyClass one.
This bug was sneakily introduced by fdb9a8e14, many thanks to
Carlos Soriano for helping spot the source of this bug.
https://bugzilla.gnome.org/show_bug.cgi?id=753431
