https://bugs.freedesktop.org/show_bug.cgi?id=79892

--- Comment #40 from Michael Meeks <[email protected]> ---
I guess improving the spreadsheet to match at the beginning of the string is
not feasible ? =) it is doing a -ton- of un-necessary work left & right in
there.

I can't see us downgrading ICU really. Beyond that if someone wants to
investigate that - they need to generate a matched pair of callgrind traces for
the same workload (with symbols) for the older (faster) version, and the newer
(slower) version - and see if indeed the problem is with ICU. If so, we then
would want to write a micro-benchmark of that, and report to the ICU people
and/or look at exactly what is wrong here.

Failing that, there are of course, lots of potential opportunities for exotic
CPU threading in the formula interpreter; we should be able to parallelize
things rather nicely - for FormulaGroups with over-complicated regexp matching
in-use - but that of course is an even bigger chunk of engineering work: if
someone wants to do that work they can poke me for code pointers =)

-- 
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Libreoffice-bugs mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs

Reply via email to