[Libreoffice-commits] core.git: editeng: Eliminate unecessary padding in classes

[Bjoern Michaelsen]

Hi,

On Wed, Nov 04, 2015 at 12:22:57PM +0100, Stephan Bergmann wrote:
> Where I consider padding-related changes to be cosmetic, unless there is a
> clear, documented benefit.

I think that (documented benefit, e.g. at least a comparison in valgrind) s the
key here. There are a bazillion micro-optimization tactics out there, many of
them providing conflicting advice. Optimization for one scenario might lead to
performance regressions in other scenarios or on other compilers. So in the
least, for such micro performance changes, we need some reference that shows
demonstratable improvements, _especially_ if the change impacts readability --
if only to make a qualified balanced decision the next time someone proposes an
conflicting improvement. Having people do wikipedia style edit wars optimizing
for two orthogonal goals asking for conflicting changes is ... unhelpful.

Best,

Bjoern

Some recent examples, where the "obviously better solution" isnt that obvious
or universally clearcut as initially thought:

http://chrisdglover.blogspot.de/2015/11/beware-of-codegen-differences-between.html

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fblog.fefe.de%2F%3Fts%3Da8c95274&edit-text=&act=url