Lars Gullik Bjønnes wrote:
Georg Baum <[EMAIL PROTECTED]> writes:
| Am Freitag, 27. Oktober 2006 17:03 schrieb Abdelrazak Younes:
| > Hello,
| >
| > This patch brings a 20% improvement when loading a big document (my test
| > document is the UserGuide copied&Pasted 3 times). This is done by
| > avoiding multiple string/vector/docstring copying.
| >
| > I can think of many other optimisation but I would like this one to go
| > in first. IMHO, most of unicode can go and only one iconv_convert()
| > should stay.
| >
| > Objections?
|
| You will probably get some from Lars,
And if I had read the whole patch I would have said: "No Go!" instead
of just complainging about string->vector.
I am not going to have a fight with you on this one. My solution works
and is elegant and more importantly fast.
If you want to revert the code to your old solution, please explain your
rationale on the list (with more pragmatic reasons than "string" is not
appropriate for containing utf8).
Or just use your prerogative as the Dictator and revert my code. I don't
really care. But if you do that I won't work any more on optimizing file
loading.
Abdel.
