On Mon, Mar 17, 2003 at 11:37:18AM +0100, Jean-Marc Lasgouttes wrote: > It would probably be a good idea to run some profiler on this and see > whether things remain sensible.
The results seem a bit odd (playing with TableExamples.lyx) : 0811dec4 594 1.26399 LyXTabular::row_of_cell(int) const 0810dfe0 597 1.27037 Paragraph::getFontSettings(BufferParams const&, int) const 08131d80 617 1.31293 LyXText::getFont(Buffer const*, Paragraph*, int) const 081134fc 645 1.37252 Paragraph::Pimpl::getChar(int) const 0825152c 696 1.48104 QLPainter::text(int, int, char const*, unsigned, LyXFont const&) 082509fc 817 1.73852 QLPainter::setPen(LColor::color, Painter::line_style, Painter::line_width) 081c6500 1004 2.13644 InsetText::draw(BufferView*, LyXFont const&, int, float&) const 08076c6e 1082 2.30242 _Rb_tree<LColor::color, pair<LColor::color const, LColor::information>, _Select1st<pair<LColor::color const, LColor::information>>>::find(LColor::color const&) const 081cc278 1158 2.46414 InsetText::getLyXText(BufferView const*, bool) const 081f2072 1872 3.98349 _Rb_tree<unsigned short, pair<unsigned short const, int>, _Select1st<pair<unsigned short const, int>>>::find(unsigned short const&) (lyx binary only top hits). Note that the ::draw has probably gone up in the profiles, but that's no surprise. The interesting one is the profile including all libraries : 001925c7 1928 0.797577 __i686.get_pc_thunk.bx /usr/lib/qt-3.0.5/lib/libqt-mt.so.3.0.5 4201f0c0 2125 0.879073 new_composite_name /lib/i686/libc-2.2.93.so 4201f3c0 2222 0.9192 _nl_find_locale /lib/i686/libc-2.2.93.so 4202a920 2303 0.952708 __GI_getenv /lib/i686/libc-2.2.93.so 4201575d 2361 0.976701 __i686.get_pc_thunk.bx /lib/i686/libc-2.2.93.so 00006ef0 2849 1.17858 __pthread_mutex_lock /lib/i686/libpthread-0.10.so 42079310 2994 1.23856 strcmp /lib/i686/libc-2.2.93.so 000070c0 3152 1.30392 __pthread_mutex_unlock /lib/i686/libpthread-0.10.so 420730c0 3255 1.34653 __malloc /lib/i686/libc-2.2.93.so 420791a0 3612 1.49422 index /lib/i686/libc-2.2.93.so 420749b0 3663 1.51531 __cfree /lib/i686/libc-2.2.93.so 00006db0 3669 1.5178 __pthread_mutex_trylock /lib/i686/libpthread-0.10.so 00488f80 3960 1.63818 qstricmp(char const*, char const*) /usr/lib/qt-3.0.5/lib/libqt-mt.so.3.0.5 4207c450 4049 1.675 memcpy /lib/i686/libc-2.2.93.so 42047f00 4822 1.99477 _IO_vfprintf_internal /lib/i686/libc-2.2.93.so 420739e0 4974 2.05765 _int_malloc /lib/i686/libc-2.2.93.so 0000a410 5316 2.19913 __pthread_alt_lock /lib/i686/libpthread-0.10.so 4207a4c0 5367 2.22023 strlen /lib/i686/libc-2.2.93.so 42022490 6772 2.80145 tolower /lib/i686/libc-2.2.93.so 00000000 6829 2.82503 (no symbol) /usr/X11R6/lib/libX11.so.6.2 0000a220 7541 3.11957 __pthread_alt_unlock /lib/i686/libpthread-0.10.so 4201ea50 11325 4.68494 __GI_setlocale /lib/i686/libc-2.2.93.so It seems the Qt frontend is manageing to be particularly stupid here. I don't have a slower machine than my Celery 850 to test on I'm afraid john
