lyx 1.3(qt) starts in slightly less than a second, assuming I have
started it before so it is in the disk cache. This is snappy enough.
lyx 1.4cvs(qt) starts in 6 seconds, putting a load on the cpu for the
whole time, no disk delays here. This is too long, and users wonder
what's going on and does double & triple starts. Then they get
confused when they get several lyx windows, because "they only
started one ???".
This is lyx starting alone, with no initial document. So paragraph breaking
and other "long document" issues does not apply here.
I used this configuration, hoping to avoid slow debug stuff.
$ ./configure --prefix=/usr/local --with-version-suffix=-1.4cvs
--with-frontend=qt --with-gnu-ld --with-qt-includes=/usr/include/qt3
--enable-assertions --disable-stdlib-debug --disable-concept-checks
I also changed VERSION to 1.4.0 (without cvs) in "configure",
that way I got the -O2 compiler flag. Still it is 6 times slower
to start than lyx 1.3. Is there any other "anti-developer" options
I should use for speed testing, or is the problem in the code?
The machine is a pentium 2.4GHz, and can print many documents
faster than it can start lyx-1.4. The compiler is debian's gcc/g++ 4.0.1
Helge Hafting
