[DotGNU]experimental patch

[Thomas Fritzsche]

Hi Neil + dotgnu hackers,

I send the current state of my DrawString/MeasureString changes to my
homepage http://www.noto.de/dotgnu/DrawString.diff and I made a update to
the screenshot
( http://www.noto.de/dotgnu/DrawString.png )
of the DrawString example in the FormTest sample. As you can see there is
a lot of work I still have to do before you can send this to CVS. I
published this because It could be usefull for other people that develop
stuff that depend on a better DrawString/MeasureString support. This is
still experimental code and a lot of things are unfinished and maybe
broken!

In my personal version I inserted a Console.WriteLine(...) call in the
Toolkit DrawString and MeasureString methods. This way I could check out
that the algorithem now measures full words for word wraping (with the
exception of textboxes) and caches nearly 100% of the MeasureString calls
after the startup, but sadly I have still a performence problem. The ilrun
process now uses 2-3 percent of my CPU in FormTest sample. I don't know
what is going on the rest of the time.

strace shows me always thousends of this calls:
---------------------------------------------------------
rt_sigprocmask(SIG_UNBLOCK, [RTMIN], [RTMIN 43], 8) = 0
gettimeofday({1064961006, 229301}, NULL) = 0
rt_sigprocmask(SIG_SETMASK, [RTMIN 43], NULL, 8) = 0
gettimeofday({1064961006, 229340}, NULL) = 0
rt_sigprocmask(SIG_BLOCK, NULL, [RTMIN 43], 8) = 0
rt_sigprocmask(SIG_UNBLOCK, [RTMIN], [RTMIN 43], 8) = 0
gettimeofday({1064961006, 229397}, NULL) = 0
rt_sigprocmask(SIG_SETMASK, [RTMIN 43], NULL, 8) = 0
---------------------------------------------------------
Perhaps this is a usefull information for someone of the dotgnu core
hacker team.
What is the reason for reading the current time that much? Is this
event/task handling? Is ilrun waiting for the result of pinvoke X calls?

Thanks,
 Thomas