From: Patrick Lam (plam@plam.lcs.mit.edu)
Date: Mon Aug 05 2002 - 08:57:09 EDT
On Mon, Aug 05, 2002 at 12:43:42AM -0700, Joaquin Cuenca Abela wrote:
> 1) I will like to know if it affects performance.
I can't measure that on my computer.
> 2) Any reason for the mutex to be allocated on the
> heap and not in the stack? That way you will not lack
> it at the end of abiword.
The mutex is static, so there's one which exists throughout the
lifetime of the program. Or that's what I meant to do; I mistyped
my patch, it should really say if (!m_grLock) m_grLock = new UT_Mutex();
> 3) Not all the operations on GR_UnixGraphics do X
> calls. Character measurement (half of them in the
> normal X build, and all them in the Xft build) don't
> makes any X call. Remember that we do *a lot* of
> character measurements...
I'm not sure which glib calls correspond to X calls or not.
e.g. I suspect that gdk_gc_set_line_attributes doesn't, but I don't
know for sure. I will remove the lock around measureUnRemappedChar.
(What about _setColor?)
> 4) Please, commit first my previous patches for
> GR_UnixGraphics, the least our patches conflicts, the
> best.
My patch is actually quite simple and conflicts should be easy to merge.
> If you're going the GR_Caret way, at least give a
> GR_CaretHidder class or something like that.
I thought I had one already. Anyway I'll implement one soon if
there isn't one already.
Bah, I have to run now and cvs is being too slow.
pat
This archive was generated by hypermail 2.1.4 : Thu Aug 08 2002 - 08:55:07 EDT