Neil Hodgson wrote:
Robert Roessler:
What I see in the GTK [on Windows] widget incarnation of Scintilla is
that a click *in* the last char on a line (but to the right of the
*midpoint*) is returning the -1 code for "no char close to here". All
other cases work as expected/wanted... in particular, horizontal
matching seems to give a "hit" if within half a char width to the
right or left of a char that is surrounded by spaces. So one might
assume that this would also work at the end of a line...
I see it too with Windows native version. There is a small chance
this was deliberate but I can't recall any reasons why this could be
helpful so it should be fixed.
Now that I can debug this, I can already see that in fact, it is
always "wrong": ALL char matches stop working after the midpoint of
the char... my client code masked this because I always take the
"match" pos and search back for a word start and forward for a word
end (which will work even in the face of this buglet).
Offhand, it looks like it would be nice to have an explicit "width"
attribute available per char in line, rather than depending on being
able to compute it from the start of the *next* char... but no doubt
there is a way of faking this.
I have tried to get debugging prints to work from
Editor::PositionFromLocationClose(), but am having no luck with my
usual approach of inserting Windows kernel debugger calls to
OutputDebugString and monitoring the output with Dbgview from
Sysinternals - I see none of my trace messages when running in the GTK
widget environment.
OutputDebugString should work. Make sure you are using the expected
...W or ...A version as I suspect GTK+ will prefer Unicode APIs.
Thanks. This is working today with either A or W (and L"..." strings)
- but I could *not* make it work yesterday. ;)
Finally, I notice that when I use a SciLexer.dll with my trace
messages with SciTE, the internal functions
Editor::PositionFromLocation() and Editor::PositionFromLocationClose()
are called at a very high rate with any mouse movement - no buttons
down etc. Is this happening to support "hotspot" styles whether they
are in use or not, or is this happening for some other reason?
Hotspots, dwell events and cursor choice.
Sure keeps things busy.
Robert Roessler
[EMAIL PROTECTED]
http://www.rftp.com
_______________________________________________
Scintilla-interest mailing list
[email protected]
http://mailman.lyra.org/mailman/listinfo/scintilla-interest
