Alfredo Braunstein wrote:
Abdelrazak Younes wrote:
Sure, IIUC ideally we should test if the inset just fits inlined with
the button or not. But the reference width here should be the
textwidth_, not the screen width (think about deeply nested stuff).
Previsouly we had (rev 14161):
openinlined_ = textdim_.wid + 2 * dim.wid <= mi.base.textwidth;
Probably the wide() thing was interfering badly with that. Might be a
good idea to come back to the old formula now. Martin, would you care to
comment?
I've tested the code above and couldn't see a flaw even with deeply
nested insets where the button is barely visible. Result of the
investigation: it works fine even without the additional refit! I also
tested tables and I could not reproduce the symptoms described in the log:
In fact seems almost correct ;-)
[ I still don't get the 2 * but probably it's a safety margin to account for
all TEXT_TO_INSET_OFFSET and such.
Yes, that is my understanding too.
One could even try to put the exact
dimension painted in ::draw ]
Yes.
Fix "inset in tabular cell sticks out over cell border" bug
So, I just put the old code back so that we can test it out extensively
as it seems to me this is the right thing to do. Martin, please don't
hesitate to touch this stuff again if you think you there is a flaw.
Thanks for pushing me Alfredo,
My pleasure ;-) You're doing a nice work.
Thanks :-)
Abdel.
