Re: PATCH: Bug 186

> I just made a guess based on all the scrollpanes I use on my computer.
> The only ones with axial slaves are tables. At least on OS X / favorite
> linux progs.

I would agree with you here -- the use of the N/E/S/W panes is typically used 
in graphics apps for rulers, spreadsheet apps for cell references, and that's 
pretty much it.  The ability to throw up a slave on a scrollpane is very very 
cool, because without Charlie's work it'd be a royal pain in the ass.

I don't recall if you'd given a solution to putting something 
mid-to-moderately complex in a slave.  I have an option that I think solves 
both problems -- make the slave template a separate template and then simply 
do something like this:

<scrollpane id="sp">
    <box structure for central slave/>
</scrollpane>

var northSlave = xwt.newBox().apply("northSlave");
$sp.northSlave = northSlave;

You can put whatever you want, and as complex as you want in the separate 
template (or later, loaded from the web or a string) and you eliminate the 
problem of altering the box's child count.

Remember that having it a property means that it works either way, too.  :-)

> Actually, I'd like to see the layout changed before you guys drop into
> serious bug smushing mode. Features are ok, but if we migrate to a
> different layout, there will be bugs introduced, and there will be
> regression bugs in layouts to deal with.

That's the scary (wonderful) part -- Charlie and Rod have been quite careful 
(exceedingly so) to not move on to another widget until the one they are 
working on is complete and working.  The widgets all use common traps and 
their behaviour seems to be tested heavily before moving on.  There simply 
isn't much to correct from what I have been working with and testing.

Offhand I don't much care whether xwt.standard.[shared|widgets][.lib] is used 
or xwt.standard.impl[.lib] is used.  I _do_, however, want to get one widget 
library completed before even thinking about renaming or rethinking a layout.  
From my perspective XWT needs *A* widget library and *A* working widget demo.  
Not what we have now, what Charlie and Rod have been doing, and now what you 
and Adam would like to move to, all at various stages of completeness.

Get one of these completed and once it's done and we have something that a 
random person finding XWT can use without cursing about incomplete widgets, 
THEN let's worry about whether xwt.standard.widgets is what we want or if 
xwt.standard.impl is the cleaner/more sane method.  I personally don't want 
to move Rod away from the work he's done to just back track and change the 
directory layout (and fix whatever bugs get introduced in the process).  I 
(from Benshaw's perspective) want a complete widget set.  Any complete widget 
set.

Charlie claims 60% completion.  From what I'm seeing (strip out the more 
esoteric widgets like table and tree) and I'm thinking 80%.  I don't want to 
backtrack just to make directory structure changes, as we already did that 
back at 5% completion.  When the widget set is done and passes 90% of the 
tests, no problem, as I at least have something I can give to others to work 
with while the changeover is taking place..

The _nice_ thing about any changeover is that it should not affect any 
applications -- Anyone writing an app which imports anything OTHER than 
xwt.standard for standard widgets is a broken app.   Everything we're writing 
only imports xwt.standard and maybe com.benshaw for our specific widgets.  
That makes me very happy.  :-)
 
> I've put aside my Sunday (GMT+10:00) for doing this move. Would you be
> up for it, if I migrated all the code you've written, and then we all
> work on cvs.xwt.org improving the widget set together? I think just
> about everyone's got keys now, and we've got a 'revamp' branch to do
> all the work under.

Rod and I would not be available (GMT-5:00), but using cvs.xwt.org brings up 
another issue -- having to wait for confirmation from you or Adam before 
checking in.  The widget audit has gone very swiftly because we were not 
worried about waiting for someone in another timezone to okay a checkin...  
that and we were all working on different parts of the widget set so 
conflicts were few and far between.

I fully understand the use of cvs.xwt.org for everything once a widget set is 
completed -- you do not want to introduce bugs into a working system -- but 
during fast development it's going to be a pain.

Regards,
Andrew

_______________________________________________
http://lists.xwt.org/listinfo/dev