[Bond-users] Problems with child in Bond 4.0

Colin Close itchka at compuserve.com
Tue Oct 6 15:31:36 EDT 2009


Dru,

I'm on MSN gemmitchka at live dot com. Be happy to chat there. You are exactly 
12 hours ahead of us time wise so if you start work at say 9.00am I'd have 
about a four hour window before I started falling asleep to try and sort this 
out. Basically if we are to do this live it will have to be fairly early in 
your working day which may not be convenient for you however I'll be on line 
from about 10:00am your time looking out.

I have sent tar.gz archive to your personal mail address containing the db 
created with pg_dump -C --format=P. Also contained in the arcive is a 
directory 'forms' put this in your /home/~ . Copy visitmaster.xml and 
bonfrontend.conf to your home as well and it should just work. You may have to 
adjust the servername of course. The apps  quite small but the version will is 
one I have hacked to try to get Bond 4.0 working so some of it will not be 
functional because I have commented out some features to simplify debugging. 
As ever some features are under 'development'.
To get to the misbehaving screen click on 'Visit List Creation'.

The file of interest is called visit_selection.xml if you search for "stdflg" 
within it this is where the group of three radio buttons begins.
The sql that updates the flags actually controls an sql statement that selects 
the source of data to insert in the main list and are not relevant to the 
problem. Try enabling some of the close statements and you will see the 
problem. While you are at try the recordnavigate in the Visits Due window and 
you will see that it is badly broken. At least it is for me. I am running SVN 
of about two days ago.

I am beginning to see what you mean about the changed method I had noticed 
that I seemed to be getting lots more refreshes than before and that indeed 
may be the problem. This particular form has all sorts of refresh paths 
including ones that use notify triggers. I am afraid that you might find some 
parts of it a bit irrational as I am still improving it as my understanding of 
Bond advances.


Best

Colin


On Tuesday 06 October 2009 05:59:39 am Dru Hill wrote:
> Do you have MSN or ICQ or yahoo chat?
>
> It maybe better to talk online.
>
> Be easier if I have a copy of the database and XML to play with to
> generate the errors with.
>
> One major change with version 4 over version 3 is that the tree
> like structure of how drawing and refreshes is done has changed.
>
> Now if you paint/refresh widget X, it paints/refreshes all widgets
> within X. This didn't use to be the old case. So previously
> if you had widgets spread out a bit, now they have to kind of be
> grouped together if they relate in a common container, or have
> individual trigger paint calls to display their contents if they
> are outside the container.
>
> Colin Close wrote:
> > Hi Dru,
> >
> > It is the most bizarre problem but I may have been mistaken in that it
> > was <child> that was broken after a days messing about I seem to have
> > nailed it down to how triggers are behaving when used in the new radio
> > button code. I have a three button radio selector that selects where data
> > will be sourced from for insertion into a list. Two of the buttons open a
> > different data input form, one of these forms is external to the file the
> > other is a <child> within the same form and the third is the form itself
> > i.e it closes either of the other two forms dependant on which is opened.
> > The <triggers> (in their most basic form) simply use the Bond API to open
> > and close the forms hence the following:-
> >
> > 	<trigger event="click" abortonfailure="true">
> > 		<api call="close">
> > 			<widget>std_visits</widget>
> > 		</api>
> > 	</trigger>
> > 	<trigger event="click" abortonfailure="true">
> > 		<api call="close">
> > 			<widget>visits_due</widget>
> > 		</api>
> > 	</trigger>
> >
> > Bear in mind that that each button has a similar trigger statement. On
> > clicking on the appropriate radio button one form should be opened and
> > another closed. Incidenteley in the original code the attribute
> > abortonfailure was set to false.
> > The expected however does not happen it appears that all the triggers for
> > all the buttons get fired sequentially so if the next statement is an
> > <api> close 'the form you are trying to open' the form gets closed as
> > soon as it is opened.
> > I have proved this bug in that I have removed all the "<api> close"
> > statements leaving only the "<api> opens" and clicking any radio button
> > opens all the forms not just the one which was desired.
> >
> > My conclusion, well clearly there is a bug which is probably in the
> > <radio> code since that is what has been changed most recently.
> >
> > Best
> >
> > Colin
> >
> > On Sunday 04 October 2009 09:37:16 pm you wrote:
> >> Could be SVN broken.
> >>
> >> Whats not working about it? is it segfaulting or just not working at
> >> all?
> >>
> >> Colin Close wrote:
> >>> Hi All,
> >>>
> >>> As far as I can tell there seems to be a problem with childwindow in
> >>> Bond-4.0. I have tries both <child> and <childwindow> tags. Previously
> >>> functional code with windows containing children worked in Bond-3.xx
> >>> not longer works. Can anyone confirm whether there has been a syntax
> >>> change or whether SVN is broken for this at the moment. Thanks.
> >>>
> >>> Best
> >>>
> >>> Colin Close
> >>>
> >>> _______________________________________________
> >>> Bond-users mailing list
> >>> Bond-users at lists.hcoop.net
> >>> https://lists.hcoop.net/listinfo/bond-users






More information about the Bond-users mailing list