Terry J. Reedy added the comment:

Please post a 'checkpoint' patch with all changes so far, even though it has 
the 'other globals' problem.

Then look at the UI parts of Debugger.py. It has a Stackviewer class that 
subclasses ScrolledLIst instead of TreeWidget and a Namespaceviewer class 
currently used for both locals and globals.

The Namespace class is a frame containing a label (a labeled frame would work 
as well), a scrollbar, and a canvas containing a frame. The inner frame has a 
two-column grid: labels with names and entry boxes with values.  Currently, the 
entry boxes do nothing other than make values look different from names, as 
editing the entry does nothing. (Ignored entries are a bad UI design.)  Perhaps 
there was once an intention to propagate changes back to the running program. 
This would not apply to post-mortem stack viewing.

Starting with the checkpoint patch, I would like to modify Stackviewer to 
display globals for any item clicked in a separate frame, like Debugger.  This 
would removing the bug in the checkpoint patch and result in a patch that I 
might commit.  It would also be a step toward merging (in a new issue).

One possibility is a tree in a new frame.  Another is to import and reuse 
Debugger.Namespaceviewer.  The issue is whether users should be able to view 
the attributes of value objects (as in Stackviewer) or not (as in Debugger).  I 
previously suggested not, because it makes the interface look busy, but I just 
rediscovered a note based on debugger experience suggesting that it might be a 
good idea.  Lets start with a second tree.

When waiting for me to respond, you could look at #17942, about improving the 
command button part of debugger.

----------

_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue24790>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

Reply via email to