<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Matt Newell wrote:
<blockquote cite="mid200611161337.32825.newellm@blur.com" type="cite">
<pre wrap="">On Thursday 16 November 2006 13:18, Brent Villalobos wrote:
</pre>
<blockquote type="cite">
<pre wrap="">I'm a PyQt newbie so maybe there's an easier way, but I wrote my own. I
saved out various view attributes, swapped in a new model, and then set the
view to those saved-out variables. It works well and to the end user the
table just updates without losing any context. It's a little annoying to
code, but I was able to save out and set selections, sort order, vertical
positions, column size, and column positions.
The problem with selections is when a new model has a different number or
order of elements. For example, if the first row is selected in your
current model, then you can easily set the first row as selected when you
reload the model. However, the first row in the new model may be a
completely different element. I worked around this by designating a unique
key for each row. So instead of recording which rows are selected, I
recorded which keys were selected. Then I had to search through the new
model and find those keys and select them in the selection model.
-Brent
</pre>
<blockquote type="cite">
<pre wrap="">On 11/16/06, Matt Chambers <a class="moz-txt-link-rfc2396E" href="mailto:chambers@imageworks.com"><chambers@imageworks.com></a> wrote:
</pre>
<blockquote type="cite">
<pre wrap="">Is it possible that using one of the View classes, to reset() the model,
but still maintain any selections the users had on items that were in
the old data, and the new data?
</pre>
</blockquote>
</blockquote>
</blockquote>
<pre wrap=""><!---->Why not just avoid calling reset? If you use beginInsertRows, layoutChanged,
or whatever is needed, then the selection will be maintained automatically.
Matt
</pre>
</blockquote>
Well, every row is updated every refresh, so it seemed easier. It is
easier, if you don't take into account<br>
all the state data gets blown away. I just wrote models to sit on top
of self updating objects I was already using.<br>
They don't know a model is watching them. Seems like my data will have
to know a lot about the model<br>
for this way to work properly, or I have to move all the updating into
the model. I suppose then you loop<br>
through your new data, updating, adding, and removing data? My
datasets can easily have thousands of rows. <br>
Has anyone done this and not lost GUI interactiveness?<br>
<br>
Matt<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
</body>
</html>