I gave it a go at writing the code myself, but I got lost, I can't find
where the event is actually handled. To trace execution I'm running
eric.py (eric-snapshot-20050917) in eric (Debian testing). <br>
<br>
Anyway, here are my points. <br>
<ul>
<li>
<p> Replace it with IPython?</p>
</li>
</ul>
IPython rocks. Awesome shell, very interactive and it provides a bunch
of profiles that can be very handy. <a href="http://ipython.scipy.org/">http://ipython.scipy.org/</a><br>
<ul>
<li><p> Can't paste more than one line</p></li>
</ul>
Shame. Makes things harder than they should be<br>
<ul>
<li><p> Can't select with shift+<arrow keys></p></li>
</ul>
Annoying. Eric just moves over without selecting anything.<br>
<ul>
<li><p> Middle click paste or typing should move focus to the current prompt</p></li>
</ul>
Instead of just overwriting the history. I don't mean the actual history, I mean what's visible in the shell.<br>
<ul>
<li><p> Ctrl-c should kill the current line</p></li>
</ul>
Instead it does nothing.<br>
<ul>
<li>
<p>Can't alter values in the code that's currently running.</p>
</li>
</ul>
test.py:<br>
-----------------------------------<br>
h = 5<br>
print 'break here'<br>
-----------------------------------<br>
Now fire up eric, run the debugger and break at the second line. Try this in the shell:<br>
>>> h<br>
5<br>
>>> h = 6<br>
>>> h<br>
6<br>
<ul>
<li><p> <arrow button up> doesn't match partials like my readline settings say to.</p></li>
</ul>
So eric doesn't use readline. Already knew that. But still, the functionality is staggeringly useful. Use case:<br>
>>> [x / 2.0 for x in range(10) if x % 2 == 0]<br>
[0.0, 1.0, 2.0, 3.0, 4.0]<br>
>>> h<br>
6<br>
>>> [<up arrow> <br>
results in <br>
>>> h<br>
<br>
That's frustrating. Yes, this case was simple. In real usage I have
lots in my history. Being able to search over some subset rather than
the whole history makes things a lot faster. Especially when you're
repeating some long sequence of actions to get some code to work right.
Yes, you could define a function, but that requires that you expect it
to not work. Also doesn't help when you're dealing with long obscure
lines that you just need to change a little bit.<br>
<br>
<br>
<br>
Again, I'm willing to help implement these features and fixes. But I
need help, I don't see how to hack the shell code yet. A standalone
version of the shell would be very useful (an app with only the
shell) as it would cut down on the depth and breadth of the problem. I
tried, but I'm not familiar enough with pykde and friends to get it
done. <br>
<br>
<br>
Thank you,<br>
Joshua C. Gilbert.<br>
<br>