A disorganized collection of TODOs, bug reports, and friends' suggestions
---

January/Feb 2010 (notes from initial release)


Maybe don't display boilerplate code when classes are being initialized?


Hubert's suggestions for permalinks:
'''
Hubert: 1) It would be cool to have the ability to link directly to an
execution point, so that when I send you a link, you could see the code
I'd written and the point in execution that I want to discuss.

me: good suggestion.  that also requires me to save your code in
something other than a regular POST variable, which i'm not doing right
now.  then i could send links of the following form:

  http://python.pgrind.com/code_id=20#4 (for line 4 of code snippet 20)

Hubert: yup, yeah the problem lies in storing the code, which brings in
a whole can of worms.  but suppose you did, consider a RESTful approach
(which I am drilling into deeply, hence it's on my mind):

http://python.pgrind.com/pg/code/20#4

...quite easy to do and manage in a framework like Django.
'''


Cory McLean: 'Might be nice to be able to automatically jump to particular step
# rather than just inc/decrement and jump to ends, particular if the
number of steps executed is significant.'


Related work:
  - PyWhip: http://pywhip.appspot.com/
    - also powered by Google App Engine
  - Crunchy: http://code.google.com/p/crunchy/

---
2010-09-05

From Hacker News feedback:

- support for raw_input() sounds like something reasonable to include,
  though, since it could simply pop up a text box and ask the user to
  input a string e.g., javascript:alert("hello, " + prompt("what is your
  name?"))

---
2010-09-29

Chas Leichner (chaoslichen@gmail.com) emailed me to show me his related
project:

  http://www.youtube.com/watch?v=ZiS4MZF5eDQ
  http://code.google.com/p/idlecarpentry/

He had an interesting idea of annotating lines of code with special
comments, and those comments would display when those lines were being
executed.  That way, you can have sorta like 'literate programming
tutorials' that intermix code and comments.  I could think about
implementing something similar within my Online Python Tutor.

e.g.,

#> declare a variable
x = 5

#> an if statement tests a condition
if x < 5:
  y = x + 5
else:
  y = x - 5

the appropriate annotations would display when you stepped through the
line of code immediately below each annotation.

---
2011-09-15

- Integrate Peter's Python 3 fork into one unified application, and
  perhaps deploy online on WebFaction, since they offer both Python 2.X
  and Python 3.X on their servers.


---
2011-09-23

- The BACKEND can easily print out a nice list of variables in each
  stack frame (and globals) by order of their APPEARANCE in the original
  program execution ... that way, the front-end doesn't have to "guess"
  an optimal order and can simply use the order that the backend has
  pre-computed.  e.g., list local stack variables by their order of
  definition in the original program execution.

---
2011-09-24

Ron's suggestion:

Ron Yeh - Feedback: It'd be nice if (for longer programs) the text
window would auto scroll to the current line. For your "Philip's
10-minute intro to Python" example, I was stepping through your code but
then the current line went off screen. I had to manually scroll down a
few times.

Philip Guo - great idea! do you know if there's any jQuery or other js
tricks to do this sort of scrolling?

Ron Yeh - Pretty sure I've done this before, a while back. Basicaly you
just need a named div or anchor in your document, and then you can
scroll to it. This page might help:
http://css-tricks.com/snippets/jquery/smooth-scrolling/

(also auto-scroll the program's stdout output as well as the code
display ... that one is easy --- always auto-scroll to BOTTOM!)

---
2011-09-26

A subset of the suggestions from MIT folks (see more in my private notes
file):

- people thought that it's important to draw arrowheads at the end of
  pointers ... try to make them relatively 'clean-looking', though

- what if the user accidentally (or purposefully) creates a HUGE list or
  other data structure?  how will the visualization handle this?  right
  now it might croak.
  - one idea is to LIMIT the size of the returned JSON data from the
    back-end???

- to make it easier to jump to certain points in execution rather than
  tediously single-step through execution.  e.g., could imagine adding a
  scrubber (like for video playback) combined with pointing at line of
  code to jump to the NEXT or PREV time that line of code as executed.

---
2011-09-29

Upgrade to Python 2.7 after moving over to Webfaction, so that I can use
the default standard library "import json"