Interactive Debugging #10
Conversation
|
|
||
| Much of the activity of programming involves **debugging**, i.e., | ||
| diagnosing and preventing undesired program behavior. | ||
| Wilkes's early realization of this fact was echoed and amplified in a 2002 |
There was a problem hiding this comment.
Results from publications need to be cited. This should be a bib-item, either directly inline here or at the end of this section with a citation.
src/live-programming.rst
Outdated
| Using these terms, debugging may be understood | ||
| more precisely as the collective processes of determining what runtime | ||
| faults led to a runtime failure, determining what software errors | ||
| led to those faults, and modifying the code to prevent the faults from |
There was a problem hiding this comment.
"modifying the code to prevent the faults from occurring", i.e. eliminating the errors?
| faults led to a runtime failure, determining what software errors | ||
| led to those faults, and modifying the code to prevent the faults from | ||
| occurring. A **debugger** is a tool that allows the programmer to observe | ||
| the machine state during program execution, thereby making it possible to |
There was a problem hiding this comment.
Might be worth explicitly connecting this to the concept of "live programming"
src/live-programming.rst
Outdated
| Interactive Debuggers | ||
| --------------------- | ||
|
|
||
| describe history of transition from physical computing |
There was a problem hiding this comment.
Make sure you include an introduction to this topic in the second draft.
src/live-programming.rst
Outdated
|
|
||
| *Algorithmic debugging* (also called *declarative debugging*) is a semi-automatic | ||
| debugging technique in which the debugger automatically generates a series of | ||
| questions to which the programmer's answers guide the search toward the bug. |
There was a problem hiding this comment.
"the debugger guides the programmer toward the bug by asking a series of questions"
| *Algorithmic debugging* (also called *declarative debugging*) is a semi-automatic | ||
| debugging technique in which the debugger automatically generates a series of | ||
| questions to which the programmer's answers guide the search toward the bug. | ||
| The debugger constructs an *execution tree* (ET), a data structure representing a |
There was a problem hiding this comment.
An ET is just one way to do algorithmic debugging.
src/live-programming.rst
Outdated
| Although algorithmic debugging can be applied in any language paradigm, it is most | ||
| suited for declarative languages, e.g., pure functional languages. | ||
| to determine whether an ET node for a pure functional program is correct, the | ||
| programmer need only check that the return value of the corresponding, referentially |
There was a problem hiding this comment.
"referentially transparent" -- you should define this if you want to use it, but I don't think you need to say it at all
|
This is as far as I'm gonna take it tonight. All bib items are completed. There are a couple more paragraphs to be inserted here and there in section intros. I've converted the intro section on early history of debugging into a TODO because I think I'm covering enough here. Debugging history + interactive breakpoints would be its own entire compendium assignment. If you disagree, I'll add something in next iteration. |
src/debugging.bib
Outdated
| author = {Lawrance, Joseph and Bogart, Christopher and Burnett, Margaret and Bellamy, Rachel and Rector, Kyle and Fleming, Scott D.}, | ||
| title = {How Programmers Debug, Revisited: An Information Foraging Theory Perspective}, | ||
| journal = {IEEE Trans. Softw. Eng.}, | ||
| issue_date = {February 2013}, |
There was a problem hiding this comment.
remove the extra information that shows up in the bibliography entries -- authors, title, venue, year, and for journal articles the vol/issue/page numbers are all that is necessary
|
|
||
| Much of the activity of programming involves **debugging**, i.e., | ||
| diagnosing and preventing undesired program behavior. | ||
| Wilkes's early realization of this fact was echoed and amplified in a 2002 |
src/live-programming.rst
Outdated
| Wilkes's early realization of this fact was echoed and amplified in a 2002 | ||
| study of the U.S. software industry, which found that software engineers | ||
| spent an average of 70-80% of their time testing and debugging, with | ||
| the average bug taking 17.4 hours to fix. |
There was a problem hiding this comment.
this is an unexpectedly large number, might need to dig into this / provide some context
There was a problem hiding this comment.
Not a surprising number imo. I definitely spent 2-3 days on several bugs as an engineer. Makes more sense when you consider the time it takes to reproduce the bug, which may be super dependent on a user's data, which you then might have to check out an encrypted copy of on a remote server, debug remotely (which is slow), repeat long repro steps each time you wanna investigate something new, etc.
I took out the mysterious 70-80% figure.
src/live-programming.rst
Outdated
| spent an average of 70-80% of their time testing and debugging, with | ||
| the average bug taking 17.4 hours to fix. | ||
| Thus, debuggers form a crucial component of an effective programming system. | ||
| (needs better finish) |
There was a problem hiding this comment.
make any changes you want to make by the final draft
src/live-programming.rst
Outdated
| over environmental and (minimally assumed) cognitive constraints---in | ||
| particular, without reliance on hypothesized mental structures. | ||
| This paper followed Sjøberg et al.'s theory-building process | ||
| :cite:`sjoberg2008building`: it adapted the constructs of information |
There was a problem hiding this comment.
every paper should have a bib-item
src/live-programming.rst
Outdated
| practical advice to builders of software engineering tools. | ||
| This paper presented a theory of programmer navigation when debugging | ||
| based on information foraging theory. | ||
| Information foraging theory is an instance of rational analysis, a |
There was a problem hiding this comment.
This definition should probably not be in the summary of the lawrence paper -- instead find the paper that they based it on and make it a bib-item
src/live-programming.rst
Outdated
| deleting program statements that do not affect the criterion variables' runtime | ||
| values at the criterion statement. | ||
|
|
||
| (another short paragraph about how program slicing has been generalized, |
src/live-programming.rst
Outdated
| values. | ||
|
|
||
| Despite their useful guarantee of bug diagnosis, algorithmic debuggers have yet to enter | ||
| widespread use. (add note about functional programming research lagging in tooling) |
src/live-programming.rst
Outdated
| :filter: key == 'nilsson1994lazy' | ||
|
|
||
| .. todo:: | ||
| write bib entry |
| an algorithm for slicing ARTs. | ||
|
|
||
| .. note:: | ||
| The debugging interface sketched in :cite:`perera2012explain` may be viewed |
There was a problem hiding this comment.
maybe make a forward reference to algorithmic debugging in the summary for roly's paper above
No description provided.