Omniscient Debugging (ODB) With Bil Lewis

Bil Lewis published an influential paper in 2003 called "Debugging Backwards in Time". It inspired me and others to build time traveling debuggers of their own. The debugger he built was called ODB, or the Omniscient debugger, implemented for Java. I was very lucky to get to chat with him over zoom, ask him some questions, and even have him demo the ODB to me. We had an amicable chat. The source code of ODB can be found at https://github.com/OmniscientDebugger/LewisOmniscientDebugger, and his paper can be found at http://www.cs.kent.edu/~farrell/mc08/lectures/progs/pthreads/Lewis-Berg/odb/AADEBUG_Mar_03.pdf. Also, he once did a talk at Google which was recorded and published here: https://www.youtube.com/watch?v=xpI8hIgOyko.

Transcript

The following transcript was automatically generated by an algorithm.

00:00 : that should make it a little bit easier
00:01 : to see
00:03 : they're very interesting posters you are
00:06 : into
00:07 : you are interested in going backwards in
00:10 : time and debugging and understanding how
00:12 : programs work cool well yeah i like to
00:16 : um enable people to do that
00:20 : so um are you um
00:23 : maybe just like are you still uh
00:26 : teaching
00:27 : as a professor how how is your schedule
00:30 : like these days
00:32 : so i am now uh
00:36 : retired and i am teaching over
00:39 : zoom so i'm actually teaching a high
00:42 : school class and computer science
00:44 : over zoom right now so oh yeah
00:47 : that that's awesome nice yeah
00:51 : so um yeah i actually think
00:54 : this technology is time traveling
00:57 : technology
00:59 : can be have a lot of potential in
01:02 : education
01:02 : in fact um i am
01:06 : uh like i spend some time
01:09 : teaching as an instructor at a coding
01:12 : boot camp
01:13 : um a couple years ago and
01:17 : and basically um
01:21 : sort of learning how the students
01:24 : learn and i like learning
01:27 : like trying a lot of different ways of
01:29 : teaching the students
01:31 : um kind of tailored to what they need
01:34 : and one of the things
01:38 : i i found that debugging to be maybe the
01:41 : hardest things
01:42 : to teach a summer
01:45 : overall and um and
01:48 : since i stopped teaching at a full-time
01:52 : basis
01:53 : i started mentoring
01:56 : at a software startup where i would work
01:59 : with
02:00 : developers that are
02:04 : mostly more junior than i am and i will
02:07 : pair a program with them
02:09 : or you know have design discussions with
02:12 : them
02:13 : um and again i'm
02:16 : i'm in the role of teaching them and a
02:18 : lot of times
02:19 : i'm trying to teach them debugging
02:22 : because
02:22 : a lot of times they would have these
02:24 : really tough problems
02:26 : these books they can't track down and i
02:29 : would work
02:29 : side by side with them trying to kind of
02:32 : find
02:34 : and again i found that well debugging is
02:38 : maybe the hardest thing to teach of of
02:40 : all of
02:41 : software so so yeah that's something i'm
02:45 : thinking about
02:46 : and that's sort of largely why i i'm
02:49 : going down this
02:50 : path chat like going for this endeavor
02:52 : um
02:53 : yeah um wonderful
02:57 : would you mind like uh so like if
03:00 : there's people watching later
03:01 : would you mind defining the term
03:04 : omniscient debugging for people
03:07 : what do you mean by that certainly
03:11 : so uh in
03:15 : 2001 or whatever it was
03:19 : um i had free time so i said oh i am
03:22 : going to do this debugger that i've
03:24 : always dreamed about
03:25 : and now that i have java it is really
03:28 : easy to do
03:29 : because of the jvm and i just whacked
03:32 : out there and did it oh my god i came up
03:34 : with this thing and it worked not very
03:36 : well but it worked
03:38 : and then i started doing research on
03:41 : other people who worked in the area and
03:44 : um
03:46 : i wanted a name and i said okay you know
03:49 : back in time debugging yeah
03:51 : okay it's been used it doesn't sound
03:53 : very good and
03:54 : i came up with the concept of
03:58 : when the debugger knows everything
04:02 : what do you want it to tell you so the
04:04 : debugger knows
04:06 : everything the burglar is omniscient it
04:08 : knows everything your challenge is to
04:10 : ask it the right question
04:12 : anyway so that's where i came up with
04:14 : the term so that was fun
04:16 : and uh makes me feel good nice
04:19 : nice yeah i i think you you kind of
04:22 : i saw the uh the youtube party
04:26 : or or the zoom party you did with with
04:29 : some
04:29 : other some other folks that are working
04:32 : in this area or are interested in
04:34 : in this technology and uh yep
04:38 : you're definitely one of the pioneers in
04:41 : this space and they
04:42 : they also know your paper probably or
04:45 : have watched your video
04:48 : um what i'm going to claim
04:52 : is my major contribution
04:55 : it's not just you know anybody can do
04:58 : bytecode instrumentation all that stuff
05:03 : i feel that my contribution was really
05:08 : how to use it in other words the idea of
05:11 : okay
05:11 : what is it that i want to know how do i
05:14 : want to see my
05:15 : information displayed how do i find my
05:18 : bug
05:19 : by going backwards in time so i feel
05:21 : that that is my real contribution
05:23 : yeah i agree yeah a lot of i've seen a
05:26 : lot of papers
05:27 : that uh focus on the recording and
05:30 : replaying
05:31 : and um that aspect of it
05:35 : whereas your focus well
05:39 : not the only focus but a big part of it
05:42 : that's different from
05:43 : theirs is you got this ui developed
05:47 : um that allows you to do this um
05:51 : and actually i i did i'm i'm gonna
05:55 : show this uh
05:59 : let me share my screen if i
06:04 : how do i do this um a green thing down
06:08 : the bottom share screen
06:09 : oh yeah you have a green thing at the
06:11 : bottom yes i do have the green thing i
06:13 : found it
06:14 : okay so oh there's a dell
06:17 : okay there we go uh yeah
06:20 : whenever whenever my internet is down i
06:23 : type hello
06:24 : in my browser and usually adele comes up
06:28 : um before adult got popular it was
06:33 : lionel richie
06:33 : i think but uh anyway um
06:37 : okay so can you see my screen all right
06:40 : yeah yeah okay yeah so i
06:44 : uh i actually got the ammunition
06:47 : debugger
06:48 : working last night and i was very
06:50 : excited
06:51 : because i i had some trouble getting it
06:53 : set up previously
06:55 : but um i got it working and i did
06:59 : a code challenge on the lead code
07:01 : website
07:02 : uh which was this one okay
07:05 : uh i did this one which was like a
07:08 : remove the
07:09 : nth node from the end of a linked list
07:13 : so yeah so i did the code
07:17 : and i ran it i i'll show
07:20 : for people watching i'll show how you
07:22 : run the debugger
07:24 : so first first i would compile my
07:27 : program
07:29 : uh like so uh java c
07:33 : remove click remove
07:37 : node.java and then uh
07:40 : oh there's a debug script
07:44 : that i adapted from what you gave in
07:47 : your source
07:49 : bundle i can say after compiling the
07:52 : program
07:55 : you say debug the name of the program
07:59 : and it'll run it'll run
08:02 : with with the
08:06 : omniscient debugger
08:09 : it'll execute the debugger class and
08:12 : passing in
08:13 : my program as the argument of the
08:16 : debugger
08:18 : okay and this this is a version of the
08:20 : program that has a bug in it
08:25 : and uh so what i what so
08:30 : there's this stack trace which i really
08:32 : found interesting
08:34 : um because you can look at this
08:38 : this actually whereas most of the stack
08:41 : traces
08:43 : that you normally see in programming
08:46 : it it only goes down
08:49 : uh it only goes inwards
08:53 : because right it it it's like a
08:56 : coastline
08:57 : that goes it indents indents and dense
09:01 : and dense
09:02 : and that's it because that's probably
09:05 : the place you threw an
09:06 : error or something this one
09:10 : it goes in and then goes back out
09:14 : um and so this is like
09:19 : you can think of this as like the
09:20 : timeline of your program
09:22 : would you say that's right
09:26 : so that is the entire
09:30 : call history so you can just imagine it
09:33 : as printing out
09:34 : a line uh at the beginning of each
09:37 : function and a line at the end of each
09:39 : function saying i came in and i came out
09:42 : and it gets deeper and deeper as you go
09:44 : in so that's
09:45 : that's the entire program right it's
09:47 : like the outline
09:48 : of the whole program yeah uh yeah so so
09:52 : here
09:52 : that's the call history um or the
09:55 : outline
09:56 : and you can click through here and then
09:59 : in this code panel
10:01 : it jumps to the place
10:04 : um i guess it jumps to the beginning of
10:07 : that of the thing that i selected
10:11 : and then here you have these uh step
10:16 : buttons
10:18 : that you can select and you can see the
10:21 : values of the
10:22 : variables um
10:25 : these locals the names are like r0
10:29 : r1 var three var four or five but they
10:33 : map back
10:33 : to uh variables in the current context
10:38 : so like rx0 is head
10:41 : and arc one is n here
10:46 : um okay that is um
10:50 : i have no idea why the names aren't
10:51 : coming in but normally the names will be
10:53 : there
10:54 : oh i see okay so there's some weirdness
10:57 : okay about my environment so so yeah so
11:02 : so normally this would just have the
11:05 : names here as
11:06 : for the locals and then when you click
11:09 : on
11:09 : when there's an object you see this
11:12 : notation
11:14 : of the object with these angular
11:16 : brackets
11:18 : and then you have a sort of like a id
11:22 : the object id and then if you double
11:26 : click on it you get to see the details
11:28 : of the objects
11:29 : on the right over here and then this is
11:32 : a linked list
11:34 : so it's kind of nested so next
11:37 : is another linked list and then you
11:39 : drill down is another linked list
11:42 : so yeah so this is
11:45 : i was able to use this debugger to um
11:48 : to solve all my bugs one one of them was
11:51 : a little tricky
11:52 : or a couple of them were a little tricky
11:55 : um
11:56 : and in this case if you drill if
11:59 : i think you can say click on here or
12:02 : click on here it goes to the
12:04 : end of the program i'm already on the
12:06 : last line
12:07 : so you're at the end yeah i'm at the end
12:10 : okay oh it threw me a little bit here
12:13 : it says um i got a null pointer
12:16 : exception at the end
12:17 : and i step backwards and i
12:20 : i don't see the line that was throwing
12:22 : the error instead
12:24 : it was the second to the last line um
12:27 : before it's thrown uh let's see
12:30 : um so so so i am going to guess
12:35 : what is the value of place to remove
12:38 : what is the value of that variable place
12:41 : to remove
12:42 : okay yeah um place to remove
12:45 : is var three
12:48 : so it's a linked list so it's it's next
12:52 : field is
12:52 : null and ah
12:56 : so i assume that's that is your uh
12:59 : where you're throwing the exception
13:01 : right there yeah i think it
13:03 : it when i plus plus and then
13:07 : and then because current node is
13:10 : probably no
13:14 : [Music]
13:16 : well anyway um why don't i
13:21 : bring up the debugger and uh
13:24 : give you a little tour as i'm doing it
13:26 : oh yeah that will be
13:27 : that will be perfect okay i'm gonna stop
13:30 : my
13:31 : uh share and now i can let you do it
13:35 : okay okay can you see the full screen
13:38 : real
13:39 : real well yeah i can
13:43 : great so um what we're looking at is we
13:47 : start
13:48 : in maine so there is maine we are at
13:51 : time right up here you can see where
13:53 : your event
13:54 : number two out of 1273 events we are on
13:58 : lines
13:59 : 42 of the file demo.java
14:03 : so that is line 42 of demo.java
14:07 : as soon as i enter a an instrumented
14:12 : method such as main i immediate record
14:16 : i entered main and here are the
14:18 : arguments there is an array of strings
14:22 : i'll double click on it it comes over
14:24 : here it's an array of zero string
14:28 : things so it's kind of boring okay um
14:32 : we can step through the code here here
14:34 : is the uh
14:35 : step into function
14:38 : and here is the step over function yeah
14:42 : and
14:42 : backwards over so we haven't gone into
14:45 : anything yet we will in a moment let's
14:46 : step
14:47 : forward forward forward and now we will
14:49 : step into
14:51 : the bad method let's step backwards
14:54 : out of the bad method and now we will
14:57 : step
14:58 : over the bad method and we
15:01 : pop down to the catch so that is what we
15:03 : expected so we will jump
15:04 : over the bad method again and up here we
15:08 : made a vector
15:08 : v is about to be a vector where is
15:12 : v v is oh v
15:15 : is right there v is
15:19 : a uh what are we wearing a static method
15:22 : so v
15:22 : is a static method a
15:26 : static variable it happened we just made
15:29 : a vector
15:29 : there my vector zero double click on
15:32 : that
15:34 : and my vector zero currently is empty
15:37 : you can see they're all
15:38 : empty oh okay but
15:42 : does that mean like it's allocated
15:45 : like space for all those things but it
15:48 : hasn't filled it up
15:51 : so what it means is that
15:54 : i have run the program to completion
15:58 : so i know the maximum number of things
16:02 : that were ever in the vector
16:03 : oh i see and so i display that
16:07 : that way because i don't want the vector
16:09 : going jumping around
16:11 : this way i can just say right here i can
16:14 : say
16:15 : show the first change to the vector the
16:17 : next value ah
16:19 : we added element zero show the next
16:21 : change to the
16:22 : vector we added we're adding bottles of
16:25 : beer on the wall
16:27 : i see okay and removing bottles of beer
16:29 : from the wall
16:30 : yeah but the idea is and so this is a
16:34 : hack
16:34 : i hacked this so that it would look good
16:37 : for the human being
16:38 : so it doesn't jump around so you see
16:41 : especially for the case of vectors
16:43 : you you handle it this way yes vectors
16:46 : and hash tables let's
16:48 : look at a hash table here the hash table
16:49 : over here
16:51 : so this is not what a hash table is
16:54 : actually
16:54 : implemented like but it shows what's
16:58 : going
16:59 : on and that's what i wanted to do i
17:00 : think the human being
17:02 : can read this a lot better and when you
17:05 : click right right right when you click
17:08 : the forward and backward button
17:10 : on the objects panel does it go to the
17:14 : next update
17:15 : okay so you select some object in this
17:18 : table
17:19 : and then you can forward and backward to
17:21 : see the next
17:23 : change of this object or the previous
17:25 : change of the object
17:27 : so this is go to zero state
17:36 : i got it okay of course it says previous
17:39 : previous previous previous the star
17:41 : indicates where you were most recently
17:44 : changed
17:44 : and you can also just look at
17:46 : individuals so i can go
17:49 : show me only the changes to four
17:53 : and i can also go over here to objects
17:58 : select instance value variable
18:01 : here are all the values of that instance
18:04 : variable
18:05 : it had four beers and it had no more
18:07 : beer
18:10 : and so i can look at it right there
18:13 : that's awesome yeah that that's a that's
18:15 : a capability that
18:17 : like we've never had basically
18:20 : just to see what what what's the next
18:23 : modification of this
18:24 : object looks like it's that's really
18:27 : amazing
18:29 : yeah so the classic thing is oh i
18:31 : divided
18:32 : by zero i took y over x and x was zero
18:37 : how did x get to be zero let's step
18:40 : backwards time and
18:41 : you set x to zero yeah exactly classic
18:44 : yes and also you jump the execution
18:48 : so to speak back to the line of the
18:50 : modification as well
18:52 : and that's really powerful yep and let's
18:55 : see so there we are
18:57 : filling up a hash table that's kind of
18:58 : boring and as we move down here
19:02 : oh we're just printing things out there
19:03 : we are taking adding things to a vector
19:07 : okay that's kind of boring but let's go
19:10 : over here because there are some other
19:11 : threads we can look at the other threads
19:13 : so now we are in a different thread
19:15 : we're in the sorter thread
19:17 : and the sorter thread is doing a quick
19:19 : sort
19:21 : so it's very anyway
19:24 : um there we go short uh elements 11
19:27 : through 9
19:28 : and we're gonna split them off so there
19:30 : we are splitting off a
19:31 : new thread to do
19:35 : the um quick sort for 11 through 15.
19:38 : and so there we can see the different
19:40 : threads coming up
19:42 : yeah and that thread oh that thread is
19:45 : already dead he's finished whatever he's
19:47 : doing
19:48 : i don't know he he started through 13.
19:51 : it did not take
19:52 : very much time and so he exited so
19:56 : there we see the threads running the one
19:58 : um
19:59 : so one how does this debugger
20:02 : work with uh different threats like will
20:07 : there
20:07 : be problems with race conditions and so
20:10 : forth would that cause
20:12 : problems for the debugger
20:15 : no as you can see these threads are all
20:17 : running in parallel
20:19 : and what i actually do is for
20:22 : every time i record information
20:27 : um i take a global lock
20:31 : so the threads are there therefore
20:33 : forced
20:34 : to be in order so i'll know
20:37 : uh i can't i won't have any um
20:41 : uh multi-threading problems okay
20:44 : don't slow that down and some problems
20:46 : will not occur
20:48 : because of this i see so so this would
20:51 : be like whenever
20:52 : whenever you log information that would
20:54 : be whenever there's a function call
20:56 : or whenever there's a modification to a
21:00 : data structure
21:01 : or whenever you're assigning
21:04 : a variable assigning a value to a
21:08 : variable somewhere
21:09 : that happens you got it
21:13 : okay nice yeah this is
21:16 : this is like incredible i i i started
21:19 : working on
21:20 : my python version of the debugger but
21:22 : there's a whole bunch of capability that
21:24 : this has
21:26 : that i haven't gotten around to yet um
21:29 : anyway
21:29 : um let's see what else is really worth
21:32 : looking at
21:33 : um as you can see i just did tty
21:36 : output i instrumented uh print line
21:40 : so if you execute print line i will
21:44 : collect it if you want in only print
21:47 : line if you do any other weight of
21:49 : output i won't see it
21:51 : yeah yeah so you can you can click on
21:54 : a out print line and it jumps to the
21:57 : line
21:58 : that prints the line right so everything
22:02 : displays according to this event time up
22:05 : here whatever
22:06 : event we're at the entire debugger shows
22:09 : that
22:10 : point in time i see nice
22:14 : um yeah very cool very cool demo
22:18 : yeah this i'm sure um
22:21 : so how did you uh like you said
22:24 : you said that like in 2001 you
22:27 : you just decided when you had some time
22:31 : you finally decided to work on this idea
22:34 : how did you come up with this idea
22:36 : to start with
22:40 : um g wilkers i don't know i was
22:44 : working in artificial intelligence when
22:47 : i first
22:48 : graduated from college and into uh
22:52 : i went to stanford research institute
22:54 : and at some point during that time when
22:56 : i was working with lisp
22:59 : and lisp had a wonderful uh ide
23:03 : environment i mean that's where all of
23:05 : our environments came from
23:07 : and um i just came up with that thought
23:10 : like oh
23:10 : i could do that and i could i sort of
23:13 : played with it and
23:14 : did a little bit of something but i
23:17 : didn't get anywhere it says
23:18 : i had work to do instead but um
23:22 : one of the other guys at sri actually it
23:25 : turns out
23:26 : he had worked on the very first one
23:30 : uh e dams i think it is back
23:33 : in 60 something it didn't really do
23:36 : anything it was sort of an
23:38 : idea and it wasn't practical or anything
23:40 : but
23:42 : i talked to him years years and years
23:44 : later that's what was really cool so
23:46 : um bowls i want to say
23:50 : um so i uh so i
23:54 : have been working on a similar thing uh
23:57 : in python as i told you and uh one
24:00 : one problem i ran into in the past is
24:04 : like i i'm having to save so much
24:07 : information
24:08 : in the process's memory that um
24:13 : at one point i was doing it in the
24:15 : browser at one point i had this
24:18 : toy programming language that would
24:21 : generate
24:22 : javascript to be executed in the browser
24:25 : and um and because i had
24:28 : the historical states of all the objects
24:32 : stored away somewhere in an array
24:36 : that was just too much for the browser
24:39 : the browser
24:39 : didn't want to give me any more memory
24:41 : and it just completely hung
24:43 : have you ran into like problems like
24:46 : that
24:47 : where you were you were simply using too
24:49 : much memory
24:52 : well of course there is
24:55 : only a finite amount of memory available
24:59 : and there's only a finite amount of time
25:01 : that i'm willing to wait for it
25:02 : anyway so yes um
25:09 : total amount of memory used is an issue
25:12 : if you look at my code my
25:15 : storage of information is
25:18 : as tight as i can possibly make it
25:22 : um it is
25:27 : each change consists only of
25:30 : its time and its new value
25:33 : um all of the information
25:36 : is a single word i do bit packing in
25:39 : words
25:40 : uh i've got 32 bits in my array
25:44 : that i bit pack to get as much
25:46 : information as i possibly can
25:48 : so um even there
25:53 : what did i say i can only collect uh you
25:56 : know
25:57 : a few million events or something
26:01 : i i i saw that you had like um
26:04 : upper limit with 32-bit computers
26:08 : but just because of the address space
26:12 : but then but then when we had 64-bit
26:14 : computers
26:15 : that kind of became an issue right
26:19 : yeah i mean uh ostensibly it will grow
26:22 : forever it's just
26:22 : the question of how long do you want to
26:24 : wait
26:27 : i see um uh and um
26:32 : let's see how how does your i kind of
26:35 : want to
26:36 : ask like how does your scheme for saving
26:39 : the historical states of
26:41 : all the objects how does that work
26:46 : okay so i do not persist anything okay
26:50 : all in memory
26:51 : only okay and so
26:55 : um what i do is every time an
26:59 : object so i come in
27:02 : i instrument your code as you
27:06 : load your code well you call debug
27:09 : on your program i have the
27:12 : um a loader i wrote a loader my loader
27:16 : loads your code
27:17 : instruments it as it comes in so i'm
27:19 : instrumenting your byte code
27:23 : and then every time i see an object
27:26 : every time the byte code
27:28 : registers a new object then i re
27:32 : build a shadow object around it or
27:35 : parallel to it it hashes to it so
27:37 : i have a parallel shadow object and in
27:39 : the shadow object i retain
27:41 : the historical information about all of
27:44 : the instance variables
27:46 : okay so it's sort of like are those
27:50 : like you have a shadow object and then
27:53 : i i think in your paper you wrote for
27:56 : every instant variable you have
27:58 : you have this history list which
28:01 : contains
28:02 : all the values of that instance variable
28:06 : and and then i assume like for for
28:09 : atomic values like numbers and
28:12 : maybe strings i don't know if you i
28:14 : don't know if you
28:15 : do strings as atomics then it'll just be
28:19 : the
28:19 : history for this is sort of like an
28:22 : array that contains all those numbers or
28:25 : all those
28:25 : arrays for for objects i imagine
28:29 : it's just like a representation
28:32 : of the object like a pointer to the
28:34 : object is that right like you just have
28:37 : an
28:38 : array of the um
28:42 : like if that if you if you assign a
28:45 : different object to that
28:47 : variable for example i guess you would
28:49 : just have
28:50 : a like at this point in time it was
28:52 : point this variable was pointing to
28:54 : object a and then later point in time
28:57 : the variable was pointing to object b
29:00 : is that accurate um
29:04 : i think so i think we're saying the
29:07 : right thing here
29:08 : okay um
29:11 : just a moment let me see if i can
29:14 : pull this off um we have the debugger
29:18 : debugger
29:20 : nice okay what i wanted to show you was
29:23 : that when i
29:24 : you can see emacs there in the
29:26 : background
29:28 : and when i go to front of line
29:33 : oh my good look at that there is a
29:36 : exception in the debugger command oh
29:40 : things are going terribly this is a bad
29:41 : thing we go back to here
29:44 : to the debugger debugger we have we're
29:46 : going to bring
29:48 : up the debugger we stop recording
29:51 : and oh we do not have
29:54 : i was going to show you recorded data i
29:57 : did not
29:58 : have um the debugger debugger
30:01 : working okay no worries
30:05 : that's okay well anyway it's not not up
30:07 : there so i was actually going to
30:09 : take you through all of the debugger and
30:12 : show you the pieces of it
30:14 : i see that's all right um
30:17 : that's what that would be a really cool
30:19 : trick though
30:21 : um i will do that i will get that
30:24 : working for you later
30:26 : well for me too i wonder what i did
30:28 : wrong okay
30:29 : yeah maybe we have another meeting late
30:32 : at some point
30:33 : that that would be really cool to see um
30:37 : so one thing you said on your website
30:39 : i'm i'm
30:40 : gonna oh actually i wanna read you this
30:43 : quote
30:43 : from your paper and uh and i get you to
30:46 : comment on it
30:47 : okay so is that notably
30:50 : an informal survey a show of hands in
30:54 : conferences classes
30:55 : etc of several thousand
30:58 : programmers reveals that roughly ninety
31:01 : percent of all programmers debug their
31:03 : programs strictly by doing
31:06 : by the use of print line uh this
31:09 : i take to be a horrendous comment on the
31:12 : state of affairs
31:16 : why why do you think um why do you think
31:19 : programmers use most programmers
31:23 : use print statements to debug their
31:25 : programs and
31:27 : what do you think and could you
31:29 : elaborate on why do you think that's a
31:31 : horrendous
31:32 : comment on the state of affairs
31:41 : bless you i'm presenting i'm going to
31:43 : different companies talking to people
31:44 : all the time
31:46 : and so i asked that question i was
31:48 : amazed
31:49 : how few people used the bugger said we
31:51 : have and
31:53 : um even our own
31:56 : gosling who our gothing
31:59 : who wrote you know um
32:02 : java yeah he uses brickline statements
32:07 : extensively and i was like you're
32:09 : kidding
32:11 : um i was and
32:14 : he was amused
32:19 : to see the debate the debugger when i
32:22 : just uh
32:23 : uh demoed it for him but he wasn't real
32:26 : interested he was sort of like okay
32:28 : that's cute
32:33 : yeah anyway so now i'm i am horrified
32:36 : that people are writing in
32:38 : uh print lines
32:41 : for all the usual reasons which is
32:45 : there are better ways to do it like this
32:48 : and
32:49 : even a regular debugger i mean
32:53 : i do print line debugging too of course
32:55 : i have
32:57 : but um oh it's a
33:00 : why do you think that developers do that
33:04 : instead of using a debugger because
33:06 : they've always done it that way and the
33:07 : bugs are scary debuggers are scary
33:11 : they've never had a good interaction
33:14 : maybe
33:15 : i'm guessing yeah i have heard that
33:18 : uh sort of comment of like oh yeah
33:21 : that's
33:22 : debuggers are something that advanced
33:25 : people use
33:26 : and i'm i'm not good enough so i just do
33:29 : print statements
33:30 : which which is an interesting comment i
33:33 : think and
33:34 : perhaps perhaps
33:37 : you know debuggers need to be more
33:40 : approachable
33:40 : or we need to give better training to
33:43 : people
33:44 : about the debugger or what like
33:48 : come to think of it i i haven't seen
33:52 : any classes that would specifically
33:55 : teach you to use the debugger
33:57 : like like in my college days i don't
34:01 : remember
34:01 : any professor teaching me how to use the
34:04 : debugger
34:06 : i don't know they might have talked
34:08 : about it in passing
34:10 : but it's certainly not something that
34:13 : was given a lot of thought
34:15 : and i wonder in the industry so like
34:18 : nowadays with
34:19 : the way the industry is a lot of people
34:22 : come into the industry
34:24 : not going through the college education
34:27 : and like how do they get your how do
34:29 : they get their education
34:31 : about how to code um
34:34 : and where does debugging or using a
34:36 : debugger fit in
34:38 : in that have people been trained to use
34:40 : this
34:41 : and um you know i don't know i started
34:44 : working in writing assembly language
34:47 : programs
34:48 : and by god you you need a debugger
34:51 : there's no
34:52 : question about it and so i didn't even
34:55 : you know for me it was just automatic
34:57 : and then i went into artificial
34:59 : intelligence working in list
35:01 : and lisp has a beautiful system
35:04 : for uh dealing with uh exceptions etc
35:08 : and it makes debugging super easy so
35:11 : i just i personally flowed in that
35:14 : direction and
35:15 : it was just natural like i get really
35:18 : pissed off
35:20 : at some of the debuggers that are out
35:21 : there yeah yeah
35:23 : it is interesting like i i work most of
35:25 : the time as a
35:27 : web developer like doing javascript on
35:30 : the web page
35:31 : and and we we use chrome
35:34 : as our you know environment and
35:38 : i'm actually surprised at how often
35:42 : i find the chrome debugger to be just a
35:45 : little bit
35:46 : buggy and i'm like dude we have a buggy
35:50 : debugger how
35:51 : how are we supposed to do our work and i
35:53 : i i'm
35:54 : kind of guessing some of the time is
35:56 : like maybe just not
35:58 : enough people are using this thing for
36:00 : for them to put in the effort
36:02 : to make this thing really good yeah it's
36:06 : uh
36:06 : it is a mystery to me but uh there it is
36:09 : yeah once i got the uh admissions
36:12 : debugger
36:14 : working kind of well um
36:17 : i started writing really sloppy code
36:21 : yeah i would not plan that like
36:25 : i would just sit there and say i'm gonna
36:27 : type in code
36:28 : and so that i would have bugs
36:30 : intentionally i would write
36:32 : too fast so i'd have bugs that i could
36:34 : look for yeah and it was great i
36:36 : it was fun oh look at the bug i made and
36:40 : and the great part is when you find the
36:42 : bug there's no
36:43 : maybes about it it is right there you
36:46 : see where it
36:47 : comes from exactly sometimes you know
36:50 : other times when you're debugging
36:51 : oh x is zero okay and that and you get
36:54 : the guess the wrong reason why x is zero
36:56 : going backwards and you yeah it's just a
37:00 : great feeling
37:01 : when you know for certain what the
37:03 : problem is
37:05 : rather than guessing because i see
37:07 : developers
37:08 : guessing before they truly truly know
37:12 : and that's such oh yeah that's such a
37:16 : bad thing yeah
37:19 : that there's like i feel like then
37:21 : they're gonna
37:22 : start writing what i call superstitious
37:26 : code
37:27 : like you don't really need the code but
37:29 : you're trying to protect yourself
37:30 : against
37:31 : something and then you're gonna do that
37:34 : all over the place whether you need it
37:36 : or not
37:37 : and yeah yeah
37:41 : a good state of the fish yeah
37:46 : when we were releasing i don't know
37:49 : sun operating system number
37:52 : who knows what number it was back in
37:56 : 1990
37:57 : or so um we're trying to uh
38:00 : make the operating system the boot
38:03 : sequence faster and the boot sequence
38:06 : was weirdly slow i i
38:08 : personally didn't work on it my friends
38:09 : did and at one point
38:12 : they came in screaming and
38:15 : just like that
38:18 : somebody years ago
38:22 : had some timing problems with bringing
38:25 : up the kernels and you know you have all
38:26 : the demons that have to run in the
38:27 : background
38:29 : and so what he did he just put a 10
38:32 : second
38:33 : delay in the kernel
38:36 : everything just brought up all the team
38:38 : and then just sat there doing nothing
38:39 : for 10 seconds
38:42 : i was like well it worked yeah i mean
38:45 : we we almost we almost did that last
38:48 : week i was debugging something
38:50 : with with a co-worker and
38:53 : you know there was this animation that
38:56 : took
38:57 : 250 milliseconds to complete and if you
39:00 : if you try to like it was like closing
39:04 : a dialogue like when you close the
39:06 : dialogue there's this animation
39:08 : that moves the dialogue away or
39:11 : something
39:12 : and that animation takes 250
39:15 : milliseconds
39:16 : and if you try to close the dialog and
39:19 : then
39:20 : open a dialog open the same dialog
39:23 : immediately before the animation is over
39:26 : there's a bug so there's a well
39:30 : okay we can patch over it by just
39:33 : waiting 200 mil 250
39:37 : seconds if you close it wait a little
39:40 : bit
39:40 : and then open another one that fixes the
39:42 : book
39:44 : guys guys guys you all know
39:47 : this is a stop gap solution right
39:51 : so i i didn't let them stop me
39:57 : yeah yeah yeah okay well it has been
40:00 : wonderful
40:01 : chatting with you i'm i'm i'm i looked
40:04 : at you your your your video i
40:06 : liked it you know and then the musical
40:07 : interviews and all that
40:09 : and you got some you're thinking this is
40:12 : the most wonderful thing in the world so
40:15 : yeah thank you so much i appreciate your
40:17 : time um
40:18 : um would you mind
40:22 : um would you mind uh answering just one
40:25 : more question
40:26 : uh um at all yeah so
40:30 : how have you found like how have people
40:32 : received
40:34 : this debugger you talked about going to
40:36 : conferences
40:37 : and showing it off and and and and
40:40 : people had really good things to say
40:42 : right can you talk a little bit about
40:44 : the like
40:45 : community aspect of this i started
40:49 : doing this in 2000 and
40:52 : i guess it was right and then
40:56 : i started going to conferences through
40:58 : 2002 3-4
41:01 : when i was in europe and i was then uh
41:06 : spent a year over there studying in
41:09 : france i was
41:10 : visiting professor et cetera so
41:13 : i got to talk to lots of people like
41:22 : talk about it a great deal and
41:25 : they love my jokes of course i'm clever
41:27 : i'm funny when i'm talking and all of
41:29 : that
41:30 : and so everybody was like really
41:32 : enthusiastic and
41:34 : nobody wanted to use it i could not
41:37 : convince at one point
41:40 : at the end of doing this i would put up
41:44 : a hundred dollar bill and say i will
41:47 : give you one hundred dollars
41:48 : if you can have a reasonable bug that i
41:50 : cannot find
41:52 : and i did not get takers
41:55 : like i i that that was a mystery to me
41:58 : so ah that's really
42:02 : interesting you you you you like
42:05 : they really like your jokes um could you
42:08 : share
42:08 : the analogy of this nick versus the
42:12 : lizard
42:14 : aha yes my favorite and then
42:18 : we'll close it we'll close it with this
42:20 : last one
42:21 : thank you so much for your time i i want
42:23 : to be respectful
42:25 : thank you so we were talking about how
42:28 : the debugger came up
42:29 : and it did not show the correct variable
42:31 : names yeah
42:33 : i am suspicious you have compiled your
42:36 : code
42:37 : in a with the um crushes you know if one
42:40 : of the compilers will crush out
42:42 : names and just do minimal name in it
42:45 : so it might be a compiler flag
42:49 : that's doing that i'm trying to remember
42:52 : i they're the reason that i put up those
42:54 : artificial names i think that was it
42:57 : some way that you don't get the actual
42:59 : names out
43:01 : anyway um
43:05 : so you were asking you just said just
43:07 : nick the snake and the lizard
43:10 : analogy so yes
43:15 : if there is a snake in the grass
43:18 : and you find the snake's tail
43:22 : you can grab the snake by the tail and
43:24 : drag it
43:25 : up and if you pull long enough you will
43:26 : get the snake's head
43:28 : and of course this is exactly like a bug
43:31 : using the ignition debugger
43:33 : you run the program and it throws an
43:36 : exception and that's a snake's tail
43:39 : so you grab onto the snake's tail you
43:41 : say why did it throw an exception
43:43 : oh because uh f was null okay
43:46 : why was f null and you pulled a snake
43:50 : whereas if you have a
43:54 : standard debugger it is more like a
43:57 : lizard in the grass because you can see
44:00 : the lizard's tail but as soon as you
44:02 : grab the lizards by a tail
44:03 : the lid that drops his tail the lizard
44:06 : runs away
44:07 : and you're left with the lizard's tail
44:09 : and you
44:10 : can't go backwards
44:14 : in time in a standard debugger so
44:18 : there you are yeah you have to try to
44:20 : hit it again
44:22 : i i i i i said one time i said like
44:27 : trying to set up breakpoint is like
44:29 : playing the price is right
44:31 : you have to try to get as close as you
44:33 : can without going over
44:37 : yep and boy we all have lots of
44:40 : experience with
44:41 : step step step oh yeah
44:45 : yeah yeah yeah good times good times
44:49 : okay thank you so much bill for your
44:51 : time i i really
44:53 : appreciate this conversation and you're
44:55 : showing the debugger
44:56 : to me and uh yeah i i'd love to see if
44:59 : you
45:00 : if you ever can get the debugger
45:04 : debugging the debugger
45:05 : i would love to see that so hit me up if
45:08 : you uh
45:09 : if you want to show that me at any point
45:11 : it will happen
45:13 : alrighty thank you so much and i hope
45:15 : you have a wonderful day rest of
45:17 : your day and you have a wonderful rest
45:20 : of your day too okay
45:21 : bye-bye bye-bye