I am writing Ruby in Emacs, but my Emacs skills are actually pretty low. What I can do, is open the project, TDD using M-x rinari-test, or play inferior Ruby in the second window using M-x run-ruby. Now I woul like to start using debugger from StdLib. I am able to summon it from irb by saying:
require 'debug'
Upon which I get into a prompt
(rdb:1)
but there my aptitude ends. I don't even know how to step into a file. Typing 'help' brought a screenful, but it didn't help me to finally start debugging my buggy gem. Online, everybody writes about things such as "rdebug" or "ruby-debug" or whatever which I firstly don't want to use and secondly, being a muggle, I am unable to install on my Debian. Please help!!!
You really need to try reading the output of help in the debugger. It explains the commands nicely.
For instance, for practice, try this at the command-line, not inside an editor/IDE:
ruby -rdebug -e 'p 1'
h
Ruby's debugger will output the help summary:
Debugger help v.-0.002b
Commands
b[reak] [file:|class:]<line|method>
b[reak] [class.]<line|method>
set breakpoint to some position
wat[ch] <expression> set watchpoint to some expression
cat[ch] (<exception>|off) set catchpoint to an exception
b[reak] list breakpoints
cat[ch] show catchpoint
del[ete][ nnn] delete some or all breakpoints
disp[lay] <expression> add expression into display expression list
undisp[lay][ nnn] delete one particular or all display expressions
c[ont] run until program ends or hit breakpoint
s[tep][ nnn] step (into methods) one line or till line nnn
n[ext][ nnn] go over one line or till line nnn
w[here] display frames
f[rame] alias for where
l[ist][ (-|nn-mm)] list program, - lists backwards
nn-mm lists given lines
up[ nn] move to higher frame
down[ nn] move to lower frame
fin[ish] return to outer frame
tr[ace] (on|off) set trace mode of current thread
tr[ace] (on|off) all set trace mode of all threads
q[uit] exit from debugger
v[ar] g[lobal] show global variables
v[ar] l[ocal] show local variables
v[ar] i[nstance] <object> show instance variables of object
v[ar] c[onst] <object> show constants of object
m[ethod] i[nstance] <obj> show methods of object
m[ethod] <class|module> show instance methods of class or module
th[read] l[ist] list all threads
th[read] c[ur[rent]] show current thread
th[read] [sw[itch]] <nnn> switch thread context to nnn
th[read] stop <nnn> stop thread nnn
th[read] resume <nnn> resume thread nnn
p expression evaluate expression and print its value
h[elp] print this help
<everything else> evaluate
The important commands to start with are s, n, c and b, and q.
s steps into a method.
n steps over a method.
c number runs (continue) until you reach line number.
b number sets a breakpoint on line number. After setting your breakpoints use c to continue running until that line is executed.
q exits the debugger.
Personally, I use the debugger gem. Others use PRY, which is similar to IRB, but with debugger-like extensions.
Knowing how to use a debugger is a good skill. There are problems you can trace down quickly using a debugger, that will take longer trying to use puts statements, because you can see what a variable contains interactively, or loop conditionally until a variable contains a certain value.
Related
When, during a pry debug session, I want to inspect step-by-step execution of FooClass.new.foo I'd do this in pry console
$ FooClass.new.foo #this gives me path and line of the method
break /path/to/foo_class.rb:LINE_WHERE_FOO_IS_DEFINED
FooClass.new.foo
This works but I need to look for the path, line and it leaves a breakpoint I sometimes have to remove.
There's a faster way:
break FooClass#foo
FooClass.new.foo
but it's still two steps, and the breakpoint stays.
Is there a way to do it in one command, like
step-into FooClass.new.foo
which would start a Pry subsession, enter the method execution and after it exits/finishes I'm back in the original session with no extra breakpoints left?
In other words: I'm in the middle of debugging, and see a method called few lines before (I can't step into it immediately). I don't want to put a binding.pry in the source (it may take a lot of time to start the debugging session again).
The problem has hidden complexity. For instance, one thing you can do is:
pry Foo.new.foo
or
Foo.new.foo.pry
This puts you into a binding on the return value of the function foo, but that is not what you want. It also shows there is complexity when your line of code Foo.new.foo runs, for a function to know what exactly to put a binding on.
So that's really the problem I feel; it would take understanding of the expression on the line to understand where to put a breakpoint. If you imagine it was working like the step function.
If your code Foo.new.foo was already a break point (e.g. you had a function with that code in and you put a breakpoint on a line like that) then you typed step, you would go into Foo.new Foo.initialize instance of foo foo.foo and so on. So, any code trying to do what you would like it to do would have a very hard time understanding where in the expression tree to stop and put your into pry.
This is why you have to tell pry where to stop using break in this context.
Since Ruby supports parallel assignments and automatic value return from functions, almost every assignment and method run ends up creating an output when working on REPLs like IRB and Pry.
Normally I prevent this echo effect by putting a semicolon at the end of each line. For instance:
JSON::parse(very_long_json_string);
This normally prevents REPL echo. But when working with very large enumerables even one mistake can generate enough output to make a mess on the screen and put all my useful command history out of memory before I have the reflex to hit the break.
Is there a way to turn this echo effect off by default in Pry? As mentioned in the comments below (#Stefan), the same can be achieved in IRB by setting conf.echo = false.
In IRB there is:
conf.echo = false
In Pry you could replace the print object with an empty proc:
_pry_.config.print = proc {}
You'll have to store the old print object in order to restore it.
In both cases, the result of the last expression is still available via _
Is it possible to detect when the value of a variable has changed using the lua debug library.
Something like A callback function which would give details like the function in which the value was changed, previous value, etc.
Is such a thing possible?
I read about hooks, but I'm not sure hooks can be set to variables.
If you don't mind using a debugger, then some debuggers allow you to set Watch expressions, which will be triggered when the condition in the expression is true. I'll show how this can be done in MobDebug (it is using lua debug library, but there is no direct way to detect a variable change as far as I know).
Let say we have a script start.lua like the one below and want to detect where foo gets value 2:
print("Start")
local foo = 0
for i = 1, 3 do
local function bar()
print("In bar")
end
foo = i
print("Loop")
bar()
end
print("End")
Download mobdebug.lua and make it available to your scripts (the simplest way is to put it into the folder with your scripts).
Start the server using lua -e "require('mobdebug').listen()" command.
Start the client using lua -e "require('mobdebug').loop()" command.
You will see the prompt in the server window: '>'. Type load start.lua to load the script.
Type step and then step again. You will see "Paused at file start.lua line 3".
Let's see what the value of foo is. Type eval foo and you should see 0.
Now we can set up our watch. Type setw foo == 2. You can specify any Lua expression after setw command; the execution of your script will be stopped when the condition is evaluated as true.
Continue execution of the script using "run" command.
The watch now fires, which will show you the message like: "Paused at file start.lua line 8 (watch expression 1: [foo == 2])". This means that the previous expression changed the value of foo to 2 and the execution is stopped at line 8. You can then inspect your script and the current values (you can use "eval" and "exec" commands to run any Lua code to be evaluated in your script environment) to find what triggered the change.
The benefit of this approach is that you are not limited to monitoring table values and can specify any expression. The main disadvantage is that your script runs under a debugger and the expression is evaluated after each step, which may get really slow.
You can do this to a certain extent in Lua by using metatables and keeping a "proxy" table, and using the __newindex function call to detect attempts to add a variable.
This is covered here in the Programming in Lua book under the section "Tracking Table Accesses":
http://www.lua.org/pil/13.4.4.html
See Also
http://www.gammon.com.au/forum/?id=10887
I am operating a huge code base and want to monitor a value of a particular variable (which is buried deep down inside one of the files)especially when it gets set to zero.
1) Variable does not belong to global scope .Is there a better option than to first set breakpoint into the function where it is defined and then set the watch point?
2) After trying the option in 1 I see that watch point gets deleted after a while saying its out of frame which used this .This way it adds to the tediousness of the procedure since I have to add it again and again?Any workarounds?
3) Is there a way to check ie watch if a particular variable is equal to 0( or any specific constant)?
want to monitor a value of a particular variable
Often this is not the best approach, especially in large codebases.
What you really likely want to do is understand the invariants, and assert that they are true on entry and exit to various parts of the code.
1) Variable does not belong to global scope .Is there a better option than to first set breakpoint into the function where it is defined and then set the watch point?
No. For automatic (stack) variables you have to be in the scope where the variable is "active".
What you can do is set a breakpoint on some line, and attach commands to that breakpoint that will set the watchpoint automatically, e.g.
(gdb) break foo.c:123
(gdb) commands 1
silent
watch some_local
continue
end
3) Is there a way to check ie watch if a particular variable is equal to 0
You can't do that with a watchpoint, but you can with a conditional breakpoint:
(gdb) break foo.c:234 if some_local == 0
I will assume that you are using Linux. You can try this:
The first step is to make the variable static, like:
static int myVar;
Then, after compiling your code using -ggdb, you must discover the address of the variable inside your binary, like so (I have used a real case as example):
readelf -s pdv | grep tmp | c++filt
In my situation, the output is:
47: 081c1474 4 OBJECT LOCAL DEFAULT 25 startProc(int)::tmp
The address in this case is 081c1474. Now you can set a watch point inside GDB:
watch *0x081c1474
Mind the "*0x" before the correct address.
I know this question is old, but I hope it helps anyway.
I don't know why I'm having so much trouble groking the documentation for the elisp debugger.
I see it has a commands to "step-into" (d). But for the life of me, I cannot see a step-out or step-over.
Can anyone help?
If I have this in the Backtrace buffer:
Debugger entered--returning value: 5047
line-beginning-position()
* c-parse-state()
* byte-code("...")
* c-guess-basic-syntax()
c-show-syntactic-information(nil)
call-interactively(c-show-syntactic-information)
...where do I put the cursor, and what key do I type, to step out of the parse-state() fn ? by that I mean, run until that fn returns, and then stop in the debugger again.
When debugging, I press ? and I see:
o edebug-step-out
f edebug-forward-sexp
h edebug-goto-here
I believe o (it is step-out) and f (like step over) are what you're looking for, though I also find h extremely useful.
'c' and 'j' work kind of like a step-out and step-over. When a flagged frame (indicated by "*") is encountered (the docs say "exited" but this doesn't seem to be how the debugger behaves), the debugger will be re-entered. When the top frame is flagged, they work like step-over; when it isn't, they work like step-out.
In your example backtrace, typing either will step out of line-beginning-position into c-parse-state. The frame flag should clear, so typing either a second time should step out of c-parse-state.
Hm. I, for one, prefer debug to edebug, but to each her own...
As to debug, I use d, c, e, and q.
If you do use debug, one thing to keep in mind, which can save time and effort, is that when you see a macro call (starts with#) you can just hit c to expand the macro -- there is normally no sense in digging into the macro expansion code (unless you wrote the macro and you are trying to debug it).
In particular, for dolist, there are two levels of macroexpansion to skip over using c: one for dolist and one for block.
HTH.