How do I read/debug FunctionClauseError? - debugging

Although there are similar questions at SO (example), for some reason the error is formatted in another way in my case (and also I understand the general meaning of FunctionClauseError), so let me ask "once more".
Here's the question I'm getting:
{:error, reason} ->
Logger.error(fn -> ["failed to fetch internal transactions for >blocks: ", inspect(reason)] end,
error_count: unique_numbers_count
)
gives (I've added line breaks)
2022-04-13T18:44:19.749 application=indexer fetcher=internal_transaction count=10 error_count=10 [error]
failed to fetch internal transactions for blocks: %FunctionClauseError{args: nil, arity: 1,
clauses: nil, function: :finalize, kind: nil, module: EthereumJSONRPC.Geth.Tracer}
so obviously reason is
%FunctionClauseError{args: nil, arity: 1, clauses: nil, function: :finalize, kind: nil,
module: EthereumJSONRPC.Geth.Tracer}
I wonder what do these bits mean and can they help with debugging?
In module EthereumJSONRPC.Geth.Tracer function finalize is defined as
defp finalize(%{stack: [top], calls: [calls]}) do
and an example of the argument passed to it (causing the error) is:
%{
calls: [[], [], [], []],
depth: 4,
stack: [
%{
"callType" => "staticcall",
"from" => nil,
"gas" => 432013,
"gasUsed" => 21661,
"input" => "0xefc4cfa.....",
"output" => "0x",
"outputLength" => 64,
"outputOffset" => 4346,
"to" => "0x0000000000000000000000000000000000000001",
"traceAddress" => [0, 0, 0],
"type" => "call",
"value" => "0x0"
},
%{
"callType" => "staticcall",
"from" => nil,
"gas" => 438139,
"gasUsed" => 1726,
"input" => "0xefc4c.......",
"output" => "0x",
"outputLength" => 64,
"outputOffset" => 4026,
"to" => "0x0000000000000000000000000000000000000001",
"traceAddress" => [0, 0],
"type" => "call",
"value" => "0x0"
},
%{
"callType" => "staticcall",
"from" => nil
"gas" => 445060,
"gasUsed" => 2521,
"input" => "0xefc4......",
"output" => "0x",
"outputLength" => 64,
"outputOffset" => 3706,
"to" => "0x0000000000000000000000000000000000000001",
"traceAddress" => [0],
"type" => "call",
"value" => "0x0"
},
%{
"callType" => "call",
"from" => "0x9a66644084108a1bc23a9ccd50d6d63e53098db6",
"gas" => 460960,
"gasUsed" => 11500,
"input" => "0xba2c.........",
"output" => "0x",
"to" => "0x841ce48f9446c8e281d3f1444cb859b4a6d0738c",
"traceAddress" => [],
"type" => "call",
"value" => "0x0"
}
],
trace_address: [0, 0, 0, 0]
}
which looks ok for me (it has both stack and calls props). Can I extract something more from the args: nil, arity: 1, clauses: nil, function: :finalize, kind: nil bit for debugging? If you point the cause, it will be helpful too, but I also want to understand how to debug this. Can extra props (depth, trace_address) be the source of the problem? Or should I look for the error cause inside the finalize function body? (as far as I understand, no)
PS A note for others seeking help with debugging: the bit before the error text
application=indexer fetcher=internal_transaction count=10 error_count=10
may contain useful info: if you look for Logger.metadata, you may find something like
Logger.metadata(fetcher: :internal_transaction)
which can give a hint about the source of the error.

By default, if you don't use try / rescue, a FunctionClauseError will be logged with all the information (values, patterns, stacktrace) needed to debug it:
> String.downcase(1)
** (FunctionClauseError) no function clause matching in String.downcase/2
The following arguments were given to String.downcase/2:
# 1
1
# 2
:default
...
But if you rescure the error and return it as a tuple, you will lose most of this information:
try do
String.downcase(1)
rescue
err -> {:error, err}
end
The result will only contain a very barebones struct:
{:error,
%FunctionClauseError{
args: nil,
arity: 2,
clauses: nil,
function: :downcase,
kind: nil,
module: String
}}
You can log directly in the rescue clause, where __STACKTRACE__ is available, using the approach explained in this official guide:
try do
String.downcase(1)
rescue
err ->
# format the exception in a readable way
Exception.format(:error, err, __STACKTRACE__)
|> Logger.error()
# return value
:error
end
That being said, this is not so common in Elixir to rely on try/rescue as explained in the guide above. Most of the time you just let unexpected things fail (the famous "Let it crash" philosophy), and use :error tuples for cases you actually expect and want to handle.

Sidenote: the answer provided by #sabiwara has a ton of helpful info and should be marked as correct.
Here is the cause of the error you got:
defp finalize(%{stack: [top], calls: [calls]})
This defines a function, accepting a map with keys, including but not limited to :stack and :calls, and their values being lists containing exactly one element.
You are passing longer lists there, hence the error. This is what you likely wanted to do:
defp finalize(%{stack: top, calls: calls})
when is_list(top) and is_list(calls)

Related

Better failing message for asserting attributes of multiple objects in the collection

For asserting that object has expected attributes I come up with custom methods
def be_active_sales_order_with(quantity)
expected = {
:order_type => "sales",
:activated_at => be_within(1.minute).of(Time.current),
:quantity => quantity
}
have_attributes(expected)
end
When method above used with one object, failing message provide clearly formatted difference which attribute's value is different than expected.
Failing message:
expected # to have attributes {:order_type => "sales", :activated_at => (be within 300 of 2019-11-26 15:56:17), :quantity => 3} but had attributes {:order_type => "sales", :activated_at => 2019-11-24 15:56:17.000000000, :quantity => 3}
Diff:
## -1,5 +1,5 ##
:order_type => "sales",
-:activated_at => (be within 300 of 2019-11-26 15:56:17),
+:activated_at => 2019-11-24 15:56:17.000000000,
:quantity => 3,
it "creates single order" do
# configuration for single order
orders = create_order.call(quantity: 3)
expect(orders.only).to be_active_sales_order_with(3)
end
But for case when multiple orders can be created failing message difficult to read without debugging failed test.
it "creates multiple orders" do
# configuration for multiple order
orders = create_order.call(quantity: 3)
expect(orders).to contain_exactly(
be_active_sales_order_with(3),
be_active_sales_order_with(3),
be_active_sales_order_with(3)
)
end
Failing message:
expected collection contained: [(have attributes {:order_type => "sales", :activated_at => (be within 300 of 2019-11-26 15:56:17), :quantity => 3})]
actual collection contained: [#Order id: 885385454, customer_id: 367693892, topic: nil, created_at: "2019-..., segment_key: "9bf937d687eaff19", stock_id: nil, notified_by: nil, use_discount: false, user_id: 548091186]
the missing elements were: [(have attributes {:order_type => "sales", :activated_at => (be within 300 of 2019-11-26 15:56:17), :quantity => 3})]
the extra elements were: [#Order id: 885385454, customer_id: 367693892, topic: nil, created_at: "2019-..., segment_key: "9bf937d687eaff19", stock_id: nil, notified_by: nil, use_discount: false, user_id: 548091186]
How I can build more readable message, which provide readable difference between expected and actual object's attributes

Checking a nested hash key

I have a hash yaml, which is sometimes like:
{"foo" => {"bar" => 10}}
and sometimes like:
{"foo" => nil}
I want to do a certain action depending on whether "bar" is present.
I write this code:
if yaml["foo"] && yaml["foo"].key?["bar"]
...
I'd like to know if there's an idiomatic way to deal with that conditional, especially the first part where I have to check the existence of the parent key.
Hash#dig comes in very handy for cases like yours:
hash = {"foo" => {"bar" => { "baz" => 10}}}
hash.dig('foo', 'bar', 'baz')
#=> 10
Note, that if at any point of digging it returns nil, the method won't blow up but just return nil as result:
hash.dig('foo', 'baz')
#=> nil

Mongo Group function call in Ruby

I have a mongodb collection like this
{"assigneeId" => 1000, "status" => 3, "starttime" => "2014 Feb 25", "numofdays => 6}
{"assigneeId" => 1000, "status" => 2, "starttime" => "2014 Jan 10", "numofdays => 6}
{"assigneeId" => 1000, "status" => 3, "starttime" => "2014 Jan 1", "numofdays => 20}
I wrote a MongoDB query to group the above collection with assigneeId whose status is 3 and add the value of numofdays like this
db.events.group ( { key: {assigneeId:1}, cond: {status: {$gte: 3}}, reduce: function (curr, result) {result.total++; result.numofdays += curr.numofdays;}, initial: {total:0, numofdays:0} })
This gives the output as expected from the Mongodb cli.
When I write the ruby code for executing the same query, I could not make it work. Am getting lot of different hits for similar problem but none of the format seems to work. I could not go past the syntax error in ruby for grouping function. Please note I am not using mongoid. If I could get this work with Aggregation framework, that is also fine with me.
This is the ruby query I wrote and the error for the same.
#events_h.group (
:cond => {:status => { '$in' => ['3']}},
:key => 'assigneeId',
:initial => {count:0},
:reduce => "function(x, y) {y.count += x.count;}"
)
Error I am getting
analyze.rb:43: syntax error, unexpected ',', expecting ')'
out = #events_h.group (["assigneeId"], { }, { }, "function() { }")
^
analyze.rb:43: syntax error, unexpected ')', expecting keyword_end
analyze.rb:83: syntax error, unexpected end-of-input, expecting keyword_end
Any help is much appreciated.
Thanks
The following test includes solutions for using both methods, group and aggregate.
I recommend that you use aggregate, it is faster and avoids JavaScript and consuming a V8 engine.
Hope that this helps.
test.rb
require 'mongo'
require 'test/unit'
class MyTest < Test::Unit::TestCase
def setup
#events_h = Mongo::MongoClient.new['test']['events_h']
#docs = [
{"assigneeId" => 1000, "status" => 3, "starttime" => "2014 Feb 25", "numofdays" => 6},
{"assigneeId" => 1000, "status" => 2, "starttime" => "2014 Jan 10", "numofdays" => 6},
{"assigneeId" => 1000, "status" => 3, "starttime" => "2014 Jan 1", "numofdays" => 20}]
#events_h.remove
#events_h.insert(#docs)
end
test "group" do
result = #events_h.group(
:cond => {:status => {'$in' => [3]}},
:key => 'assigneeId',
:initial => {numofdays: 0},
:reduce => "function(x, y) {y.numofdays += x.numofdays;}"
)
assert_equal(26, result.first['numofdays'])
puts "group: #{result.inspect}"
end
test "aggregate" do
result = #events_h.aggregate([
{'$match' => {:status => {'$in' => [3]}}},
{'$group' => {'_id' => '$status', 'numofdays' => {'$sum' => '$numofdays'}}}
])
assert_equal(26, result.first['numofdays'])
puts "aggregate: #{result.inspect}"
end
end
$ ruby test.rb
Loaded suite test
Started
aggregate: [{"_id"=>3, "numofdays"=>26}]
.group: [{"assigneeId"=>1000.0, "numofdays"=>26.0}]
.
Finished in 0.010575 seconds.
2 tests, 2 assertions, 0 failures, 0 errors, 0 pendings, 0 omissions, 0 notifications
100% passed
189.13 tests/s, 189.13 assertions/s

Hash declaration syntax error in irb

1. { :a => 10 } #=> no error
2. { a: 10 } #=> no error
3. { :"str" => 10 } #=> no error
4. { "str": 10 } #=> syntax error, unexpected ':', expecting =>
Isn't 4. same as 2? Why 2 is working and 4 throws syntax error?
My understanding is that {"key": value} is not a valid syntax as it is not clear whether it means {:"key" => value} or {"key" => value}
There is a discussion on this here. Quote from Matz in the discussion
| Iff {'key': 'value'} means {:key => 'value'} I have no objection.
| Won't that be misleading? I think the OP wants {'key': 'value'} to mean {'key' => 'value}
But considering the fact that {key: "value"}
is a shorthand for {:key => "value"}, {"key": "value"} should be a
shorthand for {:"key" => "value"}. Besides that, since it reminds me
JSON so much, making a: and "a": different could cause more confusion
than the above misleading.
matz.
Hash: Hashes allow an alternate syntax form when your keys are always symbols.
options = { :font_size => 10, :font_family => "Arial" }
You could write it as:
options = { font_size: 10, font_family: "Arial" }
In your first 3 cases all are symbols in key position,but the fourth is a string instance,not the symbol instance as key.That's the reason 4th case is invalid Ruby syntax.
{ :a => 10 }.keys[0].class # => Symbol
{ a: 10 }.keys[0].class # => Symbol
{ :"str" => 10 }.keys[0].class # => Symbol
No. (1) is standard symbol, (2) is shorthand 1.9 syntax for symbol-key hashes, (3) is shorthand for "str".to_sym, (4) does not exist and you should use the hashrocket.

What is the common way to handle optional method params?

I have a method in which i want to pass in dynamic params. The method is called in a loop and sometimes value2 is available and sometimes not.
What is the common way to handle optional method params?
my_method(:value1 => 1,
:value2 => 2 if foo, # this is not working
:value3 => 3)
I usually create a hash like this:
opts = {:value1 => 1,
:value3 => 3}
opts[:value2] = 2 if foo
my_method(opts)
The benefit of this approach is that everyone catches the if foo as it is a special case. Otherwise many programmers, like myself, will miss this at first glance and get confused why :value2 is not set.
Sometimes you have default settings, then you can use this approach:
default = {:value1 => 0,
:value2 => 0,
:value3 => 0}
opts = {:value1 => 1,
:value3 => 3}
my_method(default.merge(opts))
Or even better:
DEFAULT_OPTS = {:value1 => 0,
:value2 => 0,
:value3 => 0}
def my_method(opts)
opts = DEFAULT_OPTS.merge(opts)
# ...
end
my_method(...)

Resources