[BUG] Cannot Open SPI Interface

[TheCedarPrince]

Hi @Alexander-Barth and co,

Carrying on from a discussion from Discourse, the merge b48be39 did help me get a bit further with opening an SPI interface, but I am still receiving an error:

julia> using PiGPIO

julia> p = Pi()
[ Info: Successfully connected!
Pi("localhost", 8888, true, PiGPIO.SockLock(Sockets.TCPSocket(RawFD(17) open, 0 bytes waiting), ReentrantLock(nothing, 0x00000000, 0x00, Base.GenericCondition{Base.Threads.SpinLock}(Base.IntrusiveLinkedList{Task}(nothing, nothing), Base.Threads.SpinLock(0)), (2, 547614619512, 547614189232))), PiGPIO.CallbackThread(PiGPIO.SockLock(Sockets.TCPSocket(RawFD(17) open, 0 bytes waiting), ReentrantLock(nothing, 0x00000000, 0x00, Base.GenericCondition{Base.Threads.SpinLock}(Base.IntrusiveLinkedList{Task}(nothing, nothing), Base.Threads.SpinLock(0)), (2, 547614619512, 547614189232))), PiGPIO.SockLock(Sockets.TCPSocket(RawFD(18) paused, 0 bytes waiting), ReentrantLock(nothing, 0x00000000, 0x00, Base.GenericCondition{Base.Threads.SpinLock}(Base.IntrusiveLinkedList{Task}(nothing, nothing), Base.Threads.SpinLock(0)), (0, 547589698944, 0))), true, true, 0, 0x00000000, Any[]))

julia> PiGPIO.spi_^C

julia> h = PiGPIO.spi_open(p, 1, 100000, Int32(0))
ERROR: MethodError: no method matching iterate(::IOBuffer)

Closest candidates are:
  iterate(::Pipe, ::Int64)
   @ Base stream.jl:1263
  iterate(::Pipe)
   @ Base stream.jl:1262
  iterate(::Base.MethodSpecializations)
   @ Base reflection.jl:1148
  ...

Stacktrace:
 [1] _pigpio_command_ext(sl::PiGPIO.SockLock, cmd::Int64, p1::Int64, p2::Int64, p3::Int64, extents::IOBuffer, rl::Bool)
   @ PiGPIO ~/.julia/packages/PiGPIO/1FebP/src/pi.jl:128
 [2] _pigpio_command_ext(sl::PiGPIO.SockLock, cmd::Int64, p1::Int64, p2::Int64, p3::Int64, extents::IOBuffer)
   @ PiGPIO ~/.julia/packages/PiGPIO/1FebP/src/pi.jl:126
 [3] spi_open(self::Pi, spi_channel::Int64, baud::Int64, spi_flags::Int32)
   @ PiGPIO ~/.julia/packages/PiGPIO/1FebP/src/spiSerial.jl:93
 [4] top-level scope
   @ REPL[5]:1

After debugging, I can observe that the error occurs within this PiGPIO.jl function:

PiGPIO.jl/src/pi.jl

Lines 115 to 138 in e34746d

	"""
	Runs an extended pigpio socket command.
	* `sl`: command socket and lock.
	* `cmd`: the command to be executed.
	* `p1`: command parameter 1 (if applicable).
	* `p2`: command parameter 2 (if applicable).
	* `p3`: total size in bytes of following extents
	* `extents`: additional data blocks
	"""
	function _pigpio_command_ext(sl, cmd, p1, p2, p3, extents, rl=true)
	ext = IOBuffer()
	Base.write(ext, Array(reinterpret(UInt8, [cmd, p1, p2, p3])))
	for x in extents
	write(ext, string(x))
	end
	lock(sl.l)
	write(sl.s, ext)
	msg = reinterpret(Cuint, sl.s)[4]
	if rl
	unlock(sl.l)
	end
	return res
	end

In particular, I know it is this loop:

for x in extents 
    write(ext, string(x)) 
end 

As it is trying to iterate over a IOBuffer -- but I am unsure what it is trying to actually do or access. I tested old Julia versions starting from 1.0 and I have concluded this is just buggy code. There was not a point in time post-1.0 Julia that I could find which allows iteration over an IOBuffer.

Any thoughts here? Thanks!

~ tcp 🌳

[Alexander-Barth]

I have also seen this problem while looking at the issue:
#15

There are some changes in the i2c branch, in particular:
c80dd92

I just merged the i2c branch into the master, since the problem also affects spi as your example shows,.

Can you re-try with the master branch?

I don't know the details of the history of this code, but it was converted from the python code. Probably only a subset of the pigpio.py functionality was ever tested. Rather than looking at old julia version, it might be easier to find the issue by looking at the python code.

[aviks]

I tested old Julia versions starting from 1.0

Not to say that the code wasn't always buggy, but this code is prehistoric. From the first commit, I see a comment saying Julia Version 0.5.0-dev+3050 [c96f322] 😃 😱

[TheCedarPrince]

I have also seen this problem while looking at the issue: #15

There are some changes in the i2c branch, in particular: c80dd92

I just merged the i2c branch into the master, since the problem also affects spi as your example shows,.

Can you re-try with the master branch?

I have good news! This now appears to work!

julia> using PiGPIO

julia> p = Pi()
[ Info: Successfully connected!
Pi("localhost", 8888, true, PiGPIO.SockLock(Sockets.TCPSocket(RawFD(17) open, 0 bytes waiting), ReentrantLock(nothing, 0x00000000, 0x00, Base.GenericCondition{Base.Threads.SpinLock}(Base.IntrusiveLinkedList{Task}(nothing, nothing), Base.Threads.SpinLock(0)), (0, 0, 0))), PiGPIO.CallbackThread(PiGPIO.SockLock(Sockets.TCPSocket(RawFD(17) open, 0 bytes waiting), ReentrantLock(nothing, 0x00000000, 0x00, Base.GenericCondition{Base.Threads.SpinLock}(Base.IntrusiveLinkedList{Task}(nothing, nothing), Base.Threads.SpinLock(0)), (0, 0, 0))), PiGPIO.SockLock(Sockets.TCPSocket(RawFD(18) paused, 0 bytes waiting), ReentrantLock(nothing, 0x00000000, 0x00, Base.GenericCondition{Base.Threads.SpinLock}(Base.IntrusiveLinkedList{Task}(nothing, nothing), Base.Threads.SpinLock(0)), (0, 0, 0))), true, true, 0, 0x00000000, Any[]))

julia> h = PiGPIO.spi_open(p, 1, 100000, 0)
1

And oh my gosh @aviks -- I had no idea it was that old 😱 I'll take your advice as well regarding the original Python code @Alexander-Barth!