>From the signature, I'm guessing that it will write the address of a string
to the location that its argument points at. I would expect to call this
like so:
pp = Array(Ptr{UInt8})
ret = ccall((:ppl_version, "libppl_c"), Cint, (Ptr{Ptr{UInt8}},), pp)
# check ret for errors (presumably why it returns an int)
version = bytestring(pp[])
Your code is doing something quite different.
On Tue, Feb 17, 2015 at 5:52 PM, Zenna Tavares <[email protected]>
wrote:
> int ppl_version(const char** p)
>
> On Tuesday, February 17, 2015 at 5:38:36 PM UTC-5, Stefan Karpinski wrote:
>>
>> What is the signature of the ppl_version C function?
>>
>> On Tue, Feb 17, 2015 at 5:23 PM, Zenna Tavares <[email protected]>
>> wrote:
>>
>>> I am attempting to wrap the Parma Polyhedron Primitive Library.
>>>
>>> It has a function 'ppl_version' which modifies a set of strings,
>>> represented as a char **. An example function which calls 'ppl_version' is
>>> as follows:
>>>
>>> get_ppl_version() {
>>> const char* p;
>>> (void) ppl_version(&p);
>>> return p;
>>> }
>>>
>>>
>>> My Julia attempt is as follows:
>>>
>>> function get_ppl_version()
>>> versionnum = Array{Uint8}[Array(UInt8, 20) for i = 1:4]
>>> res = ccall((:ppl_version, "libppl_c"), Int, (Ptr{Ptr{Uint8}},),
>>> versionnum)
>>> bytestring(convert(Ptr{Uint8}, versionnum[1]))
>>> end
>>>
>>> However, I get gobbledygook back
>>>
>>> @show get_ppl_version()
>>> =>$(Expr(:call, :ppl_version)) => "Xٮ\n"
>>>
>>> Actually the result is non-deterministic, evaluating it a few times
>>> gives very strange results, e.g. ""set_bigfloat_roundin",
>>> "isppl_initialzied",
>>> ec. which seems to be random mix of my julia soure-code and c code from
>>> the api itself. I'm not sure how that is even possible. What am I doing
>>> wrong?
>>>
>>> Thanks
>>>
>>
>>
