Regenerate LibCURL.jl via Clang.jl

[nicoleepp]

The code used to regenerate is included in gen/generate.jl. Used Clang.jl on 0.7.

Closes #52

LibCURL and FTPClient tests pass.

Version of CURL/LibCURL used: 7.61.1

[omus]

The 0L literals are of type long and should be Clong(0) in Julia

[codecov-io]

Codecov Report

Merging #53 into master will decrease coverage by 13.05%.
The diff coverage is 6.34%.

@@            Coverage Diff             @@
##           master     #53       +/-   ##
==========================================
- Coverage      20%   6.94%   -13.06%     
==========================================
  Files           2       2               
  Lines          10      72       +62     
==========================================
+ Hits            2       5        +3     
- Misses          8      67       +59

Impacted Files	Coverage Δ
src/LibCURL.jl	11.11% <ø> (ø)	⬆️
src/lC_curl_h.jl	6.34% <6.34%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 28f76b5...0b61859. Read the comment docs.

[nicoleepp]

Not sure why the ANONYMOUS constants were generated or if they are needed (in lc_common_h.jl).
Edit: They have now been deleted since they came from badly parsed enums.

[nicoleepp]

There is one constant I'm not sure what to do with. It gets generated as

const CURL_ZERO_TERMINATED               = ((size_t) -1)

but this errors in julia:

julia> Csize_t(-1)
ERROR: InexactError: check_top_bit(Int64, -1)
Stacktrace:
 [1] throw_inexacterror(::Symbol, ::Any, ::Int64) at ./boot.jl:567
 [2] check_top_bit at ./boot.jl:581 [inlined]
 [3] toUInt64 at ./boot.jl:692 [inlined]
 [4] UInt64(::Int64) at ./boot.jl:722
 [5] top-level scope at none:0

I have left the constant defined for now as:

const CURL_ZERO_TERMINATED               = -1

[nicoleepp]

My other concern is that some things that used to be Union{} are now just Cvoid, and constants that were Ints are now specified as UInt32. Not sure if these changes are what we want or if it's just how things are generated.

[omus]

The C code probably wants to do:

julia> reinterpret(Csize_t, -1)
0xffffffffffffffff

[nicoleepp]

I removed the c and ctypedef macros because we would have to manually go through the generated code and change the format to use them, but, if it makes for easier reviewing, I can add them back and re-format the definitions.

[ararslan]

Just some minor stylistic pedantry 😄

[omus]

Looking good

[omus]

@nicoleepp I assume you're good with squashing the commits?

[omus]

Travis also passed but seems to have not linked back to this PR

[nicoleepp]

I'm good with squashing

[ararslan]

Nice work here!