contents broken for Go in latest released ksc v0.8

[lannybroo]

Trying to compile a .ksy that has a fixed contents field using ksc --target all or ksc --target go fails due to some issue with the go compiler.

Example broken_go.ksy:

meta:
  id: broken_go
seq:
  - id: magic
    contents: [0x44, 0x42, 0x2F]

Compiler output:

> ksc --version
kaitai-struct-compiler 0.8
> ksc --target go broken_go.ksy --verbose file
parsing broken_go.ksy...
reading broken_go.ksy...
... compiling it for go... 
List(ArraySeq(IntNum(68), IntNum(66), IntNum(32))) (of class io.kaitai.struct.exprlang.Ast$expr$List)

[KOLANICH]

kaitai-struct-compiler 0.8

Have you tried the current unstable version of the compiler?

[GreyCat]

golang support is still in relatively stages, even as of now in unstable version. Latest CI report shows that it passes only 41% of tests, which is actually pretty bad, although your particular use case now passes (see FixedContents test).

Unfortunately, there's been pretty quiet in golang compiler's land lately: last few updates to #146 were done by me, and I'm probably the person least interested in golang, as I don't use it in my everyday's life, I don't have any experience with it, etc. @lannybroo, if you feeling like you could contribute something to golang support, you'd be most welcome.

[lannybroo]

Thanks for the link to the CI report -- I had not realized that go is essentially not ready yet. I'd love to contribute but also don't have a good background in go (more interest in Construct, however). Perhaps the right approach for golang is to not include go in the ksc --target all command until it passes more useful percentage of the tests?

[GreyCat]

There is no ksc --all command. There is ksc -t all, which is 99.9% of time used for exactly for the purpose of these CI test runs — if we'll omit go from there, how would we test it?

Also, note that, unfortunately, @arekbulski stopped working on Construct: #377 (comment), so most likely Construct target will be phased out in some time :(

[lannybroo]

Ah, I interpreted ksc --target all as the library's typical way of emitting trusted parsers for all of the supported languages once I have tested my .ksy in the language I happen to be developing with. As an end user of the library, I did not have the CI test run in mind.

[lannybroo]

Closing this issue. I don't know that there are actual actions to take (other than perhaps not including unsupported/incomplete languages when someone runs ksc --target all).

[GreyCat]

KS is still an project under development, and there's no clear distinction between "unsupported/incomplete" vs "supported/complete". We can probably mark up some languages as "officially good" if they pass 100% of the tests, but then probably only Java & Ruby will be passing up to that mark — all other languages will have some percentages lower than that. This might seem terrible, but in reality if test pass rate is 85-90%, it will be still useful for some (or even many) people. Even at 41%, I know that some people have reported some success with go target.