It's not so much about whether or not we *can*, but about whether or not we
*should* generate and expose these files.
Fundamentally, Substrait aims to be a means to connect different systems
together by standardizing some interchange format. Currently that happens
to be based on protobuf, so *one*
s to protobuf classes is not Arrow's
> job. You can probably take the upstream (i.e. Substrait's) protobuf
> definitions and compile them yourself, using whatever settings required
> by your project.
>
> Regards
>
> Antoine.
>
>
> Le 03/07/2022 à 21:16, Jeroen
Hi,
I'm not sure I completely understand what you're trying to do, but if lack
of internet access is the only problem, I think you should just be able to
override the URL it tries to download by setting the ARROW_SUBSTRAIT_URL
environment variable to some local file:// URL. I think it should work
gt; wouldn't recommend doing these auto-generation and build steps on your
> own,
> > as you'd have to change the Arrow C++ build system, and specifically
> > "cpp/cmake_modules/ThirdpartyToolchain.cmake", to enable this.
> >
> >
> > Yaron.
> &
022 at 21:32, Li Jin wrote:
> Thanks both. This appears to work!
>
> With regard to linking, I also have libsubstrait.a under build/debug, but
> not in dist/lib - I suppose maybe the substrait classes are statically
> linked into libarrow.so?
>
> Li
>
> On
> At the moment there is a version at [2] which I will propose be the
> official implementation for the Apache Arrow project (although it
> needs a tiny bit of cleanup to remove a comment reference to C++).
> Assuming the discussion doesn't raise any significant concerns in the
> next week or so I'
>From a Substrait perspective, it would be up to Ibis to convert the
column names to the correct indices. This can and should be transparent
to the end user; front-ends already need to know what the schemas are
in order to produce a valid Substrait plan, so Ibis should have all the
information it n
umes substrait I'd
> >> imagine there would be similar problems because schema is part of Spark
> >> operators too and I don't think Substratit cannot convince Spark to go
> >> index based.
> >> I'd imagine during deserialization of the subtrai