[Rprotobuf-yada] do we need to require protoc ?

[Romain Francois]

On 01/23/2010 03:42 PM, Dirk Eddelbuettel wrote:
>
> On 23 January 2010 at 14:29, Romain Francois wrote:
> | Hi,
> |
> | It's been a while, but I'd like to be back on this track.
>
> It has been a while...
>
> | We absolutely need libproto, but I was wondering if we could not have to
> | require protoc and basically compile with files with protoc at build
> | time, as part of the vignette trick.
>
> But if we "compile with protoc at built time" it means we have a built-time
> dependency on protoc, hence we need protoc, hence the test in configure.in

Difference is that __we__ build, but the users install

> I am missing something. But I am post-run yet still pre-coffee...
>
> | This would also reduce the complexity of the Makevars (e.g. we would not
> | need an all target, etc ...)
>
> And you're thinking we would pre-generate them outside of the 'package build'
> and just check the resulting files in?  Could do.

That was my plan B.

These files don't change much anyway, so that might be just the easiest 
way.

> Where is the all: target nagging you? Compiler warnings?

If the files are pre-generated, then the only thing the Makevars.in has 
to contain is this:

PKG_CPPFLAGS=		@PKG_CPPFLAGS@
PKG_LIBS=		@PKG_LIBS@

Plus, as we have seen on Rcpp, the all target is the pathway to many 
abilities some consider to be unnatural (ask google...)

> Dirk
>


-- 
Romain Francois
Professional R Enthusiast
+33(0) 6 28 91 30 30
http://romainfrancois.blog.free.fr
|- http://tr.im/KfKn : Rcpp 0.7.2
|- http://tr.im/JOlc : External pointers with Rcpp
`- http://tr.im/JFqa : R Journal, Volume 1/2, December 2009