Hi All, Unfortunately, the R developers haven't (yet) made the API for interfacing with R fully public, so it is necessary to use 'internal' header files. Since these header files are (fortunately) constant for a given release of R, and most people don't have a full source installation of R available, I've found it necessary to include the header files as part of the package.
I will be working with the R core team to standardize and make public the shared library API. After this occurs, everything that rpy needs should become available in the standard set of public header files, resolving this issue. FWIW, I've added the 2.1.0 headers, and bumped up the release number to 0.4.2. This release (just uploaded to SF.net) also includeds a fix for a build error on gcc > 4.0. -Greg > -----Original Message----- > From: Dirk Eddelbuettel [mailto:[EMAIL PROTECTED] > Sent: Monday, May 23, 2005 7:26 AM > To: Don Armstrong; [EMAIL PROTECTED] > Cc: Warnes, Gregory R > Subject: Re: Bug#310198: rpy: FTBFS: Does not have headers for R 2.1.0 > > > > On 22 May 2005 at 22:33, Don Armstrong wrote: > | clone 310198 -1 > | severity -1 wishlist > | reassign -1 r-base-core > | retitle -1 please distribute Startup.h from > src/include/Startup.h in /usr/lib/R/include/ > | tag -1 = > | submitter -1 ! > | thanks > | > | > | On Sun, 22 May 2005, Dirk Eddelbuettel wrote: > | > I just checked the changelog. 0.4.1 (upstream) added headers for R > | > 2.0.1. Debian 0.4.1-3 added R-2.1.0 building on the existing > | > "batteries err generators included" framework of yielding to all > | > known recent R versions. > | > | I'd argue that this is a bad idea, because it relies on the copy > | present in the package being definitive, instead of using what is > | actually installed. [Not to mention this sort of > duplication requiring > | extra effort on the part of the maintainer to keep things > in sync...] > | > | The patch to do this is exceedingly trivial, except for one minor > | problem: > | > | include_dirs = ['/usr/lib/R/include', > | 'src' ] > | > | The minor problem is that r-base-core does not distribute Startup.h > | for some reason. [Why the headers are in r-base-core instead of > | r-base-dev, I've no idea... but I suppose there's a good reason.] > > The (upstream) R Core team is very careful not to "promise" > entry points to R > that it won't be able to support. So I doubt that you will > get them to export > other header files. > > Likewise, there are often more files and functions needed > which is why Greg > Warnes (rpy upstream) opted for the full-blown "all headers" solution, > > Lastly, you ignore that this issue needs a fix beyond Debian > where your patch > does nothing. > > I don't think I should deviate in the r-base-core package > from upstream, so I > think I will close your wishlist bug. > > Dirk > > | > | To that end, I've cloned a wishlist bug on r-base-core asking for > | Startup.h to be distributed. I've attached the patch, but > it won't fix > | the current "issue" until r-base-core implements this change. [I > | suppose you could just distribute Startup.h in the > package... but that > | seems kind of suboptimal.] > | > | > | Don Armstrong > | > | -- > | "There's nothing remarkable about it. All one has to do is > hit the > | right keys at the right time and the instrument plays itself." > | -- Bach > | > | http://www.donarmstrong.com http://rzlab.ucr.edu > | --- setup.py~ 2004-12-22 14:28:46.000000000 -0800 > | +++ setup.py 2005-05-22 22:19:10.000000000 -0700 > | @@ -133,10 +133,12 @@ > | extra_compile_args=["-shared"] > | source_files = source_files + ["src/setenv.c"] > | else: # unix-like systems, this is known to work for Linux > and Solaris > | - RSRC = get_R_SRC() > | - include_dirs = [ os.path.join(RHOME.strip(), 'include'), > | - os.path.join(RSRC.strip(), 'src/include'), > | - 'src' ] > | +# RSRC = get_R_SRC() > | + include_dirs = ['/usr/lib/R/include', > | + 'src' ] > | +# include_dirs = [ os.path.join(RHOME.strip(), 'include'), > | +# os.path.join(RSRC.strip(), 'src/include'), > | +# 'src' ] > | libraries=['R'] > | library_dirs = r_libs > | runtime_libs = r_libs > > -- > Statistics: The (futile) attempt to offer certainty about uncertainty. > -- Roger Koenker, 'Dictionary of Received Ideas of > Statistics' > > LEGAL NOTICE Unless expressly stated otherwise, this message is confidential and may be privileged. It is intended for the addressee(s) only. Access to this E-mail by anyone else is unauthorized. If you are not an addressee, any disclosure or copying of the contents of this E-mail or any action taken (or not taken) in reliance on it is unauthorized and may be unlawful. If you are not an addressee, please inform the sender immediately. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
