Re: [task #5698] First draft of CONTRIBUTING file

[John Darrington]

On Wed, Jul 05, 2006 at 04:11:44PM +0000, Jason H Stover wrote:
     
     Because of these two comments, I'd like to ask some questions about "which
     language."                                                                 
  
                                        
     John, did you mean that if we accept a contribution in, say, GNU           
  
     Octave, that putting it PSPP would require us to translate it into C? 
(That's
     what I thought you meant, but I just want to be sure.)                     
                              

Well that's what I had assumed.
                                        
     I thought we would want the language of the accepted program to be, if not 
C,
     something that gcc will compile. Right?

That would be desirable.
                                        
     If we say, "Give us code that we do not have to translate into C," what
     languages should we accept? The GNU coding standards say to use C, but what
     about Fortran? A lot of statisticians program in Fortran, and we could
     compile Fortran, so should we accept that, too?

Gcc can also compile Java and some other languages too.

From Jason's comments, I'm getting the idea that he's suggesting that
we accept contributions in a GCC supported language, and link them
into the PSPP code at the object level, thus avoiding the need to
translate them into C.   That's certainly possible, but I rather doubt
if it'll save any work.  

Firstly, in order to do that, we'd have to publish FORTRAN bindings
for the libpspp and data APIs.  Secondly, even having done that, in
most cases, I think it'll be necessary to go over contributions with a
fine toothed comb, and massage them into a shape that will suit us.
Given we have to do that,  it's not much extra effort to translate to
another language.  Translating from one procedural language to another
is a straightforward, albeit rather laborous task (a few specially
crafted sed scripts can save hours).  Of course if somebody decides to
contribute in a language such as prolog, then that's another story.

Whichever way we go, I suggest that we move the language proscription
clause from the "Requirements" to the "Recommendations" section.

J'

-- 
PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See http://pgp.mit.edu or any PGP keyserver for public key.

signature.asc
Description: Digital signature