Re: release plans

[John Darrington]

I understand that R can read SAS files, so if we're going to make PSPP
do so, then we should look into the possibility of using their code to
do the job.

J'

On Wed, Dec 20, 2006 at 10:02:07PM -0800, Ben Pfaff wrote:
     Kevin Smith <address@hidden> writes:
     
     > I have a few thoughts on some of the ideas that Jason mentioned
     > below.  First, reading SAS data sets is not a simple task.  SAS
     > data sets or specific to major version, bit ordering and
     > operating system (in some cases anyhow).  One of the major
     > problems that I have run in to as a SAS user is that old data
     > can no longer be read by newer versions of SAS, and thus users
     > with such data are left with no options.  I can see a potential
     > market for an application which can read such data, but I also
     > know that the file formats for SAS datasets (even the old ones)
     > are considered trade secrets.
     
     Usually, this just means that it takes a lot of time to figure
     out how the format works.  Thus, to make it work properly you
     need someone with a lot of time on his hands, and by preference a
     working copy of each version of SAS whose format you're
     interested in.
     
     In other words, I'd categorize SAS data file formats as "lots of
     work" not as "impossible", etc.
     
     > I assume that getting SPSS data formats is equally as difficult
     > as SAS doesn't read SPSS data.
     
     SPSS published an older version of their data file format on the
     web several years ago, giving us a head start on that format.
     Since then they've only made minor changes, so it hasn't been a
     big problem.
     
     > Is what Jason is suggesting really feasible?  As far as syntax
     > files, certainly SAS syntax is just plain text.  I don't have a
     > lot of experience with language parsers but SAS and SPSS aren't
     > fully comparible in terms of what algorithms are used on a huge
     > range of functions.  Nor is data handled in the same manner, so
     > I am uncertain how feasible this idea is.  [...]
     
     This part sounds less plausible to me, too, than understanding
     the SAS data format.  Perhaps we could start a separate ASA
     project :-)

-- 
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