FreeTDS Development Group

["James K. Lowden" <jklowden AT schemamania.org>]

On Tue, 6 Jan 2004, christos AT zoulas.com (Christos Zoulas) wrote:
> what do we do about fixing bcp for
> 0.62? Obviously my changes don't work in all cases, and break existing
> functionality. As such, they should be backed out. On the other hand,
> shipping 0.62 without the changes keeps native bcp broken.
> Unfortunately, I don't have a sybase server to test this, and my mssql
> server seems to pass the regression test.

Yes, the bcp unit test passes; no, you haven't broken anything.  

My test used "freebcp -n" with TDS 7.0.  It failed.  The question is, what
to do wrt 0.62?  

You contributed a patch designed to fix (implement, really) a feature you
need, and  people who submit patches deserve IMO to have their issue
addressed, if at all possible, in the next release.  Problem is, your
issue isn't really addressed, at least not completely, because native bcp
doesn't quite work.  

Am I wrong about that?  Is bcp working to your satisfaction?  

My feeling is this, tell me if you agree:  We'll keep ironing out issues
affecting the immediate release.  Maybe we'll get native bcp working,
maybe not, depending on our collective cleverness and Bill's time.  But
we'll iron out any 64-bit bugs and installation issues, and stamp it safe
for general consumption.  If it's otherwise ready, but native bcp isn't
yet working, we'll slap a sticker on it and start working on 0.63.  

Native bcp will be made to work again, I'm sure, in the next week or two. 
It's not like there's some giant conceptual gap or 1000 missing lines;
it's more a matter of tracking how native files are written and read.  So
your problem will be addressed, and soon.  It's only a question of where
we draw the line, of what gets included in 0.62.  

If it's important to you to be working with an "official" (we don't have
offices) release, I'm willing to keep the train in the station until
native bcp works.  If you're OK working with snapshots, I say we (might)
let it go and look to the next one.  
--jkl