FreeTDS Development Group

["Eric Deutsch" <edeutsch AT systemsbiology.org>]

Okay, well, I'm all for keeping it this way then.

I've found the first test case where the old [cplen-1] would have
worked, but I don't think we should change it.  It occurs with ODBC and
it appears that when SELECTing TEXT under DBD::ODBC, srclen is one too
large.  I don't know why that is, but since tds_convert_text() is being
lied to about how long src[] is, it shouldn't be at fault.

Anyway, I'd vote for checking this in as is and seeing what happens.

What do you mean, funny?  ;-)



> -----Original Message-----
> From: Lowden, James K [mailto:LowdenJK AT bernstein.com]
> Sent: Tuesday, July 09, 2002 12:56 PM
> To: TDS Development Group
> Subject: [freetds] RE: Losing last character of TEXT fields
> 
> > From: Eric Deutsch [mailto:edeutsch AT systemsbiology.org]
> > Sent: July 9, 2002 1:43 PM
> >
> >       case SYBTEXT:
> >       case SYBCHAR:
> >       case SYBBINARY:
> >          cplen = srclen > destlen ? destlen : srclen;
> >          memcpy(dest, src, cplen);
> >          return cplen;
> 
> Eric,
> 
> Your remarks are on the money.  Please ignore my last message.  I now
like
> your "radical" solution best, because it's both simple and funny.
> 
> I should have looked at cs.c, at cs_convert().  I was fixing dblib in
> tdslib, with the consequence of breaking ctlib.  The weird termination
> behavior described in the dblib docs shouldn't be embedded in a shared
> library.  dbconvert() is what needs attention.
> 
> Brian, I'd like to suggest a policy about conversions for the project;
if
> you like it then it should be explicitly noted somewhere.
> 
> <policy>
> tds_convert copies from src to dest and returns the output data
length,
> period.  All padding and termination is the responsibilty of the API
> library
> and is done post conversion.  There may be some shared utility padding
&
> termination functions, but they're really not "conversion".  Trying to
do
> anything nonstraightforward in tds_convert will only confuse and
confound
> the higher levels, us among them.
> </policy>
> 
> An alternative would be to pass the padding & termination policy
argument
> as
> an enumerated type, which tds_convert would carry out if the basic
> conversion succeeds.
> 
> You have more experience with this (obviously), and have some idea
where
> UTF-8 is taking us, so I'm curious how you see things.
> 
> --jkl
> 
> ---
> You are currently subscribed to freetds as:
[edeutsch AT systemsbiology.org]
> To unsubscribe, forward this message to leave-freetds-
> 127899P AT franklin.oit.unc.edu