FreeTDS Development Group

["Brian Bruns" <camber AT ais.org>]

There is a change that has to be made to run on alpha (by the way, from my 
recollection alpha does not bus error on unaligned access but it is simply 
slower than aligned access) the FAQ lists the exact change (two typedefs in 
tds.h).  so that's not much help.

If you want to test this there is a line setting row_size in 
tds_process_col_info(), tds_process_results() and tds7_process_results (TDS 
4.2, 5.0, and 7.0 respectively).  change the 4 to 8 there and see if that 
is it or not (i doubt it, but worth the try).

Brian 

On 01/25/00, "Dave Blob <blob AT saghill.com>" wrote:
> Could it be that solaris 7, while being a 32-bit OS, is 64-bit
> internally, and needs to be 8 byte aligned?  Does anyone use freetds on
> an alpha?
> 
> Another problem in running the sybperl tests seems to be that failed
> queries don't come back that way.  Is anyone using freetds and sybperl?
> 
> Thanks,
> - Dave
> 
> -----Original Message-----
> From: Brian Bruns [mailto:camber AT umcc.ais.org]
> Sent: Monday, January 24, 2000 9:51 PM
> To: TDS Development Group
> Subject: [freetds] Re: Minor issues...
> 
> 
> 
> Hmmm...dbdata() (I can only speak to the C library side of things)
> returns
> the pointer directly from the row buffer (resinfo->current_row) which is
> 4
> byte aligned (see function tds_process_col_info() for an example,
> although all 3 result functions do it).  
> 
> Someone reported this as only happening with Solaris 2.7 (2.6 works
> apparently).
> 
> Would it be possible to do a stack trace where it dumps?
> 
> Brian
> 
> On Mon, 24 Jan 2000, Dave Blob wrote:
> 
> > Has anyone had problems with alignment issues and sybperl?  In DBlib,
> it
> > looks like things come back from dbdata(), and sybperl expects them to
> > be castable to DBINTs, etc, which apparently they often are not.
> > 
> > This is on a solaris 2.7 box.  Is this a freetds or a sybperl bug?
> I'm
> > inclined to think the latter...
> > 
> > Thanks,
> > Dave
> > 
> > -----Original Message-----
> > From: Brian Bruns [mailto:camber AT umcc.ais.org]
> > Sent: Monday, January 24, 2000 8:28 PM
> > To: TDS Development Group
> > Subject: [freetds] Re: Minor issues...
> > 
> > 
> > 
> > 
> > On Mon, 24 Jan 2000, Dave Blob wrote:
> > 
> > > Hi everyone...
> > > 
> > > I'm fairly new here, so forgive me if this is all known.
> > Additionally, I'm 
> > > happy to work on a patch for what I'm pointing out, I just haven't
> > gotten 
> > > to it yet.
> > > 
> > > 1) "make check" though documented, doesn't do anything
> > 
> > Actually that's the genreic autoconf instructions.  One day those will
> > be
> > removed (once somebody writes a good install file that is :-)
> > 
> > > 2) "make test" does what "make check" is probably meant to do, but
> > only 
> > > after the entire package has been installed (so far as I can tell).
> > I thought this was fixed in CVS (to use .libs/* that is), but I'm not
> > sure.
> > 
> > > 3) On solaris, the test directories seem to get compiled sans "-lnsl
> 
> > > -lsocket"
> > Hm...I'll look at that.
> > 
> > > 4) It's really easy to get "Unknown marker:0!" with the 7.0 support,
> > so for 
> > > MS SQL, I just switched to 4.2, which seems ok...
> > TDS 7.0 support is rather new and I am temporarily without access to
> MS
> > SQL, so it's likely to be that way until I get a MS SQL setup (working
> > on
> > it)
> > 
> > > Anyhow, I've just gotten things compiled, but figured the feedback
> may
> > help 
> > > others.  Like I said, I'll try to send in a patch for some of it.
> > > 
> > > Thanks,
> > > Dave
> > > 
> > 
> > 
> > ---
> > You are currently subscribed to freetds as: blob AT saghill.com
> > To unsubscribe, forward this message to
> > $subst('Email.Unsub')
> > 
> > ---
> > You are currently subscribed to freetds as: camber AT ais.org
> > To unsubscribe, forward this message to
> $subst('Email.Unsub')
> > 
> > 
> 
> 
> ---
> You are currently subscribed to freetds as: blob AT saghill.com
> To unsubscribe, forward this message to
> $subst('Email.Unsub')