FreeTDS Development Group

["Thompson, Bill D (London)" <ThompBil AT exchange.uk.ml.com>]

FYI,

I've now proven:

SYBCHAR ->

SYBCHAR with various destlen's (dblib -1 & -2 etc)
SYBBINARY
SYBINT4
SYBFLT8
SYBNUMERIC (specifying precision & scale as discussed)

SYBFLT8 ->

SYBCHAR
SYBINT4
SYBINT2
SYBINT1
SYBNUMERIC (specifying precision & scale as discussed)
SYBMONEY

SYBNUMERIC ->

SYBCHAR
SYBFLT8



Bill


> -----Original Message-----
> From: ZIGLIO Frediano [SMTP:Frediano.Ziglio AT vodafoneomnitel.it]
> Sent: Friday, August 30, 2002 3:46 PM
> To:   TDS Development Group
> Subject:      [freetds] RE: conversion status
> 
> > 
> > All, 
> > 
> > I tested the dbconvert() tonight.  I thought you might be 
> > interested in
> > what I found.  
> > 
> > Executive summary: B+.  No segfaults.  Mostly correct 
> > outcomes when the
> > return code indicates success.  Unexotic conversions mostly succeed. 
> > Failures trigger the error handler with oh-so-helpful messages.  
> > 
> > You can reproduce my results if you like, by looking at this page:
> > 
> > http://www.SchemaMania.org/freetds/cvs/freetds/src/dblib/unittests/
> > 
> > You need: convert_acid.c and eg_handlers.h
> > also:         convert_acid.make (a shell script)
> > and   convert_acid.output (my results)
> > and   killer_query.sql 
> > 
> > The query returns one row of 18 datatypes.  The program 
> > executes a query
> > (from stdin) and attempts to convert each value in the result 
> > set to every
> > known datatype.  It consults dbwillconvert() first, and 
> > doesn't attempt to
> > convert incompatible types.  
> > 
> > Anyone think any of this should be in cvs?
> > 
> > == the results (with TDS 5.0) ==
> > 
> > Forget Binary, image, numeric, and decimal.  They're mostly broken.  
> > 
> > input               status  comment
> > =====               ======  =======
> > char                OK      No numeric/decimal, no money. 
> 
> char use same function of varchar/text... quite strange
> 
> > text, varchar       Perfect!
> > bit         OK      No numeric/decimal. 
> > tinyint             OK      No numeric, image.
> > int, smallint       OK      No numeric/decimal, image/binary.
> > real, float fair    No numeric, image, integers; maybe money.  
> 
> Added integers and fixed money
> 
> > datetime    OK      No image/binary.
> > smalldatetime       OK      No image/binary.
> > money               fair    No numeric, image, integers.  
> > money               wrong   float & real yield -283649471676416.
> 
> Fixed
> 
> > smallmoney  OK      No numeric/decimal, image/binary.       
> > smallmoney  odd     201203.04 -> real yields 201203.046875 
> > numeric             broken  char/varchar only. 
> > decimal             broken  char/varchar, float/real only.
> > timestamp   broken  char/varchar only. 
> 
> willconvert should return false for many types...
> 
> > image,varbinary     broken  char, text, binary only.  
> > 
> 
> Do you take in account that many test fail for overflow/underflow reason ?
> 
> > == conclusions ==
> > 
> > 0.  Everything can be made into a char/varchar!
> > 1.  Money needs attention.  Should convert to int, should correctly
> > convert to floating point.  This is just a debugging effort; 
> > the code is
> > there.  
> 
> Fixed. See other mail for money.
> 
> > 2.  What happened to float -> int? 
> 
> Added.
> 
> > 3.  Numeric/decimal is not ready.  
> > 
> > As far as I'm concerned, 1 & 2 have to work in 0.60.  Numeric 
> > is the other
> > big hole, but I don't consider it essential.  I know, I've 
> > many times said
> > "any<->any in 0.60".  I wasn't thinking about legal weird 
> > conversions like
> > "datetime<->image".  If we crossed those three items off the list (as
> > done), I'd consider the job complete.  
> > 
> 
> So we are ready for a pre-release ?
> 
> > And, if someone wants to make numeric work in 0.60, you've still got a
> > couple of weeks! 
> > 
> 
> freddy77
> 
> =================================
> "STRICTLY PERSONAL AND CONFIDENTIAL
> 
> This message may contain confidential and proprietary material for the
> sole
> use of the intended recipient. Any review or distribution by others is
> strictly prohibited. If you are not the intended recipient please contact
> the sender and delete all copies.
> The contents of this message that do not relate to the official business
> of
> our company shall be understood as neither given nor endorsed by it."
> 
> =================================
> 
> ---
> You are currently subscribed to freetds as: [thompbil AT exchange.uk.ml.com]
> To unsubscribe, forward this message to
> $subst('Email.Unsub')