[freetds] [PATCH] make replacements more configurable

Craig A. Berry craigberry at mac.com
Wed Jan 29 18:16:30 EST 2003


At 12:57 AM 1/28/2003 -0700, Craig Davison wrote:
>On Mon, Jan 27, 2003 at 11:37:32PM -0600, Craig A. Berry wrote:

[forward compatibility scenario snipped]

>Sorry to take this further OT, but how can you be sure that future C libraries won't define 
>some other symbol that freetds defines? 

Nothing is 100% sure in this life but it seems quite unlikely the C 
standards committee would add functions that collide either with freetds_... 
or with one of the public APIs FreeTDS uses.  In any case we are only 
discussing functions that are already in common use but don't exist 
everywhere yet.

>Nothing can be infinitely forward-compatible.

Infinite isn't necessary.  A decade or two is probably sufficient and is not 
that difficult when designed in from the start.

>Maybe the solution (fictional ignorant COBOL programmer aside) is to get autoconf for VMS 
>(according to http://vms.gnu.org/tasks.html it will build and work) and send patches to 
>configure.in etc. That would be very valuable.

That's an excellent suggestion, but unfortunately the work you cite went 
dormant several years ago.  There is a current GNU on VMS project here:  

<http://www.openvms.compaq.com/opensource/opensource.html#gnv>

but I didn't get very far trying to run FreeTDS's Configure through it.  
It's probably worth checking back there in a year or so and seeing how far 
they've gotten.

>Consider that this project is targeted to UNIX.

It seems a pity to limit it arbitrarily when very trivial changes can remove 
that restriction.




More information about the FreeTDS mailing list