Обсуждение: unique indexes and pqlib

Two seperate questions..

1) Is there any reason why I can't create a unique index on an int8?

   It tells me this:
    ERROR:  Can't find a default operator class for type 20.


2) Are there any more changes expected in the code for libpq?

Vince.
--
==========================================================================
Vince Vielhaber -- KA8CSH   email: vev@michvhf.com   flame-mail: /dev/null
       # include <std/disclaimers.h>                   TEAM-OS2
   Online Searchable Campground Listings    http://www.camping-usa.com
       "There is no outfit less entitled to lecture me about bloat
               than the federal government"  -- Tony Snow
==========================================================================

Vince Vielhaber <vev@michvhf.com> writes:
> 1) Is there any reason why I can't create a unique index on an int8?
>    It tells me this:
>     ERROR:  Can't find a default operator class for type 20.

int8 is new and not very thoroughly supported yet.  Probably there
just isn't support for making an index on it (Thomas?).

> 2) Are there any more changes expected in the code for libpq?

Ever?  Yes.  Before 6.4 release?  Perhaps not, but I wouldn't swear to
it.  We are at a bug-fixes-only stage of the release cycle, but I do not
promise there are no bugs in libpq.

Why do you care, exactly?  You shouldn't need to worry whether a
library you use is changing internally, as long as the API is stable
(and I certainly hope we do not find bugs that force an API change
at this point...)

            regards, tom lane

On 18-Oct-98 Tom Lane wrote:
> Vince Vielhaber <vev@michvhf.com> writes:
>> 2) Are there any more changes expected in the code for libpq?
>
> Ever?  Yes.  Before 6.4 release?  Perhaps not, but I wouldn't swear to
> it.  We are at a bug-fixes-only stage of the release cycle, but I do not
> promise there are no bugs in libpq.
>
> Why do you care, exactly?  You shouldn't need to worry whether a
> library you use is changing internally, as long as the API is stable
> (and I certainly hope we do not find bugs that force an API change
> at this point...)

I'm doing the libpq and psql ports for OS/2.

Vince.
--
==========================================================================
Vince Vielhaber -- KA8CSH   email: vev@michvhf.com   flame-mail: /dev/null
       # include <std/disclaimers.h>                   TEAM-OS2
   Online Searchable Campground Listings    http://www.camping-usa.com
       "There is no outfit less entitled to lecture me about bloat
               than the federal government"  -- Tony Snow
==========================================================================

> > 1) Is there any reason why I can't create a unique index on an int8?
> >    It tells me this:
> >     ERROR:  Can't find a default operator class for type 20.
> int8 is new and not very thoroughly supported yet.  Probably there
> just isn't support for making an index on it (Thomas?).

I'll try to look at it. But I'm starting to get swamped with last-minute
big additions to the documentation, and will have only limited time for
other stuff.

btw, I haven't looked yet but I hope the odbc makefile stuff still can
build standalone. Hopefully we are converging on an acceptable set of
Makefiles; I'd made changes recently - which you'd noticed broke
installations which did not do major.minor version numbering on shared
libraries - in response to major breakage introduced for some other
reason which damaged installations _with_ major.minor version numbering.
Though despite appearances my response wasn't intended to rebreak the
other systems :/
                      - Tom

"Thomas G. Lockhart" <lockhart@alumni.caltech.edu> writes:
> btw, I haven't looked yet but I hope the odbc makefile stuff still can
> build standalone.

I didn't intentionally break that --- in fact I added some stuff to
odbc's private configure script to ensure it would still support the
Makefile.  But I don't know enough about ODBC to test it, so I'd prefer
to let someone who knows what it's for make sure that it still builds
and works.  (I can only say that it does now build and install a valid
library on HPUX, which it didn't before.)

> Hopefully we are converging on an acceptable set of
> Makefiles; I'd made changes recently - which you'd noticed broke
> installations which did not do major.minor version numbering on shared
> libraries - in response to major breakage introduced for some other
> reason which damaged installations _with_ major.minor version
> numbering.

Right.  Actually the recent changes were very good --- they made it
feasible to have a large chunk of the sharedlib build rules be textually
identical in different makefiles.  I just took the next logical step
and moved the textually identical stuff into a single file.  Should be
a lot easier to maintain this way.
        regards, tom lane

> 
> TwoBB seperate questions..
> 
> 1) Is there any reason why I can't create a unique index on an int8?
> 
>    It tells me this: 
>     ERROR:  Can't find a default operator class for type 20.

This will be fixed in 6.5, due to go into beta soon.



--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026