Обсуждение: psql updates

I hope to have fixed all psql bugs that came up in the last month of my
absence. (The array syntax issue itself is not included.) I'd particularly
be interested whether the readline related compilation problem is gone,
since by readline's CHANGELOG I cannot decode when or where the problem
was introduced or removed. In addition, the frequent end user problem "I
did \dt and all my tables were gone" has been eliminated.

-- 
Peter Eisentraut                  Sernanders väg 10:115
peter_e@gmx.net                   75262 Uppsala
http://yi.org/peter-e/            Sweden

> was introduced or removed. In addition, the frequent end user problem "I
> did \dt and all my tables were gone" has been eliminated.

Huge fix for us.  Great.

--  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
 

Peter Eisentraut <peter_e@gmx.net> writes:
> ... I'd particularly be interested whether the readline related
> compilation problem is gone, since by readline's CHANGELOG I cannot
> decode when or where the problem was introduced or removed.

It was still there, but I fixed it.  Given the lack of any clear version
info for libreadline, adding a configure-time test seems to be the
way to go.

The particular problem I saw was that the exported variable
rl_completion_append_character doesn't exist in old versions of
libreadline.  (How old?  I dunno, but a RedHat 4.2 box I have access to
has a libreadline that's like that.)  I arranged to #ifdef out psql's
attempt to set the variable unless configure sees that the variable
is declared in <libreadline.h>.  With that change, psql builds
successfully against that libreadline version.  The tab-completion
behavior seems a little flaky (if you press tab when you don't have
a partial keyword typed, it wipes out whatever word you do have typed)
but I doubt it is worth trying to fix that.  I'm satisfied if psql
builds and is usable --- anyone who complains about the tab behavior
can be told they need a newer libreadline.
        regards, tom lane

On 2000-01-22, Tom Lane mentioned:

> The tab-completion
> behavior seems a little flaky (if you press tab when you don't have
> a partial keyword typed, it wipes out whatever word you do have typed)

In some cases this might be intentional. For example, when you enter
=> insert xx<tab>
then the xx gets replaced by INTO because it's the only valid thing to put
there anyway. If you observed someting different, then I'd be interested
in looking at it.


-- 
Peter Eisentraut                  Sernanders väg 10:115
peter_e@gmx.net                   75262 Uppsala
http://yi.org/peter-e/            Sweden

Peter Eisentraut <peter_e@gmx.net> writes:
> On 2000-01-22, Tom Lane mentioned:
>> The tab-completion
>> behavior seems a little flaky (if you press tab when you don't have
>> a partial keyword typed, it wipes out whatever word you do have typed)

> In some cases this might be intentional. For example, when you enter
> => insert xx<tab>
> then the xx gets replaced by INTO because it's the only valid thing to put
> there anyway. If you observed someting different, then I'd be interested
> in looking at it.

What I see with this ancient libreadline is
SELECT zz<tab>

zz is wiped out and replaced by a single space.  However, this does
*not* happen with more modern readlines; and since I don't even know
where to get source code corresponding to the readline that's on this
old Linux system, I doubt it's worth worrying about.
        regards, tom lane