Обсуждение: Command-line client sometimes misinterprets tabs for nulls
Let's say I have typed a nice formatted SQL statement like so:
select b1.blah,
b1.blah2,
b2.blah3
from blah_table as b1
inner join only blah_table2 as b2
on b1.blah =3D b2.blah;
...using tabs as the indenting medium. Now I copy that from my editor into=
=20
psql, and most of the tabs are converted to nulls, resulting in an error li=
ke=20
"Unknown value fromblah_table". Tabs should paste; they're normal=20
characters.
--=20
Casey Allen Shobe / Network Security Analyst & PHP Developer
SecureWorks, Inc. / 404.327.6339 x169 / Fax: 404.728.0144
cshobe@secureworks.net / http://www.secureworks.net
Content is my own and does not necessarily represent my company.
"Diagoras the Atheist once visited Samothrace and a friend there said to hi=
m,=20
'You think the gods have no care for humans? Why, you can see from all the=
se=20
votive pictures here how many people have escaped the fury of storms at sea=
=20
by praying to the gods, who have brought them safe to harbour.' 'Yes,=20
indeed,' said Diagoras, 'but where are the pictures of all those who suffer=
ed=20
shipwreck and perished in the waves?' On another occasion he was on a voya=
ge=20
and the crew became anxious about the bad weather and began to mutter that =
it=20
served them right for taking an atheist on board. Diagoras just pointed ou=
t=20
to them a number of other ships on the same course which were in equal=20
difficulties and asked them whether they thought that there was a Diagoras =
on=20
the passenger-list of every one of them. The fact is that a man's characte=
r=20
or way of life makes no difference at all to his good luck or bad."
- Cicero
Casey Allen Shobe <cshobe@secureworks.net> writes:
> Tabs should paste; they're normal characters.
Turn off readline (psql -n). For someone accustomed to tab completion,
the above is about as sensible as demanding we disable backspace.
It would be nice if we could distinguish pasted tabs from manually-typed
tabs, but I'm not aware of any way that psql could tell the difference.
regards, tom lane
On Tuesday 27 August 2002 01:48 am, Tom Lane wrote: > Turn off readline (psql -n). For someone accustomed to tab completion, > the above is about as sensible as demanding we disable backspace. Right, my original build of postgres did not have tab completion support as= I=20 did not have readline (a notice at the end of the configure script would ha= ve=20 been nice, maybe a summary of what will and won't be supported?). However,= I=20 love the command-line history, so readline was a must ;). > It would be nice if we could distinguish pasted tabs from manually-typed > tabs, but I'm not aware of any way that psql could tell the difference. Can tab completion be disabled without disabling command line history? I= =20 noticed -n disables both. I would gladly trade tab completion for tab=20 pastability, but I rely heavily on command line history. Thanks, --=20 Casey Allen Shobe / Network Security Analyst & PHP Developer SecureWorks, Inc. / 404.327.6339 x169 / Fax: 404.728.0144 cshobe@secureworks.net / http://www.secureworks.net Content is my own and does not necessarily represent my company.
Casey Allen Shobe <cshobe@secureworks.net> writes:
> Can tab completion be disabled without disabling command line history?
Get out the readline man page. I believe you can configure it six
ways from Sunday, but I've never tried myself ...
regards, tom lane
The problem is that most items in postgresql are tab completed. Cut and past your text into a bash or zsh prompt. It injects a bunch of filenames into your data, as per tab completion. Annoying, but for the benefits it's livable. Perhaps a toggle could be created to disable tab completion? On Mon, 2002-08-26 at 23:36, Casey Allen Shobe wrote: > Let's say I have typed a nice formatted SQL statement like so: > > select b1.blah, > b1.blah2, > b2.blah3 > from blah_table as b1 > inner join only blah_table2 as b2 > on b1.blah = b2.blah; > > ...using tabs as the indenting medium. Now I copy that from my editor into > psql, and most of the tabs are converted to nulls, resulting in an error like > "Unknown value fromblah_table". Tabs should paste; they're normal > characters. > > -- > Casey Allen Shobe / Network Security Analyst & PHP Developer > SecureWorks, Inc. / 404.327.6339 x169 / Fax: 404.728.0144 > cshobe@secureworks.net / http://www.secureworks.net > Content is my own and does not necessarily represent my company. > > "Diagoras the Atheist once visited Samothrace and a friend there said to him, > 'You think the gods have no care for humans? Why, you can see from all these > votive pictures here how many people have escaped the fury of storms at sea > by praying to the gods, who have brought them safe to harbour.' 'Yes, > indeed,' said Diagoras, 'but where are the pictures of all those who suffered > shipwreck and perished in the waves?' On another occasion he was on a voyage > and the crew became anxious about the bad weather and began to mutter that it > served them right for taking an atheist on board. Diagoras just pointed out > to them a number of other ships on the same course which were in equal > difficulties and asked them whether they thought that there was a Diagoras on > the passenger-list of every one of them. The fact is that a man's character > or way of life makes no difference at all to his good luck or bad." > > - Cicero > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) >
> Can tab completion be disabled without disabling command line history? I > noticed -n disables both. I would gladly trade tab completion for tab > pastability, but I rely heavily on command line history. Perhaps annoying, but a simple work around is to use <space><tab><space><tab> Formats like tab in the code, but space/tab combinations don't complete to anything, registering like a space in psql.
Casey Allen Shobe writes: > Can tab completion be disabled without disabling command line history? I > noticed -n disables both. I would gladly trade tab completion for tab > pastability, but I rely heavily on command line history. man psql, search for "disable" Also, you might find it easier to start an editor using \e and paste your code there. -- Peter Eisentraut peter_e@gmx.net