Обсуждение: failure of \e in psql
I see that \e no longer works as expected: test=> select * from pg_class;...test=> \e and in the editor, the query is not appearing. The 'select' query should appear in the editor because I have not entered a non-backslash command to clear the query buffer. -- 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
On 1999-11-11, Bruce Momjian mentioned: > I see that \e no longer works as expected: > > test=> select * from pg_class; > ... > test=> \e > > and in the editor, the query is not appearing. The 'select' query > should appear in the editor because I have not entered a non-backslash > command to clear the query buffer. Well, once you send it, it's sent and gone. You have to edit it before you send it. I guess I'm not following you. Of course you should somehow be able to re-edit the previous query. Hmm. -- Peter Eisentraut Sernanders vaeg 10:115 peter_e@gmx.net 75262 Uppsala http://yi.org/peter-e/ Sweden
> On 1999-11-11, Bruce Momjian mentioned: > > > I see that \e no longer works as expected: > > > > test=> select * from pg_class; > > ... > > test=> \e > > > > and in the editor, the query is not appearing. The 'select' query > > should appear in the editor because I have not entered a non-backslash > > command to clear the query buffer. > > Well, once you send it, it's sent and gone. You have to edit it before you > send it. I guess I'm not following you. Of course you should somehow be > able to re-edit the previous query. Hmm. That is how it used to work. You run the query, get an error, and \e pulls it into the editor for fixing. -- 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
> On 1999-11-11, Bruce Momjian mentioned: > > > I see that \e no longer works as expected: > > > > test=> select * from pg_class; > > ... > > test=> \e > > > > and in the editor, the query is not appearing. The 'select' query > > should appear in the editor because I have not entered a non-backslash > > command to clear the query buffer. > > Well, once you send it, it's sent and gone. You have to edit it before you > send it. I guess I'm not following you. Of course you should somehow be > able to re-edit the previous query. Hmm. Peter, before I go hunting around, can you tell me any other things psql used to do that it doesn't do anymore? We had hand-tuned psql over the years, and it would be good to know what features no longer exist so we can decide if they are needed. -- 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
On Fri, 12 Nov 1999, Bruce Momjian wrote: > Peter, before I go hunting around, can you tell me any other things psql > used to do that it doesn't do anymore? Well, let's put it this way: Everythings that used to work, that people found useful, and that doesn't work anymore is a bug. That's what it's all about after all. However: About the \e thing I simply didn't know. The \p\g was removed for consistency. You might also be interested to know that \E no longer exists, because I couldn't make sense of it. Also \d* is slated for implementation but no one wanted to respond to my request to explain what this is actually supposed to do. That's all I can come up with right now. > We had hand-tuned psql over the years, and it would be good to know what > features no longer exist so we can decide if they are needed. Well, I really comes down to what Tom said, doesn't it: If the docs don't match the code, the code it wrong. And it will get fixed. A lot of those "tunings" seemed to be of the nature "If I put \o after \x I want it to do <foo> instead". That doesn't mean that they were bad of course, but the purpose of all of this was to put a consistent face on things. Having said that, if I mess it up I'll fix it of course. -- Peter Eisentraut Sernanders vaeg 10:115 peter_e@gmx.net 75262 Uppsala http://yi.org/peter-e/ Sweden
> On Fri, 12 Nov 1999, Bruce Momjian wrote: > > > Peter, before I go hunting around, can you tell me any other things psql > > used to do that it doesn't do anymore? > > Well, let's put it this way: Everythings that used to work, that people > found useful, and that doesn't work anymore is a bug. That's what it's all > about after all. > > However: About the \e thing I simply didn't know. The \p\g was removed for > consistency. You might also be interested to know that \E no longer > exists, because I couldn't make sense of it. Also \d* is slated for > implementation but no one wanted to respond to my request to explain what > this is actually supposed to do. That's all I can come up with right now. First, let me say I am very glad you overhauled psql. It was very needed, and I like the new functionality. Already learned \echo `date`. Quite handy and very flexible. I was just curious if there was any stuff you found confusing and skipped so we could comment on it all at once. We have fixed the pager off by default, and looks like \p\g and \e need work, but that is small compared to how much functionality does work perfectly. I personally found the \e and \p\g stuff very tricky to implement. [Of course, with the new output, I am going to have to re-do every one of my SQL query outputs for the book. :-) ] No idea what \d* does, nor \E. -- 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, I no longer see the pg_description descriptions when using the \do, \df, and \dT commands. The commands are much less useful without the descriptions. Seems \dd with a string is much smarter, and pulls descriptions based on string matching. Interesting that \dd shows descriptions of everything. Not sure how to recommend you change this. The new \df and \do displays are much clearer without the descriptions. It seems \df and \do show additional information about argument types and return values, while \dd shows comments. Maybe just add descriptions to \dT, and suggest people use \dd to get info about specific operators of functions? But that kind of messes the clarity of using \dd for descriptions. I am stumped. Maybe it can't be improved. -- 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
You can turn the descriptions on by typing \set description on (or \set description foo, or whatever, as long as it's something), for example, in your ~/.psqlrc (or in your .psqlrc-7.0.0 if you don't want to interfere with the current version). The reason for having descriptions off by default was that in a number of views (I recall functions and operators), they don't fit on the screen very nicely. On the other hand, the \dd command always shows descriptions, because it's sort of the built-in manual, but it doesn't show anything else (argument types, etc.). Read the fine (SGML) manual ;) -Peter On Sat, 13 Nov 1999, Bruce Momjian wrote: > Peter, I no longer see the pg_description descriptions when using the > \do, \df, and \dT commands. > > The commands are much less useful without the descriptions. Seems \dd > with a string is much smarter, and pulls descriptions based on string > matching. > > Interesting that \dd shows descriptions of everything. > > Not sure how to recommend you change this. The new \df and \do > displays are much clearer without the descriptions. It seems \df and > \do show additional information about argument types and return values, > while \dd shows comments. > > Maybe just add descriptions to \dT, and suggest people use \dd to get > info about specific operators of functions? But that kind of messes the > clarity of using \dd for descriptions. > > I am stumped. Maybe it can't be improved. > > -- Peter Eisentraut Sernanders vaeg 10:115 peter_e@gmx.net 75262 Uppsala http://yi.org/peter-e/ Sweden
> You can turn the descriptions on by typing \set description on (or \set > description foo, or whatever, as long as it's something), for example, in > your ~/.psqlrc (or in your .psqlrc-7.0.0 if you don't want to interfere > with the current version). OK. I noticed that the existance of an .psqlrc file causes an extra newline to be printed on startup before the first prompt. Is that intentional? > > The reason for having descriptions off by default was that in a number of > views (I recall functions and operators), they don't fit on the screen > very nicely. On the other hand, the \dd command always shows descriptions, > because it's sort of the built-in manual, but it doesn't show anything > else (argument types, etc.). Got it. Yes, much clearer for \df and \do. I noticed that using \set description on and then using \dT generates an error of:test=> \set description ontest=> \dTERROR: Relation 'p' does notexist Also, the \set commands don't seem to complain about bad commands: test=> \set figgletest=> Is that intentional? > > Read the fine (SGML) manual ;) > That was part of my problem. I hadn't figured out how to generate html from the sgml ref stuff. I just spent some time and figured out I have to issue the 'make' command from the upper sgml directory because there is no Makefile in the sgml/ref directory. I can view them fine now. -- 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