Re: psql feature thought
От | Joshua D. Drake |
---|---|
Тема | Re: psql feature thought |
Дата | |
Msg-id | 44696379.2070501@commandprompt.com обсуждение исходный текст |
Ответ на | Re: psql feature thought (Tom Lane <tgl@sss.pgh.pa.us>) |
Список | pgsql-hackers |
Tom Lane wrote: > Michael Glaesemann <grzm@seespotcode.net> writes: >> What use case are you envisioning? Just saving ... *counts ... 14 >> keystrokes in this case? (I'm not saying there *isn't* a use case. I >> just don't see a big benefit here.) > > Quite aside from the compatibility and how-useful-is-it-really > arguments, I think this'd be a bad idea in the abstract. SQL is not one > of those languages that assigns semantic significance to the shape of > whitespace [1]. We should NOT introduce any such concept into psql, > because it'd fundamentally break the lexical structure of the language. > To take just one example of the difficulties you'd get into, consider > > select 1; select > 2; select 3; > > How many transactions is that exactly? And on what grounds are you > deciding? Well, my use case was that I was doing a quick batch process that had about half a dozen statements in a row that were insert/update/delete. If I had forgotten to type by begin; , I could have a problem with consistency of various things if one in the middle failed. I understand that this is completely a sugar feature and I am not actually arguing for as much as I am saying, "Hey this was an interesting thought". > > (No, I don't like python. Why do you ask? ;-)) Heh... I happen to like Python quite a bit: But even in Python you can do: select 1; select 2; select 3; From an implementation perspective I would suggest that if the line has a carriage return then it doesn't behave as a single transaction.. e.g; BEGIN; SELECT 1; SELECT 2; COMMIT; Would be the same as: SELECT 1; SELECT 2; But not the same as: SELECT 1; SELECT 2; Egad... forget I even mentioned it... These looks like a whole lot of pgsql-novice posts having to be answerd. Thanks for entertaining me though. Sincerely, Joshua D. Drake > > regards, tom lane > > [1] Mostly. There's the infamous continued-string-literal construct... > -- === The PostgreSQL Company: Command Prompt, Inc. === Sales/Support: +1.503.667.4564 || 24x7/Emergency: +1.800.492.2240 Providing the most comprehensive PostgreSQL solutions since 1997 http://www.commandprompt.com/
В списке pgsql-hackers по дате отправления: