Re: [PATCHES] allow CSV quote in NULL - Mailing list pgsql-hackers

From Andrew Dunstan
Subject Re: [PATCHES] allow CSV quote in NULL
Date
Msg-id 46AF6987.1080508@dunslane.net
Whole thread Raw
In response to Re: [PATCHES] allow CSV quote in NULL  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: [PATCHES] allow CSV quote in NULL  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers

Tom Lane wrote:
> Stephen Frost <sfrost@snowman.net> writes:
>   
>> Other, unrelated, options being or not being there doesn't really have
>> any bearing on this though.  I'm not inventing new syntax here.  I'm
>> just removing a restriction on what the user can do that doesn't need
>> to exist.
>>     
>
> I don't think you're "just removing a restriction".  What you're doing
> is exposing a whole lot of strange and arguably broken corner cases.
> If we accept this patch I think we'll be fielding bug reports as a
> result for years to come.  I *especially* dislike the part about
> allowing the delimiter character in the null string --- that will allow
> people to complain about the order in which decisions are made.
>
>             
>   

Yeah, if you allow the delimiter in the null string, what do you do if 
it's not quoted? I can't imagine what the real world case for that could 
possibly be.

Even if there's an arguable case for allowing the quote char in a null 
string (and as I indicated upthread I really think the problem being 
addressed here could be solved in a far better fashion) there is surely 
no good case for allowing the delimiter. Oh, and if we did allow the 
quote char we should surely only allow it on input - just because other 
programs produce absurd output there is not reason we should.

cheers

andrew




pgsql-hackers by date:

Previous
From: "Simon Riggs"
Date:
Subject: Re: stats_block_level
Next
From: Stefan Kaltenbrunner
Date:
Subject: Re: Reducing Transaction Start/End Contention