Обсуждение: Re: [COMMITTERS] pgsql: Document struct/union problem with pgindent.

Bruce Momjian wrote:

+ #
+ # Structure/union pointers in function prototypes and definitions have an extra
+ # space after the asterisk:
+ #
+ #    void x(struct xxc * a);


I know we should not be driven by our tools, but is there a case for a 
coding standard that requires use of a typedef name here?

cheers

andrew

Andrew Dunstan wrote:
>
>
> Bruce Momjian wrote:
>
> + #
> + # Structure/union pointers in function prototypes and definitions have an extra
> + # space after the asterisk:
> + #
> + #    void x(struct xxc * a);
>
>
> I know we should not be driven by our tools, but is there a case for a  
> coding standard that requires use of a typedef name here?

We use things like struct timeval and struct tm ... I don't think we
should be creating typedefs for those, so it would be good to find a
more general solution.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

Andrew Dunstan <andrew@dunslane.net> writes:
> Bruce Momjian wrote:
> + # Structure/union pointers in function prototypes and definitions have an extra
> + # space after the asterisk:
> + #
> + #    void x(struct xxc * a);

> I know we should not be driven by our tools, but is there a case for a 
> coding standard that requires use of a typedef name here?

I don't think so.  Shall we artificially create a typedef for standard
objects like "struct stat" in order to follow such a coding rule?  I
think that's just a recipe for confusion.

In any case, a big fraction of the places that have this issue are code
that we've imported from elsewhere (the regex engine, zic) and changing
to typedefs would mean even more drift from upstream and hence
difficulty in following their patches.

It's just a bug in pgindent that we should try to fix sometime.
        regards, tom lane

Alvaro Herrera wrote:
> Andrew Dunstan wrote:
>   
>> Bruce Momjian wrote:
>>
>> + #
>> + # Structure/union pointers in function prototypes and definitions have an extra
>> + # space after the asterisk:
>> + #
>> + #    void x(struct xxc * a);
>>
>>
>> I know we should not be driven by our tools, but is there a case for a  
>> coding standard that requires use of a typedef name here?
>>     
>
> We use things like struct timeval and struct tm ... I don't think we
> should be creating typedefs for those, so it would be good to find a
> more general solution.
>
>   

Good point.

cheers

andrew

Tom Lane wrote:
> Andrew Dunstan <andrew@dunslane.net> writes:
> > Bruce Momjian wrote:
> > + # Structure/union pointers in function prototypes and definitions have an extra
> > + # space after the asterisk:
> > + #
> > + #    void x(struct xxc * a);
> 
> > I know we should not be driven by our tools, but is there a case for a 
> > coding standard that requires use of a typedef name here?
> 
> I don't think so.  Shall we artificially create a typedef for standard
> objects like "struct stat" in order to follow such a coding rule?  I
> think that's just a recipe for confusion.
> 
> In any case, a big fraction of the places that have this issue are code
> that we've imported from elsewhere (the regex engine, zic) and changing
> to typedefs would mean even more drift from upstream and hence
> difficulty in following their patches.
> 
> It's just a bug in pgindent that we should try to fix sometime.

I would hope switching to GNU indent would fix this, but might bring new
bugs.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +