Folks,

Need soem review sorting what appears to be an oversight or something....

courier ports and tarballs appear to contain some errors in
.../authlib/README.authpostgres.html, correctable as follows, (Subject to
review):

1) As Reads:
id varchar(128) DEFAULT ' ' NOT NULL PRIMARY KEY,

Is Amended to read:

id varchar(128) DEFAULT ' ' NOT NULL,

Add after rthe last line  .....  'quota.......... NOT NULL,

CONSTRAINT id PRIMARY KEY (id)

- (with no trailing comma)

REASON: Placed per the existing README.aithpostgres.html suggestion, the
constraint will create an implicit primary key named 'passwd-pkey'.  This
may not give the same results that appear to be expected, i.e. a primary
key named 'id' is created in the MySQL module, which is referred to as of
interest to the postgresql module.  The suggested fix instead creates
primary key: id in postgresql also.

(which one of these keys is expected by the calling courtier-mta module is
yet To Be Determined - but this makes it look like MySQL).  Sam?

2) If the above fix is NOT appropriate, the trailing comma after 'quota
..... NOT NULL," must be removed.  Postgresql will not accept a comma
after the *last* entry in the list of columns.

3) As Reads:

  'int(10) unsigned DEFAULT...'  .... in the two rows establishing uid and
gid

Is Amended To Read:
  integer DEFAULT ...'

I can find no mention of such a data type as 'int(10)', nor an integer
type with the range sought, nor a constraint that (directly) permits
limiting it to a positive value.

smallint, int2 are -32768 to +32767, which will not hold the suggested
default value.

integer, int4 are -2147483648 to + 2147483647, which will (and more
besides) but may not map to the size of the variable it is to be
transferred to/from.

I don't suppose anyhting else is going to place into, or expect from
either uid or guid  values in excess of 65K or less than zero, BUT a
CONSTRAINT CHECK.... (with overhead...) could be applied that check for
such bounds in another column or a separate table if it is a concern.

Action:

Need confirmation that 'id', not 'pkey_passwd' is the name wanted for the
primary key.

Need confirmation that working with integer / int4 'as is' will not pose a
problem.

Regards,

Bill Hacker
-- 



-------------------------------------------------------
This SF.net email is sponsored by:
The Definitive IT and Networking Event. Be There!
NetWorld+Interop Las Vegas 2003 -- Register today!
http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en
_______________________________________________
courier-users mailing list
[EMAIL PROTECTED]
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users

Reply via email to