Tom Lane wrote:

Patrick Samson <[EMAIL PROTECTED]> writes:
Attribute names beginning with a dot are filtered
just in one place, in pltcl_trigger_handler().
(version 7.3.5)

I am not sure why that code is there. It is *not* there to prevent the loop from touching dropped attributes, because the same code is in the original 1.1 version of pltcl.c, long before we could drop attributes. Jan, do you remember why you put this into pltcl_trigger_handler()?

     * Ignore pseudo elements with a dot name
    if (*(ret_values[i]) == '.') {
        i += 2;

It's not documented behavior that I can see, and it doesn't seem to have
any use other than making pltcl triggers fail if a user chooses a field
name starting with a dot :-(

right, this is documented nowhere :-(

When assigning a tuple to an array, PL/Tcl creates one extra array element .tupno telling the SPI_tuptable index of the result tuple. I think I originally planned to have more of these critters ... but probably never really needed them. It is in there since 6.3!

Bottom line is, if one has a trigger, and inside the trigger he does an SPI_exec, fetches a tuple into an array and then returns [array get x] instead of new or old ... so from the back through the right chest into the left eye ... then it will fail if the .tupno isn't filtered out.


Attached is a patch to :
- Add a filter in two other places, in relation
  with the mentioned error message:

This is already done in 7.4, although for some reason pltcl_trigger_handler got overlooked - I will fix that.

- Add the same filter in the build of TG_relatts.
  This will prevent a tcl script which loops on
  TG_relattrs to fail in trying to use a dropped

This is deliberately *not* done in 7.4, because it would break the documented behavior of TG_relatts:


A Tcl list of the table column names, prefixed with an empty list
element. So looking up a column name in the list with Tcl's lsearch command returns the element's number starting with 1
for the first column, the same way the columns are customarily numbered in PostgreSQL.

I think we need to preserve the relationship to column numbers.  People
who just want a list of the live columns can get it from the OLD or NEW

regards, tom lane

# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== [EMAIL PROTECTED] #

---------------------------(end of broadcast)--------------------------- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly

Reply via email to