Re: Add on_trusted_init and on_untrusted_init to plperl UPDATED [PATCH]

On Tue, Feb 2, 2010 at 10:46 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Tue, Feb 2, 2010 at 10:42 PM, Alex Hunsaker <badalex(at)gmail(dot)com> wrote:
>> On Sat, Jan 30, 2010 at 08:49, Tim Bunce <Tim(dot)Bunce(at)pobox(dot)com> wrote:
>>> This is an update the fourth of the patches to be split out from the
>>> former 'plperl feature patch 1'.
>>>
>>> Changes in this patch:
>>>
>>> - Adds plperl.on_trusted_init and plperl.on_untrusted_init GUCs
>>>    on_trusted_init is PGC_USERSET, on_untrusted_init is PGC_SUSET
>>
>> Im not a fan of the names (I think everyone gets trusted vs untrusted
>> confused). May I humbly suggest:
>> plperl.on_init
>> plperlu.on_init
>> plperl.both_on_init <- this one is the one that throws the scheme off :(
>>
>>>    SPI functions are not available when the code is run.
>>
>> Hrm, we might want to stick why in the docs or as a comment somewhere.
>> I think this was the main concern?
>>
>> * We call a plperl function for the first time in a session, causing
>>   plperl.so to be loaded.  This happens in the context of a superuser
>>   calling a non-superuser security definer function, or perhaps vice
>>   versa.  Whose permissions apply to whatever the on_load code tries
>>   to do?  (Hint: every answer is wrong.)
>>
>>> - select_perl_context() state management improved
>>>    An error during interpreter initialization will leave
>>>    the state (interp_state etc) unchanged.
>>
>> This looked good.
>>
>>> - The utf8fix code has been greatly simplified.
>>
>> Yeah to the point that it makes me wonder if the old code had some
>> reason to spin up the FunctionCall stuff.  Do you happen to know?
>> Looks good to me otherwise.
>>
>> The tests dont seem to pass :( this is from a make installcheck-world
>> test plperl_shared        ... FAILED
>> ...
>> test plperl_init          ... FAILED
>>
>> with:
>>  SET plperl.on_trusted_init = '$_SHARED{on_init} = 42';
>> + ERROR:  unrecognized configuration parameter "plperl.on_trusted_init"
>>  -- test the shared hash
>>
>> If I throw a LOAD 'plperl'; at the top of those sql files it works...
>>
>> The only quibble I have with the docs is:
>> +      If the code fails with an error it will abort the initialization and
>> +      propagate out to the calling query, causing the current transaction or
>> +      subtransaction to be aborted. Any changes within the perl won't be
>> +      undone.  If the <literal>plperl</> language is used again the
>> +      initialization will be repeated.
>>
>> Instead of "Any changes within the perl won't be undone".  Maybe
>> "Changes to the perl interpreter will not be undone" ?
>
> With all due respect.... yuck.

OK, third time is the charm. Sigh. The "yuck" was in reference
specifically to the proposed GUC names.

I like the original ones better.

...Robert