Re: [HACKERS] ON_ERROR_ROLLBACK

On Tue, Dec 30, 2014 at 8:54 AM, Adrian Klaver <adrian(dot)klaver(at)aklaver(dot)com>
wrote:

> On 12/30/2014 07:43 AM, David G Johnston wrote:
>
>> Tom Lane-2 wrote
>>
>>> Bernd Helmle &lt;
>>>
>>
>> mailings@
>>>
>>
>> &gt; writes:
>>>
>>>> --On 29. Dezember 2014 12:55:11 -0500 Tom Lane &lt;
>>>>
>>>
>> tgl(at)(dot)pa
>>>
>>
>> &gt; wrote:
>>>
>>>> Given the lack of previous complaints, this probably isn't backpatching
>>>>> material, but it sure seems like a bit of attention to consistency
>>>>> would be warranted here.
>>>>>
>>>>
>>> Now that i read it i remember a client complaining about this some time
>>>> ago. I forgot about it, but i think there's value in it to backpatch.
>>>>
>>>
>>> Hm. Last night I wrote the attached draft patch, which I was intending
>>> to apply to HEAD only. The argument against back-patching is basically
>>> that this might change the interpretation of scripts that had been
>>> accepted silently before. For example
>>> \set ECHO_HIDDEN NoExec
>>> will now select "noexec" mode whereas before you silently got "on" mode.
>>> In one light this is certainly a bug fix, but in another it's just
>>> definitional instability.
>>>
>>> If we'd gotten a field bug report we might well have chosen to
>>> back-patch,
>>> though, and perhaps your client's complaint counts as that.
>>>
>>> Opinions anyone?
>>>
>>
>> -0.5 for back patching
>>
>> The one thing supporting this is that we'd potentially be fixing scripts
>> that are broken but don't know it yet. But the downside of changing
>> active
>> settings for working scripts - even if they are only accidentally working
>> -
>> is enough to counter that for me. Being more liberal in our acceptance of
>> input is more feature than bug fix even if we document that we accept more
>> items.
>>
>
> It is more about being consistent then liberal. Personally I think a
> situation where for one variable 0 = off but for another 0 = on, is a bug
>
>
​I can sorta buy the consistency angle but what will seal it for me is
script portability - the ability to write a script and instructions using
the most current release and have it run on previous versions without
having to worry about this kind of incompatibility.

So, +1 for back patching from me.

David J.​