Re: [PATCHES] Custom variable class segmentation fault

Bruce Momjian wrote:
>
>
>> My post below was merely to agree with Tom that in principle, patches
>> should be be reviewed before application and not after. I still think
>> that's right - I have been concerned lately that the buildfarm has been
>> broken a bit too much.
>>
>
> Well, just because they are reviewed doesn't mean they aren't going to
> break the build farm. In fact, the build farm is there to be broken ---
> if all patches worked fine on all machines, we wouldn't need the build
> farm. Let's not get into a case where keeping the build farm green is
> our primary goal, "Oh, let's not apply that patch or it might break the
> build farm". Hey, I have an idea, let's stop CVS update on the build
> farm, and it will stay green forever. :-) LOL (Of course, we don't
> want the build farm to stay broken or it masks newly introduced errors.)
>

I certainly expect buildfarm to break. But it is not intended as a
substitute for review either. We shouldn't be in the business of saying
"let's apply it and see if buildfarm breaks". We should be saying "I
have looked at this and my best guess is that it won't break." That
won't avoid all breakage, certainly. But it will keep it down.

> If you withdraw your object to the GUC patch, then with a single person
> objecting, the patch either goes in or that person takes responsibility
> for getting it into 8.2, or the blame for leaving it for 8.3.
>
>

As I pointed out - I did not object to the patch being applied, I just
stated an agreement with a general principle, so there is nothing to
withdraw.

cheers

andrew