Re: [PATCH 00/17] leak fixes: use existing constructors & other trivia

From: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date: Nov 5, 2022, 12:46 UTC
Message-ID: <221105.865yfty38w.gmgdl@evledraar.gmail.com>
In-Reply-To: <8eec228d-d392-523d-2415-149b946f642e@dunelm.org.uk>

On Fri, Nov 04 2022, Phillip Wood wrote:

Show 9 quoted lines

> On 03/11/2022 17:05, Ævar Arnfjörð Bjarmason wrote:
>> With the very minor exceptions of:
>> * 03-04/17 (which need trivial oilerplate)
>> * 05/17 (need to add trivial control flow to a free_*() function)
>> * 12/17 (narrowing scope of allocation)
>
> I've only looked at the rebase related patches. I'd really appreciate
> it if you could drop patches 12 & 14 as they conflict with [1] that
> fixes these issues by removing the setenv() calls.

Sure, sounds good to me.

I see Taylor queued it up like that, and the combination of the two passes with the "true" and "check" mode leak checking modes.

For context: I saw that you planned to fix those in some follow-up topic, but had this mostly ready before the new leak in the recent topic showed up, and that leak had big knock-on effects in what test files I could mark as passing.

So I figured I'd include some more more narrow fixes, but I'm even happier to see the root cause addressed.