Re: [PATCH] test-lib: GIT_TEST_SANITIZE_LEAK_LOG enabled by default
- From
Patrick Steinhardt <ps@pks.im>
- Date
- Jul 22, 2024, 07:52 UTC
- Message-ID
- <Zp4Pq-adl2mwA8ql@tanuki>
- In-Reply-To
- <20240710011206.GC1526498@coredump.intra.peff.net>
On Tue, Jul 09, 2024 at 09:12:06PM -0400, Jeff King wrote:
Show 17 quoted lines
> On Wed, Jul 10, 2024 at 02:51:58AM +0200, Rubén Justo wrote: > > > As we describe in t/README, it can happen that: > > > > Some tests run "git" (or "test-tool" etc.) without properly checking > > the exit code, or git will invoke itself and fail to ferry the > > abort() exit code to the original caller. > > > > Therefore, GIT_TEST_SANITIZE_LEAK_LOG must be set to true to capture all > > memory leaks triggered by the tests when SANITIZE=leak. > > > > Set it to true by default, and stop worrying about someone checking for > > leaks who isn't aware of this option and might be missing some leaks. > > I'm obviously in favor of this direction, but...why stop here? Do we > expect somebody to set it to false? If not, then can't we just get rid > of it entirely?
I'd also be strongly in favor of just removing this variable altogether. I found it quite tedious to remember setting it when working on the memory leak fixes recently, and I'm not aware of any downsides.
Patrick