Re: ds/commit-graph-gen-v2-fixes (was Re: What's cooking in git.git (Feb 2022, #08; Mon, 28))
- From
Patrick Steinhardt <ps@pks.im>
- Date
- Mar 1, 2022, 14:58 UTC
- Message-ID
- <Yh40pJ1o0JHG5Op2@ncase>
- In-Reply-To
- <a296c8fc-9701-15bd-4459-e49f157adb56@github.com>
On Tue, Mar 01, 2022 at 09:26:51AM -0500, Derrick Stolee wrote:
Show 29 quoted lines
> On 3/1/2022 8:42 AM, Junio C Hamano wrote: > > > * ds/commit-graph-gen-v2-fixes (2022-02-28) 4 commits > > - commit-graph: fix generation number v2 overflow values > > - commit-graph: start parsing generation v2 (again) > > - commit-graph: fix ordering bug in generation numbers > > - test-read-graph: include extra post-parse info > > > > Fixes to the way generation number v2 in the commit-graph files are > > (not) handled. > > > > Will merge to 'next'. > > source: <pull.1163.v2.git.1646056423.gitgitgadget@gmail.com> > > Please hold for a re-roll on this one. Ævar has some feedback > on these patches that merit a new version. His comments are > mainly about test structure, but they will help get extra > coverage in the 32-bit case. > > Patrick also had some concerns about upgrading from older, > buggy versions of Git. I think this isn't an issue [1] because of > how those older versions would ignore existing corrected > commit dates, but it would be good to get a common understanding > of the scenario before merging this. > > [1] https://lore.kernel.org/git/f50e74f0-9ffa-f4f2-4663-269801495ed3@github.com/ > > Thanks, > -Stolee
Indeed, I think my concern is not an issue. The failures I was seeing were my own fault because I had been running an intermediate version of Git, which created those faulty graphs, and they carried over into later testing.
Patrick