Peter Eisentraut <peter.eisentraut@enterprisedb.com> writes:
> On 21.07.21 21:12, Daniel Gustafsson wrote:
>>> Additionally the commit ID is stored in a global extended
>>> pax header if the tar format is used; it can be extracted using git
>>> get-tar-commit-id. In ZIP files it is stored as
>>> a file comment.
>> That does adds Git as a dependency for consuming the tarball though, which
>> might not be a problem but it's a change from what we require today.
> How so?
It's only a dependency if you want to know the commit ID, which
perhaps isn't something you would have use for if you don't have
git installed ... but I don't think that's totally obvious.
Personally I'd be more worried about rendering the tarballs
totally corrupt from the perspective of somebody using an old
"tar" that hasn't heard of extended pax headers. Maybe there
are no such versions in the wild anymore; but I do not see any
advantages to this approach that would justify taking any risk for.
regards, tom lane