tech-repository archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Proposed conversion strategy



Reinoud Zandijk <reinoud%NetBSD.org@localhost> writes:

> See f.e. https://github.com/ucb-bar/riscv-tools
>
>> If so I'd like to understand what it is you are calling a "sub-branch",
>> as I'm not sure what that is.
>
> There are various links there to various points in other repositories; those
> are the types i talked about. These points could be considered vendor branch
> imports but they are set to specific commit points to ensure that despite
> further enhancements on each of the brances this checkout contains a working
> set.

Assuming git, we could certainly decide not to use submodules.  The fact
that submodules have issues is not a reason not to use git, just to
avoid submodules.

It's relatively straighforward to do do vendor imports to a branch
disconnected from master and then to do a subtree merge to place it at
the right point.  For example, in a private repo I have 'vendor/netbsd'
with a snapshot of netbsd, and then have subtree merged it to
master:netbsd, so that it shows up under the netbsd directory in the
top-level checkout.   Of course, this would be something like
vendor/postfix and then mapped into
src/external/ibm-public/postfix/dist/.

It's unclear to me how the conversion tools deal with these vendor
imports to points below the root.  It seems that the best thing would be
to map them to vendor/foo branches with subtree merges.

Attachment: pgppWnh3nMggN.pgp
Description: PGP signature



Home | Main Index | Thread Index | Old Index