RFR: Save references to materialized branches locally

[Erik Helin]

On Wed, 27 Nov 2019 09:41:46 GMT, Robin Westberg <rwestberg at openjdk.org> wrote:

> On Wed, 27 Nov 2019 09:37:28 GMT, Erik Helin <ehelin at openjdk.org> wrote:
> 
>> On Tue, 26 Nov 2019 13:00:47 GMT, Robin Westberg <rwestberg at openjdk.org> wrote:
>> 
>>> Hi all,
>>> 
>>> Please review this change that ensure that when bots use `Repository.materialize`, they always save a named reference to what was fetched. This ensures that subsequent fetches will present these existing refs, and avoid re-fetching commits that already exist locally.
>>> 
>>> It also fixes a potential bug where the mlbridge archive reader could have fetched from an incorrect branch.
>>> 
>>> Best regards,
>>> Robin
>>> 
>>> ----------------
>>> 
>>> Commits:
>>>  - 280c73b8: Use configurable ref for the mlbridge archive
>>>  - 6b707ae4: Specify a destination ref when using materialize
>>> 
>>> Changes: https://git.openjdk.java.net/skara/pull/271/files
>>>  Webrev: https://webrevs.openjdk.java.net/skara/271/webrev.00
>>>   Stats: 54 lines in 14 files changed: 28 ins; 0 del; 26 mod
>>>   Patch: https://git.openjdk.java.net/skara/pull/271.diff
>>>   Fetch: git fetch https://git.openjdk.java.net/skara pull/271/head:pull/271
>> 
>> bots/hgbridge/src/main/java/org/openjdk/skara/bots/hgbridge/ExporterConfig.java line 172:
>> 
>>> 171:         var localRepo = Repository.materialize(scratchPath, configurationRepo.url(),
>>> 172:                                                configurationRef + ":hgbridge_config_" + configurationRepo.name());
>>> 173: 
>> 
>> Why no `+` here to to imply `--force` to `git fech`?
> 
> No reason, good catch!

Should we instead change `Repository.materialize` to always prepend the `+` to the refspec? Or do you think that makes it less clear at the call sites?

PR: https://git.openjdk.java.net/skara/pull/271