Oh wow, what an awful hack (I should know, I wrote it).

I'm a bit confused about who is copying what in there, but I trust your judgment...

In D8775#228206, @vlorentz wrote:

Dulwich calls it "raw_string", so I went with that

This is now done, the objects are fixed in the production DB and kafka.

@vlorentz I'm running the following adaptation to your script:

Yeah, I think that makes sense. Time for these tests to be written? :)

must... resist... painting the bikeshed (git_manifest? git_object? raw_object?)

Great, thanks!

Rebase on updated D8769

Variabilize task_name

Add tests for the new absolute_cooldown

This seems like a great idea.

In D8759#227735, @vlorentz wrote:

Anyway, I'm guessing it will fix itself when you send any update to the diff so let's ignore that issue for now.

This looks fine to me, though we might want to use something other than SHA1 for this.

We are currently in the process of moving the objstorage to multi-hashes, so eventually we will want to use multi-hashes here too. I wonder if we should do that before landing this diff (and D8760). It would be less effort overall, but will delay your work a bit.

eg. here it could be done by making payload a dictionary (with algos as keys and digests as values), and serialize the manifest as:

[payload_type $StrWithoutSpaces]
[payload blake2s256 $Hexdigest]
[payload sha1 $Hexdigest]
[payload sha1_git $Hexdigest]
[payload sha256 $Hexdigest]

@samplet @olasd What do you think?

In D8759#227732, @vlorentz wrote:

Hi, thanks for the diffs

In D8759#227712, @samplet wrote:

It looks like using the --draft flag when sending this with arc has confused the CI system. The build failed because it couldn't access the updated code:

fatal: Couldn't find remote ref refs/tags/phabricator/diff/31582

It was because you hadn't signed the CLA when submitting the diff

I've relaunched the latest version of the migrate_extrinsic_metadata script on getty...

Woohoo, green tests!

In D8705#227333, @ardumont wrote:

...ah... maybe, it's for other tests in another repository, right?

@vlorentz mentions that this is missing a docstring change

Yeah, I'm a bit surprised too that this would decrease overall test times, and seems like a latent bug, but this at least doesn't seem to /increase/ test times in this module, so it's probably fine?

In D8705#227328, @ardumont wrote:

If it's for test speedup only, don't you want to keep the actual behavior instead?
That is, make the auto_flush be False by default, and then you explicitely set it up to True in the calling tests.

This task was migrated to Gitlab

Migrated to GitLab

Thanks!

This project has migrated to GitLab