Retirement of Reviewboard - Transition to Phabricator

[Ben Cooksley]

Hi all,

The following is Sysadmin's suggested plan for the retirement of
Reviewboard now that Phabricator is fully up and running for hosting
of code reviews.

Phase 1: Commences September 2: All repositories are closed for
accepting new reviews on Reviewboard. A notice is added to the top of
the main page indicating that reviews should now be done on
Phabricator.

Phase 2: Commences September 16: Login to Reviewboard is disabled, and
final backups are taken. A static copy of Reviewboard is generated and
published online, and the software itself is taken down.

The vast majority of projects should now be migrated to Phabricator,
with only historical reviews needing to be cleaned up.

Note that due to how Reviewboard stores diffs and reproduces them for
use, some reviews may have decayed and may no longer be readable. This
is due to short-hashes which are used by Git/Reviewboard in diffs now
having collisions with other commits which previously did not exist.
Unfortunately there is nothing we can do about this.

Any comments on the above?

Regards,
Ben

[Albert Astals Cid]

El dijous, 24 d’agost de 2017, a les 21:07:49 CEST, Ben Cooksley va escriure:

> Hi all,
>
> The following is Sysadmin's suggested plan for the retirement of
> Reviewboard now that Phabricator is fully up and running for hosting
> of code reviews.
>
> Phase 1: Commences September 2: All repositories are closed for
> accepting new reviews on Reviewboard. A notice is added to the top of
> the main page indicating that reviews should now be done on
> Phabricator.
>
> Phase 2: Commences September 16: Login to Reviewboard is disabled, and
> final backups are taken. A static copy of Reviewboard is generated and
> published online, and the software itself is taken down.

Does this mean i can still see:
 * Diffs
 * emails of the people that made those diffs

After phase 2?

Cheers,
  Albert

>
> The vast majority of projects should now be migrated to Phabricator,
> with only historical reviews needing to be cleaned up.
>
> Note that due to how Reviewboard stores diffs and reproduces them for
> use, some reviews may have decayed and may no longer be readable. This
> is due to short-hashes which are used by Git/Reviewboard in diffs now
> having collisions with other commits which previously did not exist.
> Unfortunately there is nothing we can do about this.
>
> Any comments on the above?
>
> Regards,
> Ben

[Ben Cooksley]

On 25/08/2017 5:41 AM, "Albert Astals Cid" <[hidden email]> wrote:

El dijous, 24 d’agost de 2017, a les 21:07:49 CEST, Ben Cooksley va escriure:

> Hi all,
>
> The following is Sysadmin's suggested plan for the retirement of
> Reviewboard now that Phabricator is fully up and running for hosting
> of code reviews.
>
> Phase 1: Commences September 2: All repositories are closed for
> accepting new reviews on Reviewboard. A notice is added to the top of
> the main page indicating that reviews should now be done on
> Phabricator.
>
> Phase 2: Commences September 16: Login to Reviewboard is disabled, and
> final backups are taken. A static copy of Reviewboard is generated and
> published online, and the software itself is taken down.

Does this mean i can still see:
 * Diffs
 * emails of the people that made those diffs

Diffs should definitely be - as long as they're not broken due to the issue I noted. 

As you wouldn't be able to access them currently there is no change in that regard.

In terms of email addresses this will only be the case if Reviewboard displays them publicly. Their username will still be shown though which we can lookup on Identity.

Regards,

Ben

After phase 2?

Cheers,
  Albert

>
> The vast majority of projects should now be migrated to Phabricator,
> with only historical reviews needing to be cleaned up.
>
> Note that due to how Reviewboard stores diffs and reproduces them for
> use, some reviews may have decayed and may no longer be readable. This
> is due to short-hashes which are used by Git/Reviewboard in diffs now
> having collisions with other commits which previously did not exist.
> Unfortunately there is nothing we can do about this.
>
> Any comments on the above?
>
> Regards,
> Ben

[René J.V. Bertin]

On Friday August 25 2017 07:33:43 Ben Cooksley wrote:

Hi,

>> Note that due to how Reviewboard stores diffs and reproduces them for
>> use, some reviews may have decayed and may no longer be readable. This
>> is due to short-hashes which are used by Git/Reviewboard in diffs now
>> having collisions with other commits which previously did not exist.
>> Unfortunately there is nothing we can do about this.

This is probably a "wild thought", but would it be possible somehow to limit the issue by not letting the ReviewBoard software compare to each repo's HEAD but against whatever commit is current now (or when you retire the thing)?

And FWIW, you're aware of git-diff's --full-index option? ;)

R.

[Ben Cooksley]

On Mon, Aug 28, 2017 at 8:20 PM, René J.V. Bertin <[hidden email]> wrote:
> On Friday August 25 2017 07:33:43 Ben Cooksley wrote:
>
> Hi,

Hi Rene,

>
>>> Note that due to how Reviewboard stores diffs and reproduces them for
>>> use, some reviews may have decayed and may no longer be readable. This
>>> is due to short-hashes which are used by Git/Reviewboard in diffs now
>>> having collisions with other commits which previously did not exist.
>>> Unfortunately there is nothing we can do about this.
>
> This is probably a "wild thought", but would it be possible somehow to limit the issue by not letting the ReviewBoard software compare to each repo's HEAD but against whatever commit is current now (or when you retire the thing)?

Reviewboard to my knowledge compares against the commit sha's
mentioned in the diff.

>
> And FWIW, you're aware of git-diff's --full-index option? ;)

This is something the contributor could have done when uploading yes,
but isn't something we can do anything about now.

>
> R.

Regards,
Ben