[PATCH] strip: invalidate phase cache after stripping changeset (issue5235)

[Pierre-Yves David]

On 05/11/2016 03:38 PM, Laurent Charignon wrote:
>
> On 5/11/16, 2:21 PM, "Pierre-Yves David" <pierre-yves.david at ens-lyon.org>
> wrote:
>
>>
>> On 05/11/2016 03:16 PM, Laurent Charignon wrote:
>>> On 5/11/16, 2:10 PM, "Pierre-Yves David"
>>> <pierre-yves.david at ens-lyon.org>
>>> wrote:
>>>
>>>> On 05/11/2016 03:08 PM, Laurent Charignon wrote:
>>>>> # HG changeset patch
>>>>> # User Laurent Charignon <lcharignon at fb.com>
>>>>> # Date 1462971969 25200
>>>>> #      Wed May 11 06:06:09 2016 -0700
>>>>> # Node ID fd005346d3c86b65c690f8f987e05f44ec4ec02f
>>>>> # Parent  df838803c1d487e4601f96c6cfd85e6ad4f6291f
>>>>> strip: invalidate phase cache after stripping changeset (issue5235)
>>>> Looks like a small fix, should this be on stable?
>>> I am not sure.
>>> I tried to dig into the implications for histedit and I don't understand
>>> fully what is happening.
>>> If you run the histedit test with my fix (but not the modification to
>>> the
>>> histedit tests), then all the revisions in one of the test become
>>> magically public.
>>> This seems risky to push it to stable, isn't it?
>>  From reading this, it seems like there is an interraction betwen phase
>> root and histedit that you do not understand fully and that might still
>> be massively broken. Did I get that right?
> Yes, and it seems to have to do with descendants of the null revision.

Okay, so we need to really figure out what is going on here before 
taking that patch. Hopefully taking it on stable. I'm dropping it from 
patch work. Happy investigation!

Cheers,

-- 
Pierre-Yves David