Call for discussion: Phase names

[Pierre-Yves David]

On 11 janv. 2012, at 05:29, Kevin Bullock wrote:

> On 10 Jan 2012, at 12:30 PM, Pierre-Yves David wrote:
> 
>> On Tue, Jan 10, 2012 at 11:40:01AM -0600, Kevin Bullock wrote:
>>> On Jan 10, 2012, at 11:19 AM, Pierre-Yves David wrote:
>>> 
>>>> On Tue, Jan 10, 2012 at 10:55:06AM -0600, Kevin Bullock wrote:
>>>>> On Jan 9, 2012, at 7:01 PM, Pierre-Yves David wrote:
>>>>> 
>>>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>>>> Hash: SHA1
>>>>>> 
>>>>>> 
>>>>>> On 9 janv. 2012, at 23:25, Matt Mackall wrote:
>>>>>> 
>>>>>> People today do things like 'hg id -r tip remote' to figure out what
>>>>>>> their incoming changeset group is going to look like, and if we don't
>>>>>>> actively hide these changesets from remote clients, we're breaking that.
>>>>>> 
>>>>>> Those people should use incoming…
>>>>> 
>>>>> ?!? incoming fetches a full bundle. There are plenty of cases where you want to just see if a remote has a given cset (or what cset it thinks a given label points to), without the overhead of actually fetching changes.
>>>> 
>>>> Remote lookup is currently not being rewritten according secret. This could be
>>>> seen as a bug but I'm not enthousiasmed by the fact that the following commands
>>>> return different thing
>>>> 
>>>>   hg id -r <some-lookup> repo
>>>>   hg -R repo id -r <some-lookup>
>>>> 
>>>> (with some-lookup any of tip, <branch-name>, <bookmarks>, …)
>>> 
>>> I don't see any reason why those would have to be different if we leave
>>> secret changes out of discovery. Either command should end up with an
>>> instance of localrepo if the repo is local, and bypass the discovery code
>>> anyway, shouldn't it? (I ask naïvely, having never actually looked at that
>>> part of the code.)
>> 
>> With you behavior you describe above, you have an inconsistency between the
>> what `hg -r tip <remote>` says and `hg pull <remote>` fetch. This inconcistency
>> was the initial point of this discussion…
> 
> Then 'id' must be made to agree with 'pull', even for local repos that don't invoke the wire protocol. You can't tell users, "Just use 'incoming' instead of 'id'". The speed and bandwidth difference could be large to enormous.
> 
> If I pull from a repo on the same disk, today, do I get secret changesets? (I would expect not, but it's another issue to nail down.)

You don't

-- 
Pierre-Yves