[PATCH 1 of 3] changegroup: add changegroup4

[Martin von Zweigbergk]

On Tue, Jan 12, 2016 at 4:04 PM Gregory Szorc <gregory.szorc at gmail.com>
wrote:

> On Tue, Jan 12, 2016 at 11:40 AM, Durham Goode <durham at fb.com> wrote:
>
>> On 1/12/16, 10:49 AM, "Augie Fackler" <raf at durin42.com> wrote:
>>
>>
>>
>> >(+durham, who may have concerns around breaking format changes in cg3,
>> >but I'd rather fix cg3 pre-freeze than do cg4)
>> >
>> >On Tue, Jan 12, 2016 at 11:48 AM, Martin von Zweigbergk
>> ><martinvonz at google.com> wrote:
>> >> On Tue, Jan 12, 2016 at 8:40 AM, Gregory Szorc <
>> gregory.szorc at gmail.com> wrote:
>> >>>
>> >>>
>> >>>> On Jan 11, 2016, at 15:35, Martin von Zweigbergk <
>> martinvonz at google.com> wrote:
>> >>>>
>> >>>> # HG changeset patch
>> >>>> # User Martin von Zweigbergk <martinvonz at google.com>
>> >>>> # Date 1452553831 28800
>> >>>> #      Mon Jan 11 15:10:31 2016 -0800
>> >>>> # Node ID ee464a97569061f4c8bb4af316a0517b4feb0ad7
>> >>>> # Parent  4571c0b383378f5eec5189e3a39c672a169ce566
>> >>>> changegroup: add changegroup4
>> >>>>
>> >>>
>> >>> We haven't shipped cg3, right? If there is no backwards compat
>> concern, can we not modify cg3 instead of relegating it to an unused format
>> (superseded by cg4) that we'll need to support forever?
>> >>
>> >> You're right that there seems to be no release since cg3 was added in
>> >> 77d25b913f80 (changegroup: introduce cg3, which has support for
>> >> exchanging treemanifests, 2015-12-11) and 12f727a5b434 (changegroup:
>> >> add flags field to cg3 delta header, 2015-12-14). And I suppose we
>> >> don't care about anyone who might have picked up an unreleased version
>> >> of hg, so it seems safe to me. Thanks! Sorry, I should have checked
>> >> that myself.
>>
>> We haven't actually shipped cg3 capable code to our servers yet, so we
>> have no dependency here yet.  We may ship cg3 capable code to the server
>> before this series lands though.  Is there a config or a hotfix I can make
>> to disable cg3 for us for now?  Will just removing it from the packermap
>> prevent it from being advertised?
>>
>
> From my brief reading of exchange.py, I believe removing it from
> changegroup.packermap will be sufficient to disable it from the wire
> protocol.
>
>
Yes, I think you're right. I mentioned this to Durham on IRC, but I should
have updated this thread too. I have a second version of this series out
for review now. That version just updates cg3. Hopefully that can get in
before Durham needs to pick up a new version of hg.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20160113/c9a760e4/attachment.html>