RFR: CODETOOLS-7901672: trees extension broken after upgrade to mercurial 3.8.1

[Jonathan Gibbons]

Magnus,

Will push this later today.

-- Jon


On 5/31/17 1:36 AM, Magnus Ihse Bursie wrote:
> On 2017-05-31 10:20, Erik Joelsson wrote:
>> Hello,
>>
>> Thanks for looking at this!
>>
>> I have renamed run_command to command_cmd. I agree the naming should 
>> be consistent.
>>
>> http://cr.openjdk.java.net/~erikj/CODETOOLS-7901672/webrev.02/
>
> Looks good to me.
>
> Jon, can you push Erik's patch?
>
> /Magnus
>
>>
>> /Erik
>>
>>
>> On 2017-05-17 12:30, Magnus Ihse Bursie wrote:
>>> Looks good to me. Thanks for fixing this!
>>>
>>> A minor nit:
>>>
>>> +# This function cannot be named "command" since it clashes with the 
>>> @command
>>> +# decorator.
>>> + at command('tcommand|tcmd')
>>> +def run_command(ui, repo, cmd, *args, **opts):
>>>
>>> +# This function cannot be named list since it clashes with the 
>>> python builtin
>>> + at command('tlist')
>>> +def list_cmd(ui, repo, **opts):
>>>
>>> Maybe use a more consistent naming scheme? E.g. rename the first 
>>> command_cmd? But I'm ok if you want to keep the run_command name as 
>>> well.
>>>
>>> /Magnus
>>>
>>>
>>> On 2017-04-18 19:35, Erik Joelsson wrote:
>>>> The way a Mercurial plugin interacts with Mercurial has changed 
>>>> over time. The technique used in trees has been deprecated for a 
>>>> while and in 3.8, it no longer works.
>>>>
>>>> I have a patch that adopts the new method using decorators, 
>>>> introduced in Mercurial 1.9. It also contains a fallback workaround 
>>>> that keeps compatibility with older versions.
>>>>
>>>> For trees, there are further complications for versions between 1.9 
>>>> and 3.1 which the patch also handles. I have run the automatic 
>>>> tests for all versions from 1.1 to latest as well as manually using 
>>>> it with various versions.
>>>>
>>>> Bug: https://bugs.openjdk.java.net/browse/CODETOOLS-7901672
>>>>
>>>> Webrev: http://cr.openjdk.java.net/~erikj/CODETOOLS-7901672/webrev.01/
>>>>
>>>> /Erik
>>>>
>>>>
>>>
>>
>