Re: [boost] How about a call for bug-fix volunteers?

Subject: Re: [boost] How about a call for bug-fix volunteers?
From: Jim Bell (Jim_at_[hidden])
Date: 2010-10-28 12:48:21

On 1:59 PM, ZHUO Qiang wrote:
> 于 2010年10月26日 23:00, Jim Bell 写道:
>> Library authors shouldn't have to bear the burden of bug reports for
>> their libraries alone. Even maintaining their regression tests across
>> all platforms isn't realistic (and isn't being done).
>>
>>
>
> Sounds a good idea.
>
> I once fired a ticket 5 month ago about Boost.PropertyTree with a fix
> patch and unit test against the bug
> (https://svn.boost.org/trac/boost/ticket/4340), however, it seems no
> one takes care about it for 5 months. Frustrating me a little bit.

I think yours is a good case study for "Ticket Triage." You seem to have
done the grunt work to give some very low-hanging fruit for the boost
authors to incorporate. But you're not familiar with the ticket system,
so you don't mark it 'regression' or don't set the milestone, and it
falls through the cracks. (At least I think 'regression' the right
setting...)

Now I don't know property_tree at all, so getting my mind around your
ticket would take more time than I can spend on any random ticket, even
a promising-looking one like this. (And this would be the release
manager's shoes.)

But a volunteer who:
* understands property_tree (user-level knowledge, not developer-level),
* knew his way around its regression tests (crucial!),
* understands Trac & how we're using it...

This volunteer could:
* interact with the submitter with something he hasn't considered,
* mark it 'regression' (or 'feature request') and/or set the milestone
* give some other heads-up to the library's author or the release
manager that this is a good one to commit back to the trunk.
* point the submitter to the closest regression and challenge him to
adapt it to show the bug

The submitter is heard, the bug gets fixed, boost gets better, and the
volunteer sharpens his skills.