On Nov 15, 2010, at 2:09 PM, vicente.botet wrote:

----- Original Message ----- From: "Jim Bell" <Jim@JC-Bell.com> To: <boost@lists.boost.org> Sent: Monday, November 15, 2010 3:48 PM Subject: [boost] [Guild] Recruiting & Instructions, rough first draft

...

Here's some writing I threw down a little while ago.

Please comment, here or privately. Copy my text for reference.

It's on my website partly because I've done some formatting (emphasizing), as you'll see.

It's rough, and lacks some crucial detail.

http://jc-bell.com/contributions/boost-guild/boost-guild-recruit http://jc-bell.com/contributions/boost-guild/boost-ticket-handling http://jc-bell.com/contributions/boost-guild/boost-regression-troubleshootin...

Thanks Jim for writing this. This seems to be a good starting and I hope the Guild will have a lot of member soon.

I wanted just to add an idea that could improve the library quality. People that don't have the understanding of the internals of the library and are not interested in can as a user know how the library should behave. Even if Boost Libraries are well testeds there are always holes. This people could inspect the regression tests and the documentation and propose news test to cover more features in the library. That would result in more confidence on the quality of the library or why not, find a new bug that other will need to correct ;-). They could also ensure that any corrected tickets have its associated test.

I agree. Having more regression tests is a good thing. Writing good regression tests is surprisingly hard. -- Marshall

----- Original Message ----- From: "Jim Bell" <Jim@JC-Bell.com> To: <boost@lists.boost.org> Sent: Tuesday, November 16, 2010 12:08 AM Subject: Re: [boost] [Guild] Recruiting & Instructions, rough first draft

On 1:59 PM, Marshall Clow wrote:

...

On Nov 15, 2010, at 2:09 PM, vicente.botet wrote:

...

----- Original Message ----- From: "Jim Bell" <Jim@JC-Bell.com> To: <boost@lists.boost.org> Sent: Monday, November 15, 2010 3:48 PM Subject: [boost] [Guild] Recruiting & Instructions, rough first draft

I wanted just to add an idea that could improve the library quality. People that don't have the understanding of the internals of the library and are not interested in can as a user know how the library should behave. Even if Boost Libraries are well testeds there are always holes. This people could inspect the regression tests and the documentation and propose news test to cover more features in the library. That would result in more confidence on the quality of the library or why not, find a new bug that other will need to correct ;-). They could also ensure that any corrected tickets have its associated test. I agree. Having more regression tests is a good thing. Writing good regression tests is surprisingly hard.

I agree with you both. Making a meaningful, orthogonal regression test isn't easy. As someone learns their way around the tests, docs, and tickets, even more tests might occur to them.

Well this is not bad. Jim, would you try to integrate this idea in Guild?

Just curious: are the regression tests part of a new library's review/acceptance? (Either way, I know the need for new ones spring up.)

Reviews should take care of test also. A library without a good test coverage should not even be ready for acceptance. Vicente

At Mon, 15 Nov 2010 23:34:18 +0100, vicente.botet wrote:

The last two links are quite similar. I would say that one of the tasks will be to add a ticket, when not already present, for each failing test in the regression matrix.

* Add a ticket for each failing test * Add a test for each ticket two-sides-of-the-same-coin-ly y'rs, -- Dave Abrahams BoostPro Computing http://www.boostpro.com