Re: [boost] What are the things that we HAVE TO fix before shipping 1.56.0?

On 7/29/2014 3:39 PM, Eric Niebler wrote:

On 07/29/2014 01:36 PM, Marshall Clow wrote:

...

On Jul 29, 2014, at 8:31 AM, Eric Niebler wrote:

...

On 7/29/2014 8:18 AM, Eric Niebler wrote:

...

On 7/29/2014 12:32 AM, Andrey Semashev wrote:

...

On Tue, Jul 29, 2014 at 1:24 AM, Marshall Clow wrote:

...

We’re coming down to the end of the release process, so I’m taking a poll.

* What bugs do you think MUST be fixed before shipping 1.56? In other words, if bug XXX is not fixed, then (in your opinion) we should not bother making a release.

I think the issue in the nearby thread [1] needs to be fixed before the release. The fix is rather simple anyway.

[1] http://lists.boost.org/Archives/boost/2014/07/215581.php

Did you/someone file a bug? It's unlikely to get fixed otherwise.

Sorry, I see the pull request now.[*] I agree, it'd be nice to have this fix, but I'm not anxious to hold up the release. Marshall?

[*] https://github.com/boostorg/asio/pull/7

I’m really leery of putting something in the release that hasn’t gone through a test cycle.

Naturally. But see [1] for another case where this is a problem. It looks bad. We have two choices:

1) Hold up the release while we work on a fix and run some emergency tests. 2) Release as-is, then quickly roll a point release.

The fact that two people have reported the issue separately gives me pause.

Eric

[1]: https://groups.google.com/d/msg/boost-developers-archive/6JVNg7ZPb4k/RAlvPUe...

More information, cc'ing Neil. The problem is this commit is Boost.Range: https://github.com/boostorg/range/commit/1d91272a551df2cd1776d6447c0e39a14f6... The change to the free relational operators is causing havoc. The fix should be as simple as reverting to the 1.55 formulation of them. I'll send a patch shortly. Neil, please chime in ASAP if you think that's the wrong fix. Eric