Re: [Boost-testing] [QNX6][date-time] changes to conversions.hpp (was Can't produce final ...)

From: Jeff Garland (jeff_at_[hidden])
Date: 2005-10-28 08:53:43

On Thu, 27 Oct 2005 13:33:50 -0500, Rene Rivera wrote
> Jim Douglas wrote:
>
> > I have only been working on this QNX conversion for a couple of weeks 
> > and I have no idea about your timescales for 1.33.1. Rene Rivera is 
> > acting as my mentor/sponsor and might have a better idea about priorities.
> 
> Well the priority has been to get 1.33.1 out ASAP :-) The plan was 
> to have a candicate a month ago, so that tells you how well 
> priorities and plans work out.
> 
> > IMHO ASAP:-) Please can you make the changes to RC_1_33_0 anyway. It 
> > seems like a pretty low risk course of action.
> 
> Given that the change involves making some symbol references follow 
> what other instances of that same symbol where elsewhere I agree 
> it's a low risk change. But the decision is still up to Jeff, and Doug.

Since I didn't hear back from Doug I made an executive decision and checked
them in on the branch.  The changes didn't blow up the mainline yesterday, and
they would be trivial to revert if they cause a problem on the branch -- so
they seemed safe enough.  Jim, I assume the reason the test report showed all
green is because you had local patches?

> As for the question of priority of QNX support in 1.33.1... It would 
> be nice, but my priority would be to have enough support that it 
> makes additional testing possible, and I think we are already there. 
> I'm comparing this to the situation before where one could not even 
> get past step 0, of trying to build.

I think what I was trying to say is that really Jim should be doing this work
against the mainline and not 1.33.1 because we are really trying to do only
important bugfixes in 1.33.1.  It's not to say QNX is not important, just not
as important as actually shipping the critical bug fixes.  So if the changes
were more extensive I would have wanted them on the mainline only.  That said,
I can understand that from a porting strategy point of view you want to test
against a stable baseline -- and the RC branch is indeed the best for that. 
So it's sort of a difficult bind.  

Jeff