Re: [boost] 1.33 release regression results -- 'extra compilers'

From: Jeff Garland (jeff_at_[hidden])
Date: 2005-08-14 09:38:20

On Sat, 13 Aug 2005 23:13:28 -0500, Aleksey Gurtovoy wrote
> Jeff Garland writes:
> > On Fri, 12 Aug 2005 22:07:38 -0500, Aleksey Gurtovoy wrote
> >> Jeff Garland writes:
> >> > Why are digital mars, como and other 'non-required' compilers in the
> >> > regressions results page ( http://tinyurl.com/cdlxk )?  
> >> 
> >> Because the release was tested on them and many people would like to
> >> see what's the status of these?
> >
> > Shouldn't these be on the 'developer link' ?
> 
> I don't see why. These people are not Boost developers.

Well I guess that points out a problem with the 'titles' (developer and user)
we use to divide the results.  We really have 'supported' or 'required'
compilers and 'the rest'.

> >> > and instead I see a bunch of failures we never really tried to
> >> > resolve.  I think we should remove these from the user pages....
> >> 
> >> On what basis?
> >
> > It's just extra noise for most users -- not many people use dmc or
> > tru64.  And it could reduce their confidence in boost...
> 
> I tend to share the sentiment, but I'd still prefer the main
> page/release notes to have direct links to both sets of pages --
> correspondingly marked up, of course.

I'm fine with that although it introduces more stuff on the frontpage which
I'm generally against.  What we really need to do is rework the
compiler_status.html page to explain the distinctions and groupings better (it
has other problems too).

In the meantime, however, I'd still like to see the release status page routed
only to the 'required compilers' page for the reasons previously cited.  I
think that optimizes the 90% usage case not the 10% case...

Jeff