Re: [boost] [build] Regression tests failing across the board

Subject: Re: [boost] [build] Regression tests failing across the board
From: Vladimir Prus (vladimir_at_[hidden])
Date: 2011-02-03 07:25:28

Beman Dawes wrote:

> On Thu, Feb 3, 2011 at 6:57 AM, Vladimir Prus <vladimir_at_[hidden]> wrote:
>> Beman Dawes wrote:
>>> ...
>>> If the build test failures can't be cleared from the reporting, the
>>> tests themselves should be removed.
>>
>> I'll work on some cleanup -- presumably, only running the tests for
>> runners that are known to have no configuration issues.
> 
> Out of curiosity, what are the main configuration issues? Can they be
> detected and turned into a warning rather than a failure?

This, unfortunately, requires investigation. Basically, there are:

1. One test that fails everywhere. This one I'll just fix.

2. Two or three tests that fail for some testers. These are something I'll
look into.

3. Several test runners where most of the tests fail, with error
messages that suggest that compiler cannot be invoked (whereas it's
invoked just fine for library tests). The information in regression
table is not particularly helpful, so I either have to make tests
report more information, or adjust regression reporting, or contact
runners, or just suppress testing. 

-- 
Vladimir Prus
Mentor Graphics
[phone]