|
|
Boost Testing : |
From: Sean Huang (huangsean_at_[hidden])
Date: 2006-06-27 14:20:43
----- Original Message -----
From: "Thomas Witt" <witt_at_[hidden]>
Newsgroups: gmane.comp.lib.boost.testing,gmane.comp.lib.boost.build
Cc: <boost-build_at_[hidden]>
Sent: Tuesday, June 27, 2006 1:16 PM
Subject: Re: [1.34.0]Required platforms - Intel on Windows
>>>The question of how we distinguish between 7.1 and 8.0 based compilation
>>>is not relevant for 1.34.0. We picked from the list of available
>>>regression tests and that's the end of that.
>>>
>>
>>????
>>The current situation is that we have different results for the same
>>toolset (better: we would have them if we didn't distinguish the VC
>>version). How would you suggest to handle that without including the
>>name of the base compiler into the toolset?
>
>Well the reason for this is that someone started tp provide test results
>for the branch that created a name clash. This is the root cause of the
>problem. We were fine before that. It all boils down to the fact that we
>need stability in testing to get a release out.
Achieving stability by doing something obviously wrong? The name clash is
because the system had a bug that needs to be fixed. And that's the root
cause of the problem. I DO NOT think it was fine to get something released
by hiding a bug. To quote your required platform:
<snip>
msvc-7.1
msvc-8.0
<snip>
intel-win-9.1
intel-linux-9.0
<snip>
How can you say intel-win-9.1+VC7.1 is the required platform while
intel-win-9.1+VC8 is not based on this list.
Anybody who has used Intel's compiler knows it is not a "complete" package
meaning it has to used with another compiler. As I have said in my previous
post, not supporting intel-win9.1+VC8 would be a big dissappoint as many
peopele move to 9.1 because they want to use VC8.
>>>These questions should be resolved for the next release. Until then I
>>>kindly ask you to resolve the issue in one of the possible ways ASAP.
>>>
>>>Sorry for being rude, but every day we spend on this is a day of delay
>>>for the release.
>>>
>>
>>Nobody wants to hold the release but for me, after having seen the
>>differences between Intel on VC7 and Intel on VC8, this seems to be
>>necessary.
>
>I am not opposed to the toolset rename but in this case please fix
>explicit-failures-markup.xml ASAP.
Glad you finally said this.
Sean