Re: [Boost-testing] [iostreams][Beman] what GCC version are yourunning?

From: Douglas Gregor (doug.gregor_at_[hidden])
Date: 2005-06-13 07:53:12

On Jun 13, 2005, at 7:01 AM, Beman Dawes wrote:
> "Douglas Gregor" <doug.gregor_at_[hidden]> wrote in message
> news:0430dca6e40db881dbd8791ae3031048_at_cs.indiana.edu...
>>
>> On Jun 12, 2005, at 4:28 PM, Beman Dawes wrote:
>>
>>>
>>> "Jonathan Turkanis" <technews_at_[hidden]> wrote in message
>>> news:d8d2cd$67k$1_at_sea.gmane.org...
>>>> Hi,
>>>>
>>>> Despite the fact that you GCC toolset is not marked as required, I'm
>>>> concerned
>>>> about the runtime failure of file_descriptor_test.cpp.
>>>
>>> I've just copied gcc-tools to gcc-3-3-3-cygwin-tools and changed the
>>> batch
>>> file that runs the tests accordingly.
>>
>> gcc-3_3_3-cygwin would be slightly preferred, because it matches our
>> naming conventions.
>
> OK, I'll change it. But...
>
> Regardless of the name, copying and renaming locally didn't work.

My gcc-3_3-darwin toolset looks like this:

{
   extends-toolset darwin ;
   PYTHON_VERSION = 2.3 ;
}

Is that roughly what you're doing?

> The
> regression.py script deleted the file when it unzipped the fresh Boost 
> tree.

You can put the toolset file in the same place where regression.py 
resides, outside of the Boost tree, then regression.py will do the 
right thing.

> Also, depending on the exact toolset name is error prone and requires 
> manual
> intervention every new compiler version. Too easy for a compiler to be
> upgraded, but the tool name not updated.

Yes, you're right.

> compiler_status used to get the compiler version from the config
> information. That was automatic and reliable. Any idea why it was 
> changed?

I don't, but I'm sure someone else knows.

        Doug