Re: [Boost-testing] OSL3 will run GCC 2.95.3 tests nightly on Linux

From: Beman Dawes (bdawes_at_[hidden])
Date: 2005-06-22 07:00:28

"Martin Wille" <mw8329_at_[hidden]> wrote in message 
news:42B87DE6.5070001_at_yahoo.com.au...
> Doug Gregor wrote:
>> Martin, if you'd like to drop support for GCC 2.95.3, go right ahead.
>> We definitely don't need more than one tester on it, because we're not
>> sure how prevalent it is anymore, and you've mentioned that it sucks up
>> a lot of resources for you.
>
> Yes, this improves things for me.
>
> Will OSL also run gcc-2.95.3-stlport tests?

At some point we need to step back and ask "what is the purpose of running 
regression tests?"

Seems to me the primary purpose is to quickly identify errors in Boost code 
and quickly verify that Boost code works on the platforms important to 
users.

For gcc-2.95.3-stlport, it is hard to see how those tests would ever 
identify errors in Boost code that aren't already detected by more modern 
compilers. And while gcc-2.95.3 itself may still be important to a few 
users, I'd really be surprise if more than one or two users care about the 
gcc-2.95.3-stlport combination.

So drop the gcc-2.95.3-stlport tests, IMO.

--Beman