From: Vesa Karvonen (vesa_karvonen_at_[hidden])
Date: 2003-04-15 02:56:40
>The sheer amount of hacks contained in the pp-lib to make things work
>on broken preprocessors is evidence of this. It contains nearly four
>separate implementations of the same thing--just for that purpose.
hmm... Neither Boost.Config nor the regression logs reveal the buggy
preprocessors. I think that this is a problem. The effort that goes into
writing the workarounds goes essentially unnoticed. Also, people can not see
how buggy the compilers really are.
I think that it would be wise to write a failing regression test for every
distinct kind of workaround - not just for the PP library. Anyone, including
compiler vendors, would then be able to easily see how each compiler is
doing. I know that having to write such compiler bug tests is at least as
displeasing as writing workarounds, but otherwise the effort that goes into
producing the workarounds goes almost totally unnoticed.
The way the workaround are written in the PP library, for example, would
make it quite easy to show just how many workarounds are needed on buggy
preprocessors. Basically, one could write a simple test for every PP library
interface macro. Then those tests can easily be run both with and without
STOP MORE SPAM with the new MSN 8 and get 2 months FREE*
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk