From: Vladimir Prus (ghost_at_[hidden])
Date: 2006-10-10 03:51:58
On Sunday 08 October 2006 12:01, Bojan Resnik wrote:
> > I've just committed a large change to PCH handling. The bulk of this
> > change was done by Ilya Sokolov and adds PCH support for the gcc toolset.
> > Also, now instead of 'pch' rule, there are 'cpp-pch' and 'c-pch' rules --
> > which explicitly indicate the language of the header.
> > I've also took the change to address the need to use 'cast pchheader' --
> > now this should not be required, and won't work anyway.
> > I've also committed automated test for PCH functionality, which works on
> > gcc.
> > I was no able to test with msvc, although I carefully looked at command
> > lines, and they look correct. I'd appreciate if somebody test this on
> > msvc, and try running the 'pch.py' test as well.
> > Finally, an example 'pch' was added to the example directory.
> > - Volodya
> Hi, Vladimir.
> I ran the tests today with CVS head, using all versions of MSVC on
> my machine. The commands were
> set DO_DIFF=diff
> test_all.py msvc-6.0 > msvc-6.0.txt 2>&1
> test_all.py msvc-7.1 > msvc-7.1.txt 2>&1
> test_all.py msvc-8.0 > msvc-8.0.txt 2>&1
> There are errors and failed tests, but the results are pretty much
> the same with all these compilers. Attached is the MSVC 8 output.
A couple of issues are related to msvc.compile.pch being invoked when it
should not be, and should be fixed now. Two more failures are caused by
sequence.unique changes, I'll look at them.
There are two failures caused by different names of created targets -- I did
not know about them and will look.
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk