[Boost-testing] regression.py/Boost.Build v2 updates

From: Reece Dunn (msclrhd_at_[hidden])
Date: 2006-02-26 06:50:10

I have fixed several issues with the V2 port of regression.py which I have 
made available at 
https://zigzag.cs.msu.su/boost.build/wiki/BoostRegressionsV2. This includes:

*  the mvsc --> msvc fix;
*  the fix for deleting unicode directories (already in the V1 version);
*  a fix for preventing a Traceback when using --monitored (although I 
haven't been able to get monitored builds working yet);
*  a fix for cygwin/python which doesn't appear to have platform.system();
*  fixed the process_jam_log source_dir where regression.py looks for the 
exe (it is in boost_dir/dist/bin);
*  fixed the log function to keep the output in sync, so you don't end up 
with output like:

# Building "process_jam_log.exe" 
("c:\devel\regression\boost\tools\jam\src\bin.n
tx86\bjam.exe" --v2 "-sBOOST_BUILD_PATH=c:\devel\regression" 
"-sBOOST_ROOT=c:\de
vel\regression\boost" msvc)...

c:\devel\regression>cd "c:\devel\regression\boost\tools\regression\build"

C:\devel\regression\boost\tools\regression\build>"c:\devel\regression\boost\tool
s\jam\src\bin.ntx86\bjam.exe" --v2 "-sBOOST_BUILD_PATH=c:\devel\regression" 
"-sB
OOST_ROOT=c:\devel\regression\boost" msvc
warning: toolset python initialization: can't find tool python2.4
warning: initialized from
Building Boost.Regex with the optional Unicode/ICU support disabled.# 
Searching
for "process_jam_log.exe" in "c:\devel\regression\boost\dist\bin"...
# process_jam_log.exe succesfully built in 
"c:\devel\regression\boost\dist\bin\p
rocess_jam_log.exe" location

Please refer to the Boost.Regex documentation for more information
(and if you don't know what ICU is then you probably don't need it).
...patience...
...found 459 targets...
# Making "c:\devel\regression\results" directory...

The last fix is done by chanding log to:

def log( message ):
    sys.stdout.flush()
    sys.stderr.flush()
    sys.stderr.write( '# %s\n' % message )
    sys.stderr.flush()

NOTE: This is not an ideal fix if the stdout and stderr streams are 
interleaved before the call to flush (e.g. OEOE would become OOEE where O is 
stdout and E is stderr output).

There are probably a few other fixes I have made, but they are the main 
ones.

- Reece