From: Rene Rivera (grafik.list_at_[hidden])
Date: 2003-10-19 20:45:08
[2003-10-19] Robert Ramey wrote:
>I've recently updated my CVS tree for the development version of
>boost and found the following differences when I try to build
>and run tests from with in a current test directory
>a) The test executables are built in
Yep, per design.
>b) process_jam_log seems to work as one would
>expect on the new directory
>c) I can't generate compiler status tables for one test
>$ compiler_status --compiler vc7 c:/boost-dev/ compiler_status.html
>**** exception(205): std::exception: boost::filesystem::directory_iterator
>ructor: "c:\boost-dev\libs\serialization\test\bin": The system cannot find
>******** errors detected; see standard output for details ********
>suggesting that this program hasn't be updated to reflect a change
>in expected directory structure.
It was updated. I just was not aware of the use case of only generating
results for a subset of the tests at the time.
>among trying different options I get
>what is the recommended way to build and run tests for one library?
Just as you did before... Like I mentioned in a earlier email... This is now
>While we're on the subject of compiler_status program, I would much like
>to see a switch added "--target=<target name>" so that I could run
>the tests built in release mode. Perhaps other switches as well. This
>would permit me to run tests in release mode with varying combinations
>of libraries and maintain one table.
Not sure what you mean? I run the release version of the regression tests.
See the "Linux_RSI" results. Are you using the "-sBUILD=release" option?
-- grafik - Don't Assume Anything
-- rrivera (at) acm.org - grafik (at) redshift-software.com
-- 102708583 (at) icq
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk