Re: [boost] Issues with compiler_status

From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-10-23 09:24:27

At 03:47 PM 10/22/2003, Robert Ramey wrote:

 >A couple of questions regarding compiler_status and the build system in
 >general.
 >
 >a) Here is the output from a test
 >
 >- <test-log library="" test-name="test_contained_class_xml_archive"
 >test-type="" test-program=""
 >target-directory="bin/boost/libs/serialization/test/test_contained_class_xml
 >_archive.test/gcc/debug" toolset="gcc" show-run-output="true">
 ><lib result="fail" timestamp="2003-10-22 00:38:09
...
 >when processed by compiler_status.cpp it is marked as "Passed" so 
something
 >doesn't seem right here.

There is clearly a problem, but I'm somewhat unsure exactly where.

 >This was a test on the gcc compiler 3.2

Does the same problem happen with other compilers?

 >b) The directory structure ...

Rene or one of the other build experts needs to answer those questions. 
Note that there is a mailing list (jamboost at yahoogroups.com) just for 
build issues.

...
 >I have to say I've found the build/test/status system incredibly useful 
for
 >library development. I realise
 >that its a tough job and that efforts are underway to improve it.  I 
would
 >much like to see it have "first class status" as a boost library 
component
 > - much like the test system does. I would find it
 >very useful in other work I do besides boost. It seems that most if of it
 >is there.  Its just that it doesn't seem "polished".   I would guess that
 >much of this would be remedied by better documents.   Maybe
 >or maybe something has to be re-thought to make it easier to understand
 >and/or trace.  I don't know.

The current approach is not very robust. Since bjam doesn't know anything 
about building the .xml output, changes to toolsets or bjam message outputs 
tend to break XML generation (process_jam_log.cpp).

Once Boost.Build V2 goes into production use, we might want to look at 
generating the XML directly. We now know what the XML needs to contain, so 
that eliminates one of the past impediments.

--Beman