Tom Kent wrote:

I brought it up last week on the testing list. I think it is somehow related to symlinks being generated for the headers. I'm going to try removing everything from that directory and trying a fresh run.

Tom, you might want to omit the --force-update option: Command Line run.py --runner=teeks99-01a-win2008-64on64 --force-update \ --toolsets=msvc-11.0 "--bjam-options=\"-j4 address-model=64\"" \ --comment=..\\info.html --tag=develop It's SVN-specific and might be causing the problems.

On Sat, Mar 1, 2014 at 2:05 AM, Andrey Semashev wrote:

On Monday 24 February 2014 11:55:59 Tom Kent wrote:

...

On Mon, Feb 24, 2014 at 11:45 AM, Andrey Semashev

wrote:

...

Hi,

I'm seeing test compilation errors saying that some headers are missing. For example:

http://www.boost.org/development/tests/develop/developer/output/teeks99-01 a-win2008-64on64-boost-bin-v2-libs-log-test-attr_attribute_value_impl-test -msvc-11-0-debug-threading-multi.html

says that boost/smart_ptr/intrusive_ptr.hpp does not exist. There are other missing headers. The tester is teeks99-01a-win2008-64on64. Is this a known problem?

I brought it up last week on the testing list. I think it is somehow related to symlinks being generated for the headers. I'm going to try removing everything from that directory and trying a fresh run.

Apparently, that helped, thanks. At least teeks99-01a-win2008-64on64 is all green now. But all teeks99-02* testers exhibit the same problem.

The --force-update didn't do anything. Turning it off didn't change any results. I tried removing the headers' directory (boost_root/boost) between each run, but that did not help. Finally, I had to simply delete the entire boost_root repository from each of my runs, that seems to have helped. Unfortunately, it would be bandwidth prohibitive to re-clone the repo each run, so I have a master copy of the repo that I am copying down into each run directory before the run. Not an idea situation, but because of the way the submodules are setup, I can't do local clones :-( I'm worried that the process of bootstrapping boost (and doing a regression run) hasn't been thoroughly tested on windows and the problems I was seeing might come back to bite us in the future. If people think they are getting an updated copy, but it has issues, that's not a good situation. Tom