Boost 1.54 build problem: --without-context not honoured?
Hi, I'm digging into a build problem and noticed that even though I specify "./bjam --without-context", Boost.Context is still built! The bjam configuration output states that Context is not to be built: Component configuration: - atomic : building - chrono : building - context : not building - coroutine : building - date_time : building - exception : building But it is built nonetheless. This causes a build failure for architectures which are not supported; see build log [1]. Compare that to the build for 1.53 [2] which works fine. I've diffed the libs/context directory between 1.53 and 1.54 and nothing jumps out at me as the cause. Has anyone noticed and fixed this already? Thanks, -Steve [1] https://buildd.debian.org/status/fetch.php?pkg=boost1.54&arch=ia64&ver=1.54.... [2] https://buildd.debian.org/status/fetch.php?pkg=boost1.53&arch=ia64&ver=1.53....
I'm digging into a build problem and noticed that even though I specify "./bjam --without-context", Boost.Context is still built! The bjam configuration output states that Context is not to be built:
Component configuration:
- atomic : building - chrono : building - context : not building - coroutine : building - date_time : building - exception : building
But it is built nonetheless. This causes a build failure for architectures which are not supported; see build log [1]. Compare that to the build for 1.53 [2] which works fine.
Perhaps it's because coroutine depends on context, so building coroutine triggers a build of context anyways? Try adding --without-coroutine as well. Regards, Nate
Hi, What I understand from this thread is that you need to specify all the dependent library from not being build if you want to disable some components from build. This means that the user should know the library dependency explicitly when compiling in order to build boost. From the users' point of view, I think that this behavior is not good. The build system should honor the users' explicit argument. On Sun, Jul 7, 2013 at 9:34 AM, Nathan Ridge <zeratul976@hotmail.com> wrote:
I'm digging into a build problem and noticed that even though I specify "./bjam --without-context", Boost.Context is still built! The bjam configuration output states that Context is not to be built:
Component configuration:
- atomic : building - chrono : building - context : not building - coroutine : building - date_time : building - exception : building
But it is built nonetheless.
When I run bjam with the argument "--without-context" explicitly, I assume that context is not built regardless of the dependency of "context", since this is what I specify explicitly. Of course, the warning on "coroutine", something like: WARN: coroutine: not building (depends on context) might be necessary. On the other hand, if you specify like "--with-coroutine", "context" should be built and the info about "context" being built should be in the log. Best regards, -- Ryo IGARASHI, Ph.D. rigarash@gmail.com
On July 6, 2013 07:04:26 PM Steve M. Robbins wrote:
Hi,
I'm digging into a build problem and noticed that even though I specify "./bjam --without-context", Boost.Context is still built!
Checking further, I see that Boost.Coroutine uses Boost.Context so I have to disable both to avoid building Context. I guess the fact that the 1.53 build worked was a fluke. -Steve
AMDG On 07/06/2013 05:04 PM, Steve M. Robbins wrote:
Hi,
I'm digging into a build problem and noticed that even though I specify "./bjam --without-context", Boost.Context is still built! The bjam configuration output states that Context is not to be built:
The reason it's being built is because Boost.Coroutine depends on Boost.Context.
Component configuration:
- atomic : building - chrono : building - context : not building - coroutine : building - date_time : building - exception : building
But it is built nonetheless. This causes a build failure for architectures which are not supported; see build log [1]. Compare that to the build for 1.53 [2] which works fine.
I've diffed the libs/context directory between 1.53 and 1.54 and nothing jumps out at me as the cause. Has anyone noticed and fixed this already?
In Christ, Steven Watanabe
participants (4)
-
Nathan Ridge -
Ryo IGARASHI -
Steve M. Robbins -
Steven Watanabe