[Buildroot] RFC: package/boost: build failure

[Buildroot] RFC: package/boost: build failure

[Andreas Ziegler]

Dear all,

package/boost currently fails to build for certain architectures (64bit 
w/ 128bit integer processing), resulting in the following build failure 
[1]:

/home/autobuild/autobuild/instance-4/output-1/host/mips64-buildroot-linux-gnu/sysroot/usr/include/boost/intrusive/hashtable.hpp: 
In static member function 'static std::size_t 
boost::intrusive::prime_list_holder<Dummy>::position(std::size_t, 
std::size_t)':
/home/autobuild/autobuild/instance-4/output-1/host/mips64-buildroot-linux-gnu/sysroot/usr/include/boost/intrusive/hashtable.hpp:222:69: 
error: 'sizes' was not declared in this scope; did you mean 'size'?
   222 |          return fastmod_u32(hash, inv_sizes32[size_index], 
uint32_t(sizes[size_index]));
       |                                                                  
    ^~~~~
       |                                                                  
    size

The development branch of boost/intrusive contains a fix for this 
failure which, unfortunately, is rather longish [2]. Easy extraction of 
the relevant part of the patch is not possible, due to massive changes 
to the whole code.

I can think of three options to deal with this issue:

   (1) Use the commit [2] as is, even if it changes more than the failing 
code
   (2) Create a custom diff for prime_list_holder::position()
   (3) Live with the failure until boost 1.81 is released

Incidence rate of (3) is low; numbers (1) and (2) both deviate from the 
Buildroot preferred approach, so my question would be: how to proceed 
here?

Kind regards,
Andreas

[1] 
http://autobuild.buildroot.net/results/6cb/6cbc0672f3c5046405793d8cdc8f961d2ffb1d3e
[2] 
https://github.com/boostorg/intrusive/commit/825aedad557b7e3bb27614335b6ced16887103de
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot

Re: [Buildroot] RFC: package/boost: build failure

[Thomas Petazzoni via buildroot]

On Mon, 07 Nov 2022 06:38:43 +0000
Andreas Ziegler <br015@umbiko.net> wrote:

> The development branch of boost/intrusive contains a fix for this 
> failure which, unfortunately, is rather longish [2]. Easy extraction of 
> the relevant part of the patch is not possible, due to massive changes 
> to the whole code.
> 
> I can think of three options to deal with this issue:
> 
>    (1) Use the commit [2] as is, even if it changes more than the failing 
> code

To me the commit [2] is not that large, and would be OK to backport as
an interim solution until 1.81 is released.

Thanks a lot for this investigation work, much appreciated!

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot