On Thu, Dec 21 2017, Jeremie Courreges-Anglas <[email protected]> wrote:
> On Thu, Dec 21 2017, Jeremie Courreges-Anglas <[email protected]> wrote:
>> On Fri, Dec 15 2017, Landry Breuil <[email protected]> wrote:
>>> On Thu, Dec 14, 2017 at 10:58:11PM +0100, Rafael Sadowski wrote:
>>>> Hi All!
>>>> 
>>>> After a long time and many setbacks I'm happy to send this mail. Please 
>>>> find
>>>> attached a diff as tarball to update boost to the latest stable version 
>>>> 1.65.1.
>>>> 
>>>> Notable port changes.
>>>> 
>>>> - Of course bump all shared libs from 6.0 to 7.0.
>>>> - COMPILER= base-clang ports-gcc (Replace ONLY_FOR_ARCHS)
>>>
>>>
>>> Fwiw i fully support this moving on, eventual fallout can be dealt later
>>> in-tree. But how much was it built-tested on a ports-gcc arch, at least
>>> macppc|sparc64 ? :)
>>
>> Using Rafael's second tarball, it builds and packages fine on sparc64,
>> games/bastet builds and runs (using ports-gcc or ports-clang, btw).
>> The machine I'm doing tests on doesn't have much resources so don't
>> expect more results.
>
> net/powerdns also seems to build fine on sparc64.

yaml-cpp, innoextract...

>> For $reasons I have not tried building the updated boost on arm yet,
>> tarball currently unpacking...
>
> Fails to build on arm.  I don't understand why sparc64 is fine but arm
> isn't, both use ports-gcc.  The symbols could be found in libunwind, or
> we could disable the backtrace functionality.

After a chat, Rafael wrote an updated diff[1] that disables the
boost/stacktrace lib on all architectures.  This lib has been added in
boost-1.65.0 so probably no application uses it yet.

Diff attached for convenience, packages fine on amd64, build still
ongoing on arm.

[1] http://ix.io/DgF

Attachment: boost-1.65.1-rsadowski-3.diff.gz
Description: Binary data

-- 
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF  DDCC 0DFA 74AE 1524 E7EE

Reply via email to