Re: [boost] Reminder: The Master branch will be closing for the 1.77.0 release on WEDNESDAY 4-August
On 3/08/2021 11:57 am, Andrey Semashev wrote:
The docker image that the tools get run from doesn’t have “linux/magic.h”
More as we discover it.
I would like to know more about this issue. In particular, what Linux system lacks this file (distro, kernel version, etc.). Maybe, this is worth fixing before the release.
Git history suggests that the file started existing in linux kernel 2.6.19 and remains since. Although it moved from one folder to another in the internal kernel source tree in kernel 3.7, but that shouldn't affect its existence in the userspace include dirs, unless something is using the wrong include paths.
On 8/3/21 3:58 AM, Gavin Lambert via Boost wrote:
On 3/08/2021 11:57 am, Andrey Semashev wrote:
The docker image that the tools get run from doesn’t have “linux/magic.h”
More as we discover it.
I would like to know more about this issue. In particular, what Linux system lacks this file (distro, kernel version, etc.). Maybe, this is worth fixing before the release.
Git history suggests that the file started existing in linux kernel 2.6.19 and remains since.
Although it moved from one folder to another in the internal kernel source tree in kernel 3.7, but that shouldn't affect its existence in the userspace include dirs, unless something is using the wrong include paths.
Thanks. 2.6.19 was released nearly 15 years ago, I wouldn't think such old systems would be building new Boost. I can add a workaround if that's needed.
On 8/3/21 11:55 AM, Andrey Semashev wrote:
On 8/3/21 3:58 AM, Gavin Lambert via Boost wrote:
On 3/08/2021 11:57 am, Andrey Semashev wrote:
The docker image that the tools get run from doesn’t have “linux/magic.h”
More as we discover it.
I would like to know more about this issue. In particular, what Linux system lacks this file (distro, kernel version, etc.). Maybe, this is worth fixing before the release.
Git history suggests that the file started existing in linux kernel 2.6.19 and remains since.
Although it moved from one folder to another in the internal kernel source tree in kernel 3.7, but that shouldn't affect its existence in the userspace include dirs, unless something is using the wrong include paths.
Thanks. 2.6.19 was released nearly 15 years ago, I wouldn't think such old systems would be building new Boost. I can add a workaround if that's needed.
Here is the workaround: https://github.com/boostorg/filesystem/commit/9794725bda2e612f1fdcadd98cdce5... I would like to merge it to master please.
On 8/3/21 04:50, Andrey Semashev via Boost wrote:
On 8/3/21 11:55 AM, Andrey Semashev wrote:
On 8/3/21 3:58 AM, Gavin Lambert via Boost wrote:
On 3/08/2021 11:57 am, Andrey Semashev wrote:
The docker image that the tools get run from doesn’t have “linux/magic.h”
More as we discover it.
I would like to know more about this issue. In particular, what Linux system lacks this file (distro, kernel version, etc.). Maybe, this is worth fixing before the release.
Git history suggests that the file started existing in linux kernel 2.6.19 and remains since.
Although it moved from one folder to another in the internal kernel source tree in kernel 3.7, but that shouldn't affect its existence in the userspace include dirs, unless something is using the wrong include paths.
Thanks. 2.6.19 was released nearly 15 years ago, I wouldn't think such old systems would be building new Boost. I can add a workaround if that's needed.
Here is the workaround:
https://github.com/boostorg/filesystem/commit/9794725bda2e612f1fdcadd98cdce5...
I would like to merge it to master please.
The docker image that was having issues combines the test runner results into the test matrix. It doesn't build all of boost, just a couple utilities for that purpose. It didn't have the kernel headers installed. The workaround looks fine. Please merge to master. Thanks Andrey. -- Michael Caisse Ciere Consulting ciere.com
participants (3)
-
Andrey Semashev -
Gavin Lambert -
Michael Caisse