Omit CPU type from man page

[bmwiedemann]

When a reproducible build is wanted,
we should not add the build CPU type to man pages like ldmd2.1

See https://reproducible-builds.org/ for why this is matters.

The specification for SOURCE_DATE_EPOCH is at
https://reproducible-builds.org/specs/source-date-epoch/
and everyone doing reproducible builds, sets it.

[bmwiedemann]

Alternative would be to apply a sed in man page generation to drop the CPU line

[kinke]

Lgtm, any objections?

[JohanEngelen]

This to me looks strange. The output of ldc2 --version should not need this special case. While generating the man page, just sed the line out. (I guess the problem is that ldc2 --help is used for generating the man page?)
If you still want to continue with this, at the very least it needs an explaining comment of what is going on here.

[bmwiedemann]

ldc2 --help does not capture the CPU type. But ldmd2 --help does. And technically, your own machine's CPU type is not part of the ldc version or help and most other compilers do not include it there, so a third option would be to avoid the special case and just drop it.

[bmwiedemann]

some context from the ldc2 man-page:

       Options  marked with (*) also have a -disable-FOO variant with inverted
       meaning.

              based on DMD v2.080.1 and LLVM 6.0.1 built with LDC - the LLVM D
              compiler  (0.17.5) Default target: x86_64-unknown-linux-gnu Host
              CPU: bdver1 http://dlang.org - http://wiki.dlang.org/LDC

              Registered Targets:

       aarch64
              - AArch64 (little endian)

              aarch64_be [...]

[bmwiedemann]

https://travis-ci.org/ldc-developers/ldc/jobs/420270765 failed with

gnumake: *** [/Users/travis/build/ldc-developers/ldc/runtime/druntime-test-shared/finalize.done] Illegal instruction: 4

[bmwiedemann]

@JohanEngelen added an explaining comment in the code
my tests with the patch look good.

[bmwiedemann]

Can this be merged now?

[kinke]

Yeah, I guess the chance of some other distro/packager benefitting from this is higher than someone using SOURCE_DATE_EPOCH by accident and caring about the host CPU output.