[Buildroot] [PATCH v3] package/trace-cmd: bump to version 2.9.1
Arnout Vandecappelle
arnout at mind.be
Wed Dec 2 17:18:01 UTC 2020
On 02/12/2020 17:37, Norbert Lange wrote:
> Am Di., 1. Dez. 2020 um 21:35 Uhr schrieb Arnout Vandecappelle <arnout at mind.be>:
>>
>> Hi Norbert,
>>
>> On 03/11/2020 16:53, Norbert Lange wrote:
[snip]
>>> define TRACE_CMD_BUILD_CMDS
>>> - $(MAKE) $(TARGET_CONFIGURE_OPTS) \
>>> - CFLAGS="$(TRACE_CMD_CFLAGS)" \
>>> - CPPFLAGS="$(TRACE_CMD_CPPFLAGS)" \
>>> - $(TRACE_CMD_MAKE_OPTS) \
>>> - -C $(@D) all
>>> + $(TARGET_CONFIGURE_OPTS) \
>>> + CFLAGS="$(TRACE_CMD_CFLAGS)" \
>>> + CPPFLAGS="$(TRACE_CMD_CPPFLAGS)" \
>>
>> Can you explain in the commit message why these are passed in the environment
>> now instead of on the command line?
>
> I do a "normal" build now, instead of using a fragment in a subdirectory.
Err... Before it was "make all", where do you see the anything about a fragment
in a subdirectory?
> Means the Makefile would like to do stuff like adding include paths which doesnt
> work if the variables are overriden.
Yes, that's exactly the kind of explanation I would expect in the commit message.
>
> As rationale I would use: "Thats the way you normally build packages",
> aswell as "Thats how you would DL the source" (instead of doing a git
> fetch, in regards to last point).
It was git before because kernel.org has only recently started supporting
tarball downloads from git (where "recently" is about a year ago I think, maybe
even two).
Regards,
Arnout
> Of course I don't know why it was done in a entirely different manner
> before, so I might be missing something
>
> Norbert
>
More information about the buildroot
mailing list