[SRU][O/N][PATCH v2 0/1] reenable libtraceevent

Bethany Jamison bethany.jamison at canonical.com
Fri Aug 30 16:21:12 UTC 2024


Hello Stefan

Sorry for the delay in response - I had never made my own PPA and built 
with it so I got to learn how to do that first. I uploaded my revert to 
both a Noble and Oracular PPA on Launchpad and they built successfully. 
However, since my first submission one of the files in my Oracular patch 
was modified and there's a conflict with my first patch set, so I'll be 
re-sending the updated patches.

Thanks -- Bethany


On 8/16/24 6:31 AM, Stefan Bader wrote:
> On 15.08.24 19:41, Bethany Jamison wrote:
>> BugLink: https://bugs.launchpad.net/bugs/2076190
>>
>> Reenable LIBTRACEEVENT - it was disabled in commit 92550b568d (UBUNTU:
>> [Packaging] perf: build without libtraceevent) to temporarily reduce
>> the amount of build dependecies. According to LP#2076190 disabling
>> libtraceevent has reduced much of the functionality of perf. Perf is
>> no longer able to perform any operations on tracepoints.
>>
>> It looks like libtraceevent was originally disabled for no other
>> reason than to reduce the number of build dependencies - and it even
>> seems like this disablement was not intended to be permenent (from
>> commit message: "Disable libtraceevent in perf to avoid adding extra
>> build dependencies for now.") I think there is no significant tradeoff
>> in reenabling libtraceevent as it is fixing a regression in
>> functionality introduced by commit 92550b568d.
>
> Ok, that is something to work with. I guess the problem is potentially 
> not avoiding but reducing the dependencies. That can save setup time 
> and maybe compile time. Which is something one never is completely 
> sure what the initial reason was. I vaguely recall that there was talk 
> about intending to package things differently. Unfortunately that is 
> nothing I am following closely.
> One question here, have you tried to upload source packages with the 
> revert to a PPA? Why I am asking is that build dependencies are 
> declared in debian/control which is generated partially from 
> debian.*/debian.control.stub. If libtraceevent was disabled to avoid 
> dependencies, those likely need to be added again together with the 
> revert. But test compiling with cbd or our builders does not catch 
> this as they work with pre-set setup.
>
>>
>> v2: explain purpose of patch in greater detail as well as include a
>> buglink
>
> The BugLink should be in each patch. But that could be done when 
> applying. Provided there is no other reason to re-submit (above).
>
>>
>> Bethany Jamison (1):
>>    UBUNTU: [Packaging] perf: reenable libtraceevent
>>
>>   debian/rules.d/2-binary-arch.mk  | 4 ++--
>>   debian/rules.d/3-binary-indep.mk | 4 ++--
>>   2 files changed, 4 insertions(+), 4 deletions(-)
>>
>
> - Stefan
>



More information about the kernel-team mailing list