From: Jeff Hostetler <git@jeffhostetler.com>
Date: Thu, 07 Dec 2017 15:43:02 GMT
Subject: Re: partial_clone_get_default_filter_spec has no callers
Message-ID: <7a7400a1-0743-152a-fa3b-740c3b2b8e04@jeffhostetler.com>
In-Reply-To: <44d0c556-d5ba-186f-7493-f9900eccb442@ramsayjones.plus.com>



On 12/6/2017 8:59 PM, Ramsay Jones wrote:
> 
> 
> On 06/12/17 21:07, Jeff Hostetler wrote:
>>
>>
>> On 12/6/2017 12:39 PM, Ramsay Jones wrote:
>>> Hi Jeff,
>>>
>>> commit f1862e8153 ("partial-clone: define partial clone settings
>>> in config", 2017-12-05), which is part of your 'jh/partial-clone'
>>> branch, introduces the partial_clone_get_default_filter_spec()
>>> function without any callers. Could you please confirm that this
>>> is intentional and that, presumably, a future series will include
>>> a call to this function.
>>
>> I'll double check.  Thanks.
>>
>> BTW is there another tool that you're using to find these?
>> I know I ran make DEVELOPER=1 and make sparse on everything
>> and didn't see that come up.
> 
> In addition to sparse (which finds some of these), I also run a perl
> script over the object files after a given build. (The script was
> posted to the list by Junio, many moons ago, and I have made several
> changes to my local copy).
> 
> I am attaching a copy of the script (static-check.pl). Note that the
> 'stop list' in the script (%def_ok) is _way_ out of date. However, the
> way I use the script, that does not matter; I run the script over the
> master->next->pu branches and (ignoring the master branch) diff the
> result files from branch to branch. For example, tonight I have:
> 
>    $ wc -l sc nsc psc
>      74 sc
>      73 nsc
>      75 psc
>     222 total
>    $
>    $ diff sc nsc
>    44d43
>    < oidmap.o	- oidmap_remove
>    $
>    $ diff nsc psc
>    43a44
>    > list-objects-filter-options.o	- partial_clone_get_default_filter_spec
>    58a60
>    > sequencer.o	- sign_off_header
>    $
> 
> You also have to be careful with leaving stale object files
> laying around from previous builds ('make clean' sometimes
> doesn't). Actually, it may be simpler to read a previous mailing
> list thread on exactly this subject [1].
> 
[...]
> 
> ATB,
> Ramsay Jones
> 
> [1] https://public-inbox.org/git/%3Cb21c8a92-4dd5-56d6-ec6a-5709028eaf5f@ramsayjones.plus.com%3E/
> 

thanks!  maybe you could post something (in contrib/ perhaps)
that would run your script on a pair of commits like t/perf/run.sh.
just a thought.

Jeff