Re: [isar 0/1] Fix non-reproducible issue due to localepurge

[Jan Kiszka <jan.kiszka@siemens.com>]

On 14.04.22 10:08, Henning Schild wrote:
> Am Wed, 13 Apr 2022 12:15:47 +0000
> schrieb <Venkata.Pyla@toshiba-tsip.com>:
> 
>> Hi Henning Schild,
>>
>> Thank you for the comments.
>>
>> Please find my comments inline.
>>
>>> -----Original Message-----
>>> From: isar-users@googlegroups.com <isar-users@googlegroups.com> On
>>> Behalf Of Henning Schild
>>> Sent: 11 April 2022 16:24
>>> To: pyla venkata(ＴＳＩＰ) <Venkata.Pyla@toshiba-tsip.com>
>>> Cc: isar-users@googlegroups.com; jan.kiszka@siemens.com; dinesh
>>> kumar(ＴＳＩＰ) <dinesh.kumar@toshiba-tsip.com>; hayashi kazuhiro(林
>>> 和宏 □ＳＷＣ◯ＡＣＴ) <kazuhiro3.hayashi@toshiba.co.jp>
>>> Subject: Re: [isar 0/1] Fix non-reproducible issue due to localepurge
>>>
>>> Am Sun, 10 Apr 2022 22:32:28 +0530
>>> schrieb venkata.pyla@toshiba-tsip.com:
>>>  
>>>> From: venkata pyla <venkata.pyla@toshiba-tsip.com>
>>>>
>>>> Hello all,
>>>>
>>>> Replying to this mail conversation,
>>>> (https://groups.google.com/g/isar-users/c/JCgpWU_pFJU/m/IwlYTmzQBQAJ)
>>>>
>>>> For the problem "debconf cache non-reproducible", I figured out a
>>>> solution in isar system itself.  
>>>
>>> Generating configuration items and placing them before a package gets
>>> installed is very common. It is not only done by Isar but very
>>> likely by many packages in debian that cater for other packages.
>>>
>>> What you are suggesting is not a generic solution to "debconf cache"
>>> being non-reproducible. It is just a quirk to hide the problem in
>>> that one case for that one package.  
>>
>> Yes I agree with you, Ideally this should be fixed from each package
>> if they have such non-reproducible behaviour.
> 
> I would want to bet that there are some packages which show similar
> problems directly from upstream, but likely more in downstream isar
> layers. Because a debian package on its own will simply bring its
> default config, every deviation will come from something else ... like
> ansible, puppet, isar ... 
> 
>> I reported a bug today in the Debian package bug-tracking system,
>> though I don’t have solution for that, but I want to start discussion
>> and get some solution going forward.
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1009639
> 
> I am not sure that is a valid debian but either, they could just tell
> you to install first and configure later. The non default config is not
> coming from debian.
> 
>> Meanwhile we can use this trick to bypass non-reproducible problem,
>> what's your opinion?
> 
> I would prefer a way/idea to deal with such problems in
> general. Like a quality checker script that would warn about a
> populated conf cache, or a postprocess rule to clear the cache.
> 
> That would be a real good thing, while your patch is only a workaround
> for an exotic potential problem most people do not really care too much
> about.
> We know anyways that an isar build is never 100% reproducible ... just
> try a wic run and binary diff the resulting images instead of diffing
> the files on mounted partitions.
> 

Are wic-related differences coming from the Isar-way of using wic? Or
are we sharing issues with OE/Yocto? They claim to be reproducible. We
will have to eventually as well.

Jan

-- 
Siemens AG, Technology
Competence Center Embedded Linux