bug#42141: blast+ is not reproducible

[Maxim Cournoyer]

Hi,

Ricardo Wurmus <rekado@elephly.net> writes:

> zimoun <zimon.toutoune@gmail.com> writes:
>
>> Hi,
>>
>> On Thu, 29 Sep 2022 at 23:09, Maxim Cournoyer <maxim.cournoyer@gmail.com> 
>> wrote:
>>> Ricardo Wurmus <rekado@elephly.net> writes:
>>>
>>>> Fixed with commit 1ee2d117d8fa9e2e0d4ec46cc5555497bb5e6337.
>>>
>>> Yay!  Thank you!  And for my curiosity, how did I get the two build
>>> farms to agree on an identical build, as reported in 'guix challenge'?
>>>
>>> That is odd.
>>
>> Because blast+ is multi-outputs and ’out’ is fine contrary to ’include’:
>>
>> $ guix challenge blast+:include
>> /gnu/store/0kbsdr61qpj0vkc6s8g2kbp4dq936n0p-blast+-2.11.0-include contents 
>> differ:
>>   local hash: 0q4nssknlmc54m8abndn9bhrlwm4m28lkb75i1wnwr0ghbalj02x
>>   
>> https://ci.guix.gnu.org/nar/lzip/0kbsdr61qpj0vkc6s8g2kbp4dq936n0p-blast%2B-2.11.0-include:
>>  0q4nssknlmc54m8abndn9bhrlwm4m28lkb75i1wnwr0ghbalj02x
>>   
>> https://bordeaux.guix.gnu.org/nar/lzip/0kbsdr61qpj0vkc6s8g2kbp4dq936n0p-blast%2B-2.11.0-include:
>>  0cakizfsqb1lla62cmwnng1h9gvqgf3lyjk0k7lkiaisj713mpzx
>>   differing files:
>>     /include/ncbi-tools++/ncbi-tools++/ncbi_random_macro.h
>>     /include/ncbi-tools++/ncbi-tools++/ncbiconf_unix.h
>>
>> 1 store items were analyzed:
>>   - 0 (0.0%) were identical
>>   - 1 (100.0%) differed
>>   - 0 (0.0%) were inconclusive
>
> Ah, thanks for explaining!

Ah!  Shouldn't the default be to compare all outputs?  It seems that'd
be less surprising and useful.

What do you think?

-- 
Maxim