[erlang-questions] relx extended deps

e@REDACTED e@REDACTED
Tue Jan 27 17:07:41 CET 2015


On 01/27/2015 05:04 PM, Anthony Ramine wrote:
> Le 27 janv. 2015 à 16:58, e@REDACTED a écrit :
>
>> On 01/27/2015 04:53 PM, Anthony Ramine wrote:
>>> Le 27 janv. 2015 à 16:49, e@REDACTED a écrit :
>>>
>>>> On 01/27/2015 04:47 PM, Anthony Ramine wrote:
>>>>> Le 27 janv. 2015 à 16:44, e@REDACTED a écrit :
>>>>>
>>>>>> how am i running ranch without ssl now?
>>>>>
>>>>> By complaining to Ranch's author that he should either remove the code depending on ssl and put it in another separate project, or properly make his project depend on ssl and thus just not let you use ranch without ssl.
>>>>
>>>> hold on, i am not into political question, i want to keep it technical.
>>>> i am running ranch without ssl.
>>>> doesn't it contradict the statement of dependency?
>>>
>>> And anyone actually using ranch with ssl now needs an additional non-standard 'sequence' key in their app file.
>>
>> on the contrary!
>> it is ranch who needed this additional 'sequence' key.
>> because 'ranch' knows the requirement,
>> while a top-app does not have to know.
>>
>> you may call it a 'conditional-dep'
>> where the condition is triggered by the top-app
>> (without awareness)
>
> They need the machinery behind it, and which is unknown to OTP itself, even if they didn't write the actual 'sequence' property themselves. I am against adding conditional dependencies, given they are mostly a way for developers to be lazy when it comes down to splitting their own projects decently.

sorry, i failed to understand the roles behind 'they' in your statement.




More information about the erlang-questions mailing list