[erlang-questions] Erlang code loading

[Max Bourinov]

I tend to think that Ladislav Lenart is right... At least I don't see any
reason why actual code should be loaded more that once.

Waiting for Erlang Super Master to clarify....

Best regards,
Max




2011/11/21 Max Bourinov <bourinov@REDACTED>

> Does anybody knows how it exactly runs?
>
> Best regards,
> Max
>
>
>
>
> 2011/11/21 CGS <cgsmcmlxxv@REDACTED>
>
>> **
>> Sorry for not being explicit. I was referring not only to the data, but
>> also to the code itself. As far as I noticed, Erlang does not make any
>> difference between loading the same module in different processes and
>> different modules in different processes. Each process instance is
>> independent and does not share any bit of its memory with other instances
>> (it can copy, but does not share). I might be wrong in my observation,
>> though (only a developer can be 100% sure).
>>
>> My test consisted in starting the same piece of code in two different
>> Erlang sessions (processes as you call them) and I noticed the same memory
>> consumption for each (top -d 1 -p <PID of the system process where your
>> code is loaded>). You can test that as well and compare it with the beam
>> size. As I said, I might be wrong because my code has a certain level of
>> granularity which doesn't allow very large functions (I avoid large
>> functions for better monitoring the states).
>>
>> I hope this answer will at least help you to convince yourself about
>> which answer is correct.
>>
>> CGS
>>
>>
>>
>>
>>
>>
>> On 11/21/2011 12:51 PM, Max Bourinov wrote:
>>
>> About the process data it is absolutely clean how Erlang handles it. I
>> just want some clarification about the code.
>>
>>
>> Best regards,
>> Max
>>
>>
>>
>>
>> 2011/11/21 CGS <cgsmcmlxxv@REDACTED>
>>
>>>  One instance per process, one memory allocation per process (even if it
>>> is the same code running). Erlang has no explicit knowledge of beam shared
>>> objects as far as I know. If somebody knows differently, please, correct me.
>>>
>>> CGS
>>>
>>>
>>>
>>>
>>>
>>> On 11/21/2011 12:40 PM, Max Bourinov wrote:
>>>
>>> Thank you for your reply! But in Erlang processes, not threads. There
>>> are some differences between them...
>>>
>>>
>>>
>>> Отправлено с iPhone
>>>
>>> 21.11.2011, в 15:05, CGS <cgsmcmlxxv@REDACTED> написал(а):
>>>
>>>   Hi, Max,
>>>
>>> Not an expert, but I think the memory is multiplied by the number of
>>> threads. Meaning, you start with an initial heap for that thread and add to
>>> that the states and other internal variables (repeat that for all the
>>> threads). At least that I noticed from my codes by now. If I am wrong, I
>>> apologize.
>>>
>>> Cheers,
>>> CGS
>>>
>>>
>>>
>>>
>>> On 11/21/2011 11:57 AM, Max Bourinov wrote:
>>>
>>> Hi Erlangers,
>>>
>>>  I am wondering how my Erlang app would behave in this case:
>>>
>>>  I have a lot identical processes (the same module, state of couse is
>>> different in each process). The module that I swap many times has a huge
>>> function. I want to clarify how much memory it will take? It will take
>>> memory for that huge function in each process or only once?
>>>
>>>  I believe it will take memory only once, but I want to make sure.
>>>
>>> Best regards,
>>> Max
>>>
>>>
>>>
>>> _______________________________________________
>>> erlang-questions mailing listerlang-questions@REDACTED://erlang.org/mailman/listinfo/erlang-questions
>>>
>>>
>>>   _______________________________________________
>>> erlang-questions mailing list
>>> erlang-questions@REDACTED
>>> http://erlang.org/mailman/listinfo/erlang-questions
>>>
>>>
>>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20111121/8011910c/attachment.htm>