capacity/limit is misdocumented

[ExplodingCabbage]

client_reference.rst:

• Still documents the old limit parameter, not the new capacity one
• Lists the default value for all the connector subclasses as None, rather than 20.

I'd PR a fix, but as I noted in #1601 (comment) I disagree with @fafhrd91's change to the limit/capacity functionality so I'll wait for that argument to be resolved first - no sense documenting something that I'm currently arguing you guys should revert!

[fafhrd91]

i agree with your reasoning, but i still think we need capacity feature.

i suggest rename capacity to max_connections and add max_connections_per_host

[asvetlov]

Let don't invite very_long_names_in_java_style.
This is documentation issue.
capacity is good name, if we document limit as obsolete and deprecated the meaning of attribute is obvious. limit is also perfect name but already used.

@ExplodingCabbage regarding to your specific case (web scrapper throttling) -- please use explicit semaphores or something like this. Or, maybe, create a ClientSession instance per site.

aiohttp should have perfect defaults for generic cases.
I believe it means that ClientSession should behave like browser.
Most users don't care about connection limits but they are very sensitive to memory/resource leaks.

[fafhrd91]

i think we should support per host limit.
only reason is, limit per host it is not that trivial to implement.

[asvetlov]

good point.
Then let's choose capacity and capacity_per_host as alias for deprecated limit.
They are shorter than max_connections and add max_connections_per_host.
I believe total limit is much more wider used than limit per host.

[fafhrd91]

but maybe we should just leave limit in place, just change it's meaning.
and add limit_per_host for old limit meaning.

[fafhrd91]

i renamed capacity back to limit and added limit_per_host

[lock]

This thread has been automatically locked since there has not been
any recent activity after it was closed. Please open a new issue for
related bugs.

If you feel like there's important points made in this discussion,
please include those exceprts into that new issue.