[jira] [Updated] (IGNITE-14911) Inconsistent and confusing timeout properties in python thin client

["Ivan Daschinsky (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IGNITE-14911?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ivan Daschinsky updated IGNITE-14911:
-------------------------------------
    Summary: Inconsistent and confusing timeout properties in python thin client  (was: Inconsistent and confused timeout properties in python thin client)

> Inconsistent and confusing timeout properties in python thin client
> -------------------------------------------------------------------
>
>                 Key: IGNITE-14911
>                 URL: https://issues.apache.org/jira/browse/IGNITE-14911
>             Project: Ignite
>          Issue Type: Task
>          Components: python, thin client
>    Affects Versions: python-0.4.0
>            Reporter: Ivan Daschinsky
>            Assignee: Ivan Daschinsky
>            Priority: Blocker
>              Labels: python, thin
>             Fix For: python-0.5.0
>
>
> Currently, in python thin client timeout values can be set in int and float.
> 1. Timeouts for sql and cache properties are set using int as milliseconds
> 2. Timeouts for tx and expiry policy can be set either in float (as seconds) and in int as milliseconds (not released yet)
> But in python traditionally all timeouts are set using float and int as *seconds*. Milliseconds precision are achieved using fractions (f.e. 120 ms are set as 0.12)
> We can solve this issue in many ways
> 1. Completely change to traditional way (and probably broke backward compatibility)
> 2. Change logic as for transactions for all timeouts with deprecation warning and with advice using only float.
> 3. Change logic for transaction or expiry_policy traditionally, for all others timeouts as in point 2. (This is not consitent)
> 4. Set all timeouts in ints and milliseconds (except for socket timeout in `Connection`) and optionally for ExpiryPolicy accept also `datetime.timedelta`. 
> I suppose, that forth variant is the best option.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)