Items for Friday

[Brent Putman]

On 1/30/17 5:28 AM, Rod Widdowson wrote:
> I know we have discussed this to death (since before V3 it seems), and I know Brent and Ian have spent quite some time doing
> research but I am struggling to find the result.  I'm hoping that someone can point me at it and I can close off IDP-1115 and remove
> the agenda item, otherwise we need to bottom this out and then I can close off IDP-1115

I have to run to a meeting and will look into in detail later, but real
quick:  We did "resolve" (at least at that time) the HttpClient timeout
issues from JSPT-48.  It's in the linked issue JSPT-66.  The default
internal in the HttpClientBuilder is 60 seconds for all 3 timeouts.

Haven't looked at IDP-1115 yet in detail, but glancing at the summary
I'm 99.9% sure the delta is deliberate.  We make the timeout different
for the batch vs dynamic cases, that was part of what we did back in
the fall before the release. We decided 60s is too long a timeout for
the dynamic case, since it's directly user-visible, being "inline" in
the user request flow.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://shibboleth.net/pipermail/dev/attachments/20170130/d0bb1fdf/attachment.html>