time to reconsider CURL_GLOBAL_ACK_EINTR ?

Hi,

As we're starting to discuss fixing the thread-safety issues in
curl_global_init(), we need to clean up the parts of that logic that is our
own creation.

One of those things is the CURL_GLOBAL_ACK_EINTR flag. It is stored in a
global variable and thus not thread-safe. It is documented as:

   "When this flag is set, curl will acknowledge EINTR condition when
   connecting or when waiting for data. Otherwise, curl waits until full
   timeout elapses."

1. Can we perhaps change to *always* abort these select() loops on EINTR ?
Surely we don't get signals that often so that it will make a significant
difference?

2. Should we make this a curl_easy_setopt() option instead? There are some
call path done in the multi interface that doesn't actually have an easy
handle. Either we have it as a curl_multi_setopt() option as well or maybe we
just acknowledge it by default there?

-- 
  / daniel.haxx.se | Get the best commercial curl support there is - from me
                   | Private help, bug fixes, support, ports, new features
                   | https://www.wolfssl.com/contact/
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette:   https://curl.haxx.se/mail/etiquette.html