On Tue, 26 Sep 2017, Ray Satiro via curl-library wrote:

> Ok how about if I make a PR so when pthread option is enabled first check
> for -lpthread, then fall back on -pthread (FYI I think this is from the 90s,
> I don't know how much difference it will make), then if that doesn't work
> try checking for the function without either (in case there's a glib that
> rolls it up or something).

Will that really make any difference?

To me, the source of the problem seems to be that the test program doesn't
cause any link error when -fsanitize=address is used. Thus the need to provide
an option to add pthreads isn't detected by configure.

I would probably consider it a slightly better idea to just draw the
conclusion that when the threaded resolver is used on non-windows, we MUST
link with pthreads so we add that option unconditionally without checking if
we really need to add the option.

-- 
  / daniel.haxx.se
-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette:   https://curl.haxx.se/mail/etiquette.html