2018-02-05 18:29 GMT+03:00 Daniel Stenberg <[email protected]>: > On Fri, 2 Feb 2018, Paul D Rotter wrote: > >> We define USE_OPENSSL in our project, so USE_WIN32_CRYPTO has always been >> off as we do use OpenSSL. The problem with USE_WIN32_CRYPTO being >> unconditionally defined is it results in USE_NTLM2SESSION not being defined >> in curl_ntlm_core.h: >> >> Was this change by design? > > > It looks like this was a (side-)effect of 24bba4045677 [1] brought in PR > #1943 [2] and I don't think it was intentional. > > Any suggestions on how we should proceed to fix it? > > [1] = https://github.com/curl/curl/commit/24bba4045677 > [2] = https://github.com/curl/curl/pull/1943 >
Initially, I've addressed the issue with OpenSSL <-> WinCrypt for NTLM implementation here https://github.com/curl/curl/commit/a8135f0768f4723db7bfdc1fecc7c054a73a2dbd#diff-af3b638bc2a3e6c650974192a53c7291 I still think that those options should be mutually exclusive until OpenSSL is not a "first-class citizen" on Windows. But I can be wrong. The main issue lack of tests for functionality which those flags enable on Windows. IMO, it is too many flags. Any objections are welcome. > -- > > / daniel.haxx.se > ------------------------------------------------------------------- > Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library > Etiquette: https://curl.haxx.se/mail/etiquette.html -- Best Regards, Sergei Nikulov ------------------------------------------------------------------- Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library Etiquette: https://curl.haxx.se/mail/etiquette.html
