I was in doubt whether to split it into 2 patches (because they touch single file), ok, I'll do that
вт, 11 окт. 2022 г. в 10:19, Willy Tarreau <w...@1wt.eu>: > Hi Ilya, > > On Sat, Oct 08, 2022 at 08:17:26PM +0500, ???? ??????? wrote: > > Hello, > > > > when OPENSSL_VERSION=latest (same for LIBRESSL_VERSION=latest) was > > introduced, I made a mistake, and resolved version was generated as > > OPENSSL=3.0.5 which makes no sense to build-ssl.sh, proper version should > > have been OPENSSL_VERSION=3.0.5 > > > > temporarily we stick to LIBRESSL=3.5.3, because of few rergressions in > > recently released 3.6.0 > > > > cheers, > > Ilya > > > From 4298e0251a08649dd650294a2bdd1bb700f38e42 Mon Sep 17 00:00:00 2001 > > From: Ilya Shipitsin <chipits...@gmail.com> > > Date: Sat, 8 Oct 2022 20:00:41 +0500 > > Subject: [PATCH] CI: SSL: use proper version generating when "latest" > semantic > > is used > > > > both "OPENSSL_VERSION=latest" and "LIBRESSL_VERSION=latest" processing > > introduced errors when build-ssl.sh script was invoked. that error > > in turn led to skipping custom openssl build and haproxy was linked > against > > stock openssl, i.e. openssl-1.1.1 > > > > also, we temporarily stick to LibreSSL=3.5.3, because recently released > > 3.6.0 introduced some regression which must be resolved first > > These are two completely independent things, they should be in separate > patches, because if we apply as-is, once the libressl issue is solved, > if we just revert this patch, it'll reintroduce the issue with openssl > that this patch solves. > > Would you please care to split that in two ? > > Thanks, > Willy >