This is a question that's been asked before, and that has popped up again in https://github.com/openssl/openssl/issues/6765
Our current mechanism for creating tarballs for a new OpenSSL release is to run 'make dist' in any given build tree... it's a bit clumsy, as it needs a wasted configuration if it's done in a newly checked out work tree, but is designed to work correctly from any build tree. The original intention (way back, I think we're even talking SSLeay time here, but at the very least pre-1.0.0 time) was to make a tarball that can be built directly with just a 'make' on any Unix box and without requiring perl. Since 1.1.0, though, the tarballs do require perl to generate certain files, such as include/openssl/opensslconf.h. That makes a pre-configured distribution less benefitial. Also, if anyone tries to run 'nmake' on Windows without configuring first, they get a nasty and quite confusing surprise... I think the same happens on VMS, although I haven't tested that. I can see two way to fix this: 1. Don't release pre-configured tarballs. This is a very simple thing to do, all we have to do is use 'make tar' to create tarballs instead of 'make dist'. We could remove the dist target entirely while we're at it. 2. Restore the no-perl benefit with a tarball distributed with 'make dist' (which is very simple due to 'make build_all_generated'). 3. Have the 'dist' config target generate a really dumbed down Makefile that contains the same well known targets as the usual build files, but makes sure to run some kind of fancy script (supposedly in perl) that runs a proper configuration for the platform at hand. (actually, the first item doesn't depend on the rest, but the answer will direct our focus) Cheers, Richard -- Richard Levitte levi...@openssl.org OpenSSL Project http://www.openssl.org/~levitte/ _______________________________________________ openssl-project mailing list openssl-project@openssl.org https://mta.openssl.org/mailman/listinfo/openssl-project