Package: nim
Version: 0.17.2-1
Severity: serious
Tags: sid buster
User: pkg-openssl-de...@lists.alioth.debian.org
Usertags: openssl-1.1-trans
Control: block 871056 by -1

Please migrate to libssl-dev in the Buster cycle. I am very sorry for
this late report but this package was never on my list. It slipped
because it never B-D on libssl1.0-dev.

I looked at the package and the easy build fix is to add the 1.1 prefix
to the libs it looks (in lib/wrappers/openssl.nim). The way it is now,
the package (nimble binary) should not work if libssl-dev is installed
because it parses the .so versions from left to right and first one is
empty. So there is this which already should affect Stretch.

There are a lot of defines and function names copied from the openssl's
header files. Someone needs to double check those that there still the
same.

Function wise:
SSL_library_init() and a few other macros towards "OPENSSL_init_ssl(0, NULL)"
  so "normal" C will work but if nim is accessing the functions directly
  then it will fail.

SSLv23_client_method() and friends are also macros. If I read this right
  then if it is not available, then it will invoke TLSv1_method(). This
  will work in 1.1 but please don't do this. The problem with
  TLSv1_method() is that it will give you _only_ TLSv1 and _never_
  TLSv1.1, and/or TLSv1.2 like SSLv23_client_method(). The "v23" functions
  should be around (for 1.0.2 and earlier) and for 1.1 you would to use
  TLS_client_method().
  The v2 and v3 (and the TLSv1) could be disabled at build time. If you
  want to exclude a certain TLS version you should use something like 
  SSL_OP_NO_TLSv1 to disable TLSv1 only (and keep other version like
  v1.1 and v1.2 around).

Data strucures. All structures are opaque and you need to tell libssl to
allocate it and free it. This means you can't have them on stack or
dereference them. It seems that md5_* (md5_File) uses them on Stack.

A larger collection of what changed in OpenSSL 1.0.2->1.1 is at
   https://wiki.openssl.org/index.php/OpenSSL_1.1.0_Changes

Sebastian

Reply via email to