Remmina 1.2 backport (was: Re: Bug#771784: Additional info)

[Andreas Metzler]

On 2017-09-15 Raphaël Halimi <raphael.halimi at gmail.com> wrote:
[backporting Remmina]
> * libssh 0.8 builds under Stretch, but one of the tests failed with
> libgcrypt 1.7.6, so 1.7.8 (or later) had to be backported also.
[...]

Hello,

I have just spent some time debugging this. torture_options failing is
not related to a specific libgcrypt version. The test fails if
GCRYPT_VERSION in gcrypt.h does not match "[0-9]+.[0-9]+.[0-9]+".

This happens when building gcrypt from git, e.g. a build of current
LIBGCRYPT-1-7-BRANCH head will have

#define GCRYPT_VERSION "1.7.10-beta"

If the abovementioned match does not succeed config.h ends up with
/* #undef HAVE_GCRYPT_ECC */
/* #undef HAVE_ECC */

I suspect from there on follows some contradicting views on what gcrypt
is believed to support. - These two #defines are missing, but there are
other condionals defines relating to GCRYPT_VERSION_NUMBER.
Anyway, this is a bug in cmake/Modules/FindGCrypt.cmake.

So a gcrypt backport is not needed.

You might be wondering why libgcrypt 1.7.6-2+deb9u2 triggers the bug.
Well, that is
https://bugs.launchpad.net/ubuntu/+source/libgcrypt20/+bug/1700157 which
was fixed 1.7.8-1.

cu Andreas
-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'