Bug 546124 – [TRACKER] >=net-libs/gnutls-3.5.0 breakage

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Patrick Lauer 2015-04-10 07:37:21 UTC]

# Patrick Lauer <[e-mail]> (10 Apr 2015)
# Breaks pretty much all consumers, like samba
# Mask until it's more usable
>=net-libs/gnutls-3.4.0

[Comment 1 Graham Murray 2015-04-10 07:47:14 UTC]

I think the problem is that the openssl compatible library is no longer built by default, so either a USE flag for this should be introduced or the library built unconditionally

[Comment 2 Alon Bar-Lev (RETIRED) 2015-04-10 08:31:16 UTC]

(In reply to Graham Murray from comment #1)
> I think the problem is that the openssl compatible library is no longer
> built by default, so either a USE flag for this should be introduced or the
> library built unconditionally

thanks!

added openssl USE for but gnutls-3.3.14 and gnutls-3.4.0.

[Comment 3 linedot 2015-04-11 16:32:35 UTC]

*** Bug 546274 has been marked as a duplicate of this bug. ***

[Comment 4 Lars Wendler (Polynomial-C) (RETIRED) 2015-04-13 12:41:15 UTC]

Let's make this a tracker bug...

[Comment 5 Martin Jansa 2015-04-14 11:09:03 UTC]

There is also an issue in dev-libs/efl, when built with gnutls USE flag (even when gnutls itself is built with openssl USE flag):
https://www.mail-archive.com/[e-mail]/msg1314316.html

[Comment 6 Jeroen Roovers (RETIRED) 2015-04-17 04:25:11 UTC]

(In reply to Martin Jansa from comment #5)
> There is also an issue in dev-libs/efl, when built with gnutls USE flag

Please file a new bug report for each issue.

[Comment 7 Lars Wendler (Polynomial-C) (RETIRED) 2016-05-10 11:46:38 UTC]

commit b9a33361371825153728d97d0503a42eb8c62d10
Author: Lars Wendler <[e-mail]>
Date:   Tue May 10 13:44:32 2016

    package.mask: Unleash >=net-libs/gnutls-3.4.0 to ~arch.
    
    Signed-off-by: Lars Wendler <[e-mail]>

[Comment 8 Alon Bar-Lev (RETIRED) 2016-05-10 11:49:58 UTC]

Hi,
I committed gnutls-3.5.0, so basically nobody will test 3.4 now.
Do you think we should mask the 3.5?
Thanks!

[Comment 9 Graham Murray 2016-05-10 11:55:04 UTC]

(In reply to Alon Bar-Lev from comment #8)
> Hi,
> I committed gnutls-3.5.0, so basically nobody will test 3.4 now.
> Do you think we should mask the 3.5?
> Thanks!

According to the release notes and documentation, 3.5 is backward compatible (at least at source code level) with 3.4 - in other words none of the removal of deprecated functions which plagued the 3.3 to 3.4 transition. So if it builds with 3.4 then it should build and work with 3.5.

So if 3.4 is going ~arch then there should be no reason to mask 3.5.

[Comment 10 Alon Bar-Lev (RETIRED) 2016-05-10 11:57:10 UTC]

Do you want to skip stabilization of 3.4?

[Comment 11 Kristian Fiskerstrand (RETIRED) 2016-05-11 17:38:09 UTC]

(In reply to Alon Bar-Lev from comment #10)
> Do you want to skip stabilization of 3.4?

3.4 is current stable so I would recommend stabilizing that over 3.5 at the present time

[Comment 12 Alon Bar-Lev (RETIRED) 2016-05-11 19:59:58 UTC]

(In reply to Kristian Fiskerstrand from comment #11)
> (In reply to Alon Bar-Lev from comment #10)
> > Do you want to skip stabilization of 3.4?
> 
> 3.4 is current stable so I would recommend stabilizing that over 3.5 at the
> present time

So we need to mask >=3.5
Polynomial-C?

[Comment 13 Kristian Fiskerstrand (RETIRED) 2016-05-12 09:12:05 UTC]

(In reply to Alon Bar-Lev from comment #12)
> (In reply to Kristian Fiskerstrand from comment #11)
> > (In reply to Alon Bar-Lev from comment #10)
> > > Do you want to skip stabilization of 3.4?
> > 
> > 3.4 is current stable so I would recommend stabilizing that over 3.5 at the
> > present time
> 
> So we need to mask >=3.5


Yeah, that sounds like a prudent choice to me to ensure proper testing of 3.4 before stabilization.

[Comment 14 Lars Wendler (Polynomial-C) (RETIRED) 2016-05-12 23:10:38 UTC]

I have no issues with masking gnutls-3.5.x although I am under the impression that there's no real difference in testing gnutls-3.4.x or gnutls-3.5.x.

[Comment 15 Alon Bar-Lev (RETIRED) 2016-05-13 05:06:15 UTC]

I masked >=gnutls-3.5.0, we leave this as long as upstream does not treat the 3.5 series as stable.
This will enable us to stabilize the 3.4 in the mean time.