Re: [gentoo-dev] old "masked for testing" entries
> "KF" == Kristian Fiskerstrand > writes: KF> I'm not familiar with any large difference. I only mentioned sks because it is the only heavy user of berk db I currently run. Most either moved on to other libs or I use w/ pg. I did get the impression from the sks list that db5 worked better than db4, though. Or perhaps that was something which sleapycat fixed in more recent versions of 4, too? KF> I'm testing with 5.2 for my live SKS ebuild which I've been using KF> for quite some time on a few of my servers as backends of the KF> load-balanced without any issues, KF> Upgrading is relatively easy, mostly involving cleaning the KF> environment, which will be re-generated with the updated version. The issue seen on debian was that the tools for 5.1 were used by the upgrade script when the sks-dependent-on-5.3 was released, but there was no dependency so apt didn't know to ensure that the binary dpkg required was installed. That shouldn't be an issue on Gentoo, given that the programs installed with a given db SLOT are not dependent on any USE flags and the parallel versions tend to remain longer. It seems, even though I only mentioned it in an aside, I could have thought of a better example. -JimC -- James Cloos OpenPGP: 0x997A9F17ED7DAEA6
Re: [gentoo-dev] old "masked for testing" entries
> "MG" == Michał Górny writes: MG> Dnia 2014-06-30, o godz. 17:40:16 MG> James Cloos napisał(a): >> So it should be fine to unmask 5.3 and slowly update reverse >> dependencies to depend on 5.3 instead of whichever 4.x they >> currently demand. >> Unmasking the earlier 5.x releases seems unnecessary, though. MG> While at it, please don't unmask <5.3.28-r1 (EAPI<5 ebuilds). Multilib MG> relies heavily on not having anything EAPI<5 over 4.8.30-r1. Good point. I should have more precise and said "the most recent ebuild in the 5.3 SLOT", which is what I was thinking. -JimC -- -JimC -- James Cloos OpenPGP: 0x997A9F17ED7DAEA6
Re: [gentoo-dev] old "masked for testing" entries
Dnia 2014-06-30, o godz. 17:40:16 James Cloos napisał(a): > I've read that some heavy users of db, such as sks, work better with 5.3 > than they did with older versions. But upgrading needs to be done with care. > > So it should be fine to unmask 5.3 and slowly update reverse > dependencies to depend on 5.3 instead of whichever 4.x they currently > demand. > > Unmasking the earlier 5.x releases seems unnecessary, though. While at it, please don't unmask <5.3.28-r1 (EAPI<5 ebuilds). Multilib relies heavily on not having anything EAPI<5 over 4.8.30-r1. -- Best regards, Michał Górny signature.asc Description: PGP signature
Re: [gentoo-dev] old "masked for testing" entries
-BEGIN PGP SIGNED MESSAGE- Hash: SHA512 On 06/30/2014 11:40 PM, James Cloos wrote: >> "MG" == Mike Gilbert writes: > > I've read that some heavy users of db, such as sks, work better > with 5.3 than they did with older versions. But upgrading needs to > be done with care. > I'm not familiar with any large difference. I'm testing with 5.2 for my live SKS ebuild which I've been using for quite some time on a few of my servers as backends of the load-balanced without any issues, I've not tried 5.3 yet. The net-misc/sks package for ~arch is still on 4.8 without any issues on the rest of the servers. Upgrading is relatively easy, mostly involving cleaning the environment, which will be re-generated with the updated version. - -- - Kristian Fiskerstrand Blog: http://blog.sumptuouscapital.com Twitter: @krifisk - Public PGP key 0xE3EDFAE3 at hkp://pool.sks-keyservers.net fpr:94CB AFDD 3034 5109 5618 35AA 0B7F 8B60 E3ED FAE3 - Nomina stultorum scribuntur ubique locorum Fools have the habit of writing their names everywhere -BEGIN PGP SIGNATURE- iQIcBAEBCgAGBQJTsd0hAAoJEPw7F94F4TagCmsP/0fNyRcCDgpEpscr2nM3gAnC ccO8vmB44lAXNh4Zq01hb334vBl+qDT7fJDQDt/Jb39NKA/2zMQskKzVkDa1vnaj 4YHE7pfx128GtAuUCWZgKipkMUG1hKEuL3GZL2KaaoJ/zk6JYLW/z2ZZuo99N7JI ks0ltdIwPNTmj8YVxFQKqo9Nbb9yrrMNYGfxPmFQdSRr5ZkM+ieOihkh95ICTz5C qfyn2eh85mVPS+aMN8MzuQQLE5hMauQ+YohiZ3Q4NIx0hdjTesEbzBi1hDIOng74 nbxnCZ/7E2VNMRC/wOFRVcYxr/D3gETVlwvm2VIsALxF6jdYkWcJmkKM19fURAaO fS5FQVRcXRBL2/1spBSH6Z8/EoYATl1YsylLmncGfOwaD3gAUHZhRqXar82HvSFC uH5MYS96D63loicUE4VDPGekfbKzm8jOhhPXWOCvNDdyQIpitvwe4YdJefNQv3B5 8ybmFygQO9/zL0pUwBx8z2V0ciJ+Jb6Vx1+nMrjxYakkJUVsJAnGodeEODh7E4eY 3vAst/005dPPPuQbHbxZgS+01ZnRu1Ed1uqfxM0mQpTNKyUg+Dr4lLzEblzGP0AT ct//vkAREZGftWnegsSGGVcT1YHm5GVN3gvtIF5UEWLUJvzlJZlS/rE9rTQopq4U 2QyV6VUKzk9NHnDYf9YE =AHJf -END PGP SIGNATURE-
Re: [gentoo-dev] old "masked for testing" entries
> "MG" == Mike Gilbert writes: MG> For example, I think the major reason for the sys-libs/db mask is a MG> weird licensing issue. It's still nice to have it in the tree. That only applies to db:6.0. I know debian and ubuntu primarily use 5.3 these days, with the only issues being related to upgrading existing stores to the newer formats as they release newer versions and re-compilations of the reverse dependencies linked against 5.3. I've read that some heavy users of db, such as sks, work better with 5.3 than they did with older versions. But upgrading needs to be done with care. So it should be fine to unmask 5.3 and slowly update reverse dependencies to depend on 5.3 instead of whichever 4.x they currently demand. Unmasking the earlier 5.x releases seems unnecessary, though. -JimC -- James Cloos OpenPGP: 0x997A9F17ED7DAEA6
Re: [gentoo-dev] old "masked for testing" entries
On Mon, Jun 30, 2014 at 04:46:04PM -0400, Mike Gilbert wrote: > On Mon, Jun 30, 2014 at 12:38 PM, William Hubbs wrote: > > All, > > > > Rich Freeman asked, in another thread, for specific examples of old > > package.mask entries that just have "masked for testing" as the > > description. > > > > Here is what I found with a quick look through package.mask. These > > should be cleaned up by either 1) removing the mask or 2) booting the > > affected packages or versions from the tree. > > > > Or 3) Improve the mask message. > > For example, I think the major reason for the sys-libs/db mask is a > weird licensing issue. It's still nice to have it in the tree. If there is a weird licensing issue, it would be better to get the answers needed to resolve that issue and remove the mask. p.mask should never be permanent. William signature.asc Description: Digital signature
Re: [gentoo-dev] old "masked for testing" entries
On Mon, Jun 30, 2014 at 12:38 PM, William Hubbs wrote: > All, > > Rich Freeman asked, in another thread, for specific examples of old > package.mask entries that just have "masked for testing" as the > description. > > Here is what I found with a quick look through package.mask. These > should be cleaned up by either 1) removing the mask or 2) booting the > affected packages or versions from the tree. > Or 3) Improve the mask message. For example, I think the major reason for the sys-libs/db mask is a weird licensing issue. It's still nice to have it in the tree.
[gentoo-dev] old "masked for testing" entries
All, Rich Freeman asked, in another thread, for specific examples of old package.mask entries that just have "masked for testing" as the description. Here is what I found with a quick look through package.mask. These should be cleaned up by either 1) removing the mask or 2) booting the affected packages or versions from the tree. # Sergei Trofimovich (05 Jan 2013) # Masked for testing. Is not compatible with cvsps-2 (bug #450424). # But can be used on it's own! Try 'cvsps --fast-export'. >=dev-vcs/cvsps-3 # Tim Harder (27 Nov 2012) # Masked for testing =media-libs/libsfml-2* # Markos Chandras (03 Nov 2012) # Masked for testing app-benchmarks/ltp # Robin H. Johnson (08 Oct 2012) # Masked for testing =sys-libs/db-6.0* =sys-libs/db-5.3* =sys-libs/db-5.2* =sys-libs/db-5.1* =sys-libs/db-5.0* # Ultrabug (16 May 2012) # Masked for testing >=sys-cluster/corosync-2.0.0 # MATSUU Takuto (27 Oct 2011) # Mask for testing >=sys-devel/distcc-3.2_rc1 # Christian Faulhammer (12 Mar 2011) # Mask for testing >=www-apps/joomla-1.6.0 # Michael Sterrett (20 Jan 2010) # testing mask for upcoming exult release >=games-engines/exult-1.3 # Vlastimil Babka (20 May 2008) # Masked for testing app-arch/rpm5 Thanks, William signature.asc Description: Digital signature