Re: [gentoo-dev] pet feature: mtime preservation (was [gentoo-dev-announce] Council Summary from meeting on April 09, 2009)

2009-04-16 Thread Ciaran McCreesh
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On Wed, 15 Apr 2009 02:15:46 +0200
Marijn Schouten (hkBst) hk...@gentoo.org wrote:
 and from #gentoo-council:
 [do apr 9 2009] [22:41:47] dberkholz those of you with a pet
 feature, feel free to respond to the summary on -dev if you have some
 compelling reason why it's super easy to get in and can't be pushed
 back. keep in mind there's no reason we need to take forever for EAPI
 4, either

Can we at least hold off on these until EAPI 3 is approved? I doubt
any of the many groups involved in EAPIs will be able to keep up with
the workload otherwise...

- -- 
Ciaran McCreesh
-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.9 (GNU/Linux)

iEYEARECAAYFAknnrusACgkQ96zL6DUtXhEZBgCffO/OlEDubIUdVyyOtJ73gwuW
+BQAoKz/D5u2ihDY9icN+ZOyQpoATFce
=jB6X
-END PGP SIGNATURE-


[gentoo-dev] pet feature: mtime preservation (was [gentoo-dev-announce] Council Summary from meeting on April 09, 2009)

2009-04-14 Thread Marijn Schouten (hkBst)
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Thomas Anderson wrote:
 Hi,
 
 Here is the summary from Thursday's council meeting. The full log along
 with the summary will appear shortly at  
 http://www.gentoo.org/proj/en/council.
 
 Regards,
 Thomas

and from #gentoo-council:
[do apr 9 2009] [22:41:47] dberkholz those of you with a pet feature, feel
free to respond to the summary on -dev if you have some compelling reason why
it's super easy to get in and can't be pushed back. keep in mind there's no
reason we need to take forever for EAPI 4, either


pet feature: mtime preservation

supereasy: yes, this is already the current behavior of both portage and pkgcore
so nothing needs to be implemented here.

can't be pushed back: the mtime of to-be-installed files is currently
unspecified. Since the behavior of portage isn't changing nothing will break due
to this non-change, but it will make hacks that work around the package manager
(officially) obsolete and make it simpler to keep track of compiled code for
dynamic languages.

There is an extended proposal for changing only those mtimes that are too old or
too new (in the future). If that is indeed desirable this could be changed in
EAPI4, but it seems easy if not better/cleaner if mtime mangling is done on a
per-ebuild basis for those few that need it.

If the package manager preserves mtimes they can still easily be changed in
individual ebuilds.
However if the package manager does not preserve mtimes and they need to be
recovered, then the mtimes need to be saved at a point where mtimes have not yet
been mangled and then restored after the package manager is done with mtime
mangling. This is possible but the exact time of mangling would then need to be
specified. This is clearly a hack.

I know I personally asked for mtime preservation on 7-5-2008 on this list[1] and
the issue has surely existed even longer. Let's get rid of this problem already.

Thanks,

Marijn

[1]:http://thread.gmane.org/gmane.linux.gentoo.devel/55953

- --
If you cannot read my mind, then listen to what I say.

Marijn Schouten (hkBst), Gentoo Lisp project, Gentoo ML
http://www.gentoo.org/proj/en/lisp/, #gentoo-{lisp,ml} on FreeNode
-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAknlJzEACgkQp/VmCx0OL2yokQCfRAjuMnssCQ1sVajiX+VVdHBC
FBUAnjUFcmBXQMvpqiFxrSNB4w69a3Fm
=RcL4
-END PGP SIGNATURE-