On 6/23/13 7:30 PM, Alexander Hansen wrote:
> On 6/21/13 8:08 PM, Daniel Johnson wrote:
>>
>> On Jun 20, 2013, at 9:15 AM, Daniel Macks <dma...@netspace.org> wrote:
>>
>>> On Thu, 20 Jun 2013 01:52:10 -0700, David Lowe
>>> <doctorjl...@verizon.net> wrote:
>>>
>>>> On Jun 19, 2013, at 8:02 PM, Alexander Hansen wrote:
>>>>
>>>>> In view of simplifying our lives we've got a few options here:
>>>>>> 1) EOL 10.5, and defer 10.6 for a while. This could be done >
>>>> essentially immediately, with all of the 10.5 packages being stashed
>>>> in > a 10.5-EOL directory similarly to we do for 10.4.
>>>>>> 2) EOL 10.5 and 10.6/i386, keeping 10.6/x86_64 around for a
>>>> while > longer. This will take additional tweaks to fink but I don't
>>>> know of a > reason at this point that precludes this.
>>>>>> 3) EOL 10.5 and 10.6. This is probably the simplest option,
>>>> because it > just requires people to stop committing to the 10.4/
>>>> tree, and a new > fink release which is set only to acknowledge 10.7
>>>> and later.
>>>>>> Anyway, feedback would be appreciated.
>>>>
>>>> For my own selfish reasons, i prefer #1. I have several working
>>>> machines that aren't acceptable to 10.7 or newer, and frankly i've
>>>> come to loath Lion on the one that is. I have no intention of moving
>>>> to 10.8 as it seems to be moving further in the wrong direction, and
>>>> will probably end up reinstalling 10.6 when support for 10.7 is
>>>> dropped. Sébastien rightfully mentions the tradition of supporting
>>>> only two recent versions of the OS. Well, yeah, Apple used to only
>>>> support two recent versions. It is, however, *still* providing
>>>> updates to Snow Leopard. I don't have any numbers, but browsing web
>>>> forums leads me to believe that masses of Macs are stuck at 10.6.
>>>
>>> 10.6 was the last system to support Rosetta (thanks for the reminder,
>>> cirdan), and I know a bunch of sites are keeping some machines at 10.6
>>> because they still need that. I can't think of a reason to keep
>>> 10.6/i386 though. Now that we've moved so many years in the x86_64
>>> world, is there anything in active development that is not in the new
>>> arch? Would be good to check and see is there are any things we missed
>>> though. And 10.6/i386 is harder to support (requires actual
>>> test-building and sometimes arch-specific tweaks) than 10.6/x86_64
>>> because it's not the native arch for the machine (whereas "it works on
>>> 10.{>6}/x86_64 therefore it'll probably be okay as-is on 10.6/x86_64"
>>> is usually true). So I lean towards #2, with #3 my second choice
>>> (because it really is easy and we really are over-extended with what we
>>> can actually support by a lot).
>>>
>>> dan
>>
>> 10.5 should definitely go. It should have gone when 10.8 came out. :)
>>
>> 10.6/i386 is a maintainer's nightmare, especially for someone who
>> maintains a lot of perlmods. I have so many packages with crazy hacks
>> just to deal with 10.6/i386/pm5100. I personally will NOT put any more
>> effort into them. I have no way to test that combination and haven't
>> for quite some time.
>>
>> I have no particular problem with 10.6/x86_64, but if we do need to
>> start a 10.9 tree, I think it's ridiculous to have to maintain THREE
>> separate trees. Is there any way that 10.6/x86_64 can be integrated
>> into the existing 10.7 tree? There should be minimal differences
>> between them. I have way too many packages to try to maintain 3 nearly
>> identical copies of them.
>>
>> Daniel
>>
>>
>
> It wasn't really Apple's tradition Sebastian was referring to, it was
> Fink's tradition prior to 10.6, too. Since 10.4-10.6 had so much in
> common for a while, we kept 10.4 around longer than we had in the past.
>
> Right now 10.9 package testing is being done in the 10.7 tree; this may
> change, however.
>
> Just so that we're clear: official support for a platform being dropped
> means:
> 1) We'll try to keep things in a working state.
> 2) Maintainers who want to do updates certainly can.
> 3) The project disclaims any responsibility for breakage introduced by
> #2. :-)
> Just like 10.4.
>
> Maintainers should feel free to make package updates for a platform that
> the project no longer supports, if they are so inclined, keeping in mind
> that all bug reports will be forwarded to them.
>
> In principle 0.34.9 could be the last for 10.5, and we could do that
> essentially immediately I'd have liked to have some new features for
> the supported platforms in 0.35.0, like the dpkg/apt updates that TheSin
> is working on, or git selfupdates, but I don't think they're ready on a
> short timeframe.
>
> We definitely don't want to wait until "Mavericks" (WTF!) comes out to
> put 10.5 out to stud. 10.6, on the other hand...
OK. Since nobody is expressing any enthusiasm for keeping 10.5 around,
I propose a retirement date of around July 1 (Canada Day) via the
simultaneous release of fink-0.34.9 for 10.5, fink-0.35.0 for 10.6+, and
a 10.5-EOL subtree of the 10.4 main tree.
--
Alexander Hansen, Ph.D.
Fink User Liaison
My package updates: http://finkakh.wordpress.com/
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
Fink-devel mailing list
Fink-devel@lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel
